Examples of Open-Ended Tasks and Mistakes Students Make with ThemExample of Task Examples of Frequent Mistakes Made by Students
Students’ faulty responses reflect a lack of experience with the types of tasks—tasks that require students to read closely and attend to the evidence in the text. To become competent at these tasks requires experience with such tasks and deliberate instruction of strategies and close reading of tasks. Three actions on the part of teachers will support students in developing the competence that will keep them on the road to college and career readiness:
Look again at the typical mistakes students make, such as those listed in Table 1. The related mistakes that students are likely to make in responding to the tasks reflect two major problems:
Teachers can help students avoid these problems by helping them to understand who their readers will be and by demonstrating for them how to frame the responses in ways that make explicit connections between their ideas and information from the text.
Writing for “remote readers” is a new experience for young students who are accustomed to sharing their writing with teachers and peers who can give feedback about clarity on the spot. Teachers need to help their students understand that as they write responses on a large-scale assessment, they are writing for readers who are unfamiliar with them personally and who will not available to ask for clarifications or to point out shortcomings of their writing. Indeed, students need to know that their responses might even be “read” and scored by a computer.
In addition, students, especially younger students, are not aware of the importance of providing clear indications of their thinking in their writing. During class discussions of text-related questions, students can ask for clarifications and have incomplete or vague responses corrected. When writing answers for a stranger to read, clarity is essential. Showing students some unclear responses to questions and discussing how to fix them is one step in developing both their awareness of the need for clarity and their skill in providing it. Having them work in groups to improve the clarity of their own responses and those of peers is another approach that can help focus student attention on how to apply this skill.
Finally, prompting students to self-monitor by asking questions is an especially effective way to help them keep in mind the need for clarity as they write. A guiding checklist can provide them with hints such as the following:
Making connections between ideas in writing is a key aspect of clarity. As in the examples above, most CCSS/ELA-related assessment tasks ask students to give evidence from the text to support their responses. Examination of student work shows that those who are unfamiliar with this kind of test question commonly provide just a conclusion and list two details from the text. They seldom offer any information as to how these details support their conclusion.
Direct instruction and practice with both written and oral responses can develop students’ skill in making connections explicit. The following are some practices and activities that teachers can use both to help students develop a model for thorough, complete answers and to learn about the aspects of their writing that trigger confusion in readers:
Modified bloom’s taxonomy [ [ 11 ] ].
Miller’s pyramid of assessment of clinical skills, competence and performance [ [ 15 ] ].
Assessment is central to the educational process, and has benefits beyond that of measuring knowledge and competence alone; principally in directing and stimulating learning, and in providing feedback to teachers and learners [ 17 ]. Recent research supports a critical role for assessment in consolidating learning, and strengthening and facilitating memorisation and recall. There is accumulating evidence that the process of stimulating recall through testing enhances learning and retention of learned material. This has been termed the testing effect , and several hypotheses have been put forward to explain it, including increased cognitive effort, conceptual and semantic processing, and increased attention to the properties distinguishing the learnt item from similar items, which strengthens the relationship between the cue which triggers the memory and the memory item itself [ 18 ],[ 19 ]. It appears to be principally the act of retrieving information from memory which strengthens knowledge and knowledge retention [ 20 ],[ 21 ], irrespective of whether retrievable is covert or overt [ 22 ]. Importantly, high-level questions appear to stimulate deeper conceptual learning and better learning retention then those pitched at a lower level [ 23 ]. A number of strategies have been proposed to exploit this in educational practice, including those recently summarised for use in medical education [ 24 ]. This is in a sense related to the “generation effect”, where it has been shown that spontaneously generating information as opposed to learning it passively improves subsequent recall [ 18 ],[ 19 ].
It is accepted that standards of assessment are inherently variable. There is therefore an obligation, in summative assessment, to ensure that assessment meets certain minimum criteria [ 25 ]. Achieving this in the individual instance is challenging, given the wide range of skills and knowledge to be assessed, marked variation in the knowledge of assessment of those who must assess and the highly variable environments in which the assessment takes place. There is now an extensive literature on assessment, in terms of research, guidelines and recommendations [ 26 ],[ 27 ]. Importantly, modern approaches recognise that no single form of assessment is suitable for every purpose, and stressed the need for programmatic assessment , which explicitly recognises that assessment is best served by a careful combination of a range of instruments matched to a particular purpose at each stage of the learning cycle, such as for formative, diagnostic or summative purposes [ 25 ],[ 26 ],[ 28 ].
Despite the proliferation of assessment methodologies which attempt to test the competence of medical students directly, such as OSCE, OSPE, case-based assessment, mini-CEX and workplace-based assessment, written assessments remain in widespread use. Much of the knowledge base required by the clinician is not necessarily testable in the performance format. Additionally, in comparison with most practical assessment formats, written tests are easier to organize and deliver, requiring little more than pen and paper or a computer, a venue, question setters and markers who need not be physically present.
In general, all forms of written assessment may be placed into one of two categories. Constructed response or open-ended questions include a variety of written formats in which the student is required to generate an answer spontaneously in response to a question. The prototypical example is the essay. There are many variants including short answer questions (SAQ), mini-essay questions, single-word and single-sentence questions and the modified essay question (MEQ). The selected-response or closed-ended format is typified by the multiple-choice question (MCQ) assessment, where candidates select the most appropriate answer from a list of options rather than generating an answer spontaneously. Many variants of the multiple-choice format have been used: current best practice recommends the use of one-best-answer (of three, four or five possible answers), and extended matching item (EMI) formats [ 29 ]. In this debate I shall use the term open-ended when referring to the constructed-response format, and multiple-choice as a synonym for the selected-response format.
All high-stakes assessments should meet an adequate standard in terms of quality and fairness, as measured by a number of parameters, summarised recently in a consensus statement [ 30 ]. Principal among these are the classic psychometric parameters of reproducibility (reliability or consistency; that a result would not essentially change with retesting under similar conditions), and validity or coherence, which I describe in detail below. Other important measures by which assessments should be judged are equivalence (assessments administered at different institutions or during different testing cycles produce comparable outcomes), feasibility (particularly in terms of efficiency and cost effectiveness), educational effect (the student who takes the assessment is thereby motivated to undertake appropriate learning), catalytic effect (the assessment provides outcomes that, when fed back into the educational programme, result in better teaching and learning) and acceptability to both teachers and learners.
It is generally accepted that the multiple-choice format, in contrast to the open-ended format, has high reliability and is efficient, a consequence primarily of wide sampling, and to a lesser extent, of its objectivity. In support of the open-ended format, it has been widely held that this format is superior at testing higher cognitive levels of knowledge and has greater validity. This belief is intuitively appealing and appears to represent the viewpoint of many of those involved in medical assessment, including those with extensive knowledge and experience in medical education. In an attempt to gain the best of both formats, there has been a shift from the prototypical essay towards newer formats comprising a larger number of short, structured questions, a development intended to retain the perceived benefit of the open-ended question with the superior reliability of the MCQ.
Thus the two formats are generally seen to be in tension, MCQ being significantly more reliable, the open-ended format having greater validity. In this debate I will compare the performance of the open-ended format with MCQ in summative assessment, particularly in final exit examinations. I draw attention to the large body of evidence which supports the view that, in summative assessment, the multiple-choice format is intrinsically able to provide all the value of the open-ended format and does so more reliably and cost effectively, thus throwing into question the justification for the inclusion of the open-ended format in summative assessment. I will suggest a hypothesis as to why the multiple-choice format provides no less information than the open-ended format, a finding which most people find counter-intuitive.
A critical concept is that assessment is not only of learning, but also for learning [ 27 ],[ 31 ]. In the first case, the purpose of assessment is to determine whether that which is required to be learnt has in fact been learnt. In the second case, it is acknowledged that assessment may in itself be a powerful driver for learning at the cognitive level. This is supported by a body of evidence indicating the powerful effect of assessment on strengthening memorisation and recall [ 20 ],[ 22 ],[ 23 ]. In this debate I concentrate primarily on summative assessment in its role as assessment of learning ; one must however remain aware that those methods of assessment best suited to such summative assessment may not be identical to those best suited to assessment for learning ; indeed, it would be surprising if they were.
For the first part of the 20 th century, written assessment in medicine consisted largely of essay-writing [ 30 ]. Multiple-choice assessment was developed for psychological testing by Robert Yerkes immediately before the First World War and then rapidly expanded for the testing of army recruits. Yerkes was interested in assessing learning capacity—not necessarily human—and applied it to crows [ 32 ] and pigs [ 33 ] as well as psychiatric patients and mentally challenged subjects, a group among whom it was widely used for a number of years thereafter [ 34 ],[ 35 ]. Application to educational assessment has been credited to Frederick J. Kelly in 1914, who was drawn to it by its efficiency and objectivity [ 36 ].
Throughout its history, the multiple-choice format has had many detractors. Their principal arguments are that closed-ended questions do not stimulate or test complex constructive cognitive processes, and that if the ability to construct rather than choose a correct answer is not actively assessed, there is a potential that it will be neither taught nor learnt [ 37 ]-[ 41 ].
As Rotfield has stated: "Students proudly show off their high grades, from multiple-choice exams, as if their future careers will depend on knowing which choice to make instead of discerning which choices exist" [ 42 ]. Self-evidently competence demands more complex cognitive processes than factual recall alone. The ability to invoke these higher levels of cognition is clearly a skill which should be explicitly assessed. Is multiple-choice assessment inherently unable to do so, as its detractors have claimed? The belief that open-ended questions test high-order cognitive skills whereas multiple-choice questions do not and that therefore by inference open-ended questions evoke and test a reasoning process which is more representative of real-life problem-solving than multiple-choice, is a serious concern which I address in this review. We begin however with a comparison of the two formats in terms of reproducibility and feasibility.
Wider sampling greatly increases reproducibility, compensating as it does for unevenness in a candidate’s knowledge, varying quality of questions and even the personality of examiners [ 43 ],[ 44 ]. That the reproducibility of the multiple-choice format is much higher than that of the open-ended format is borne out in numerous studies comparing the two formats [ 45 ]-[ 47 ]. Recognition of these shortcomings has led to the design of open-ended-formats specifically intended to increase reproducibility and objectivity, while maintaining the supposed advantages of this format in terms of validity. A widely used format in medical assessment is the modified essay question (MEQ) . The format is of a clinical scenario followed by a series of sequential questions requiring short answers. This was expressly designed to bridge a perceived gap between multiple-choice and SAQ as it was believed that it would prove better at testing high-order cognitive skills than multiple-choice while allowing for more standardised marking than the standard open-ended question [ 45 ].
Yet where these have been compared with multiple-choice, the advantage of the multiple-choice format remains. A large number of questions and multiple markers are required in order to provide acceptable reliability for MEQs and essay questions [ 45 ]. Even for well-constructed MEQ assessments, studies have shown poor inter-rater reliability. Thus in an MEQ paper in a final undergraduate medical exit examination marked in parallel by several assessors, statistically significant differences between the scores of the different examiners were shown in 50% of the questions, as well as significant differences in the median scores for the examination as a whole [ 47 ]. Nor were these differences trivial; a substantial difference in outcome in terms of likelihood of failure were shown. This is cause for concern. Schuwirth et al . have stressed the necessity for interpreting reliability in terms of outcome, particularly in terms of pass/fail misclassification, and not merely in terms of numeric scores such as Cronbach’s alpha [ 27 ]. In this and other such studies the open-ended questions were of the highest possible quality practically achievable, typically MEQ's carefully prepared by skilled question writers working in teams, reviewed for appropriateness and scored using an analytic scoring scheme designed to minimise inter-rater variability. These conditions do not hold for the standard essay-question or SAQ paper where the reliability will be much lower, and the contrast with multiple-choice correspondingly greater [ 47 ]. Open-ended items scored on a continuum, such as 0-100%, have much lower inter-rater reliability than those scored against a rigid marking schedule. Therefore the discrepancy in reliability for the "graded essay" marked on a continuum versus multiple-choice is much larger than it is for more objectively scored open-ended formats.
In contrast to the open-ended question format, the multiple-choice is objective and allows multiple sampling of a subject. The result is high reproducibility. Furthermore it substantially reduces the potential for a perception of examiner bias, and thus the opportunity for legal challenge by the unsuccessful candidate [ 48 ]. The multiple-choice format is efficient. Lukhele et al . studied a number of national university-entrance examinations which included both multiple-choice items and essay questions [ 49 ]. They found that 4-8 multiple-choice items provided the same amount of information as a single essay, and that the essay’s efficiency in providing information about the candidate’s ability per minute of testing was less than 10% of that of an average multiple-choice item. For a middle-level examinee, approximately 20 times more examination time was required for an essay to obtain the same information as could be obtained from a multiple-choice assessment. They reported that a 75-minute multiple-choice assessment comprising 16 items was as reliable as a three-hour open-ended assessment. Though the relative gain in efficiency using multiple-choice in preference to essay questions varies according to subject, it is an invariable finding [ 49 ].
Though the initial development of an multiple-choice assessment is labour-intensive, this decreases with increasing experience on the part of item-writers, and decreases further once a question bank has been developed from which questions can be drawn for re-use. The lower efficiency of the open-ended question is not restricted to examination time but also the requirement for grading by examiners. Typically an open-ended test requires from 4 to 40 times as long to administer as a multiple-choice test of equivalent reliability [ 50 ]. In one study, the cost of marking the open-ended items was 300 times that of the multiple-choice items [ 49 ]; the relative cost of scoring the papers may exceed a factor of 1000 for a large examination [ 50 ].
The multiple-choice format thus has a clear advantage over open-ended formats in terms of reproducibility, efficiency and cost-effectiveness. Why then are open-ended questions still widely used? Principally this is because of a belief that essay-type questions, SAQ and their variants test higher-order cognitive thinking in a manner that MCQ cannot, and consequently have higher validity. It has been repeatedly stated that the MCQ format is limited in its ability to test deep learning, and is suitable for assessing facts only, whereas open-ended questions assess dynamic cognitive processes such as the strength of interconnected rules, the use of the mental models, and the mental representations which follow [ 37 ]-[ 39 ]; in short that open-ended questions permit the assessment of logical and reasoning skills in a manner that multiple-choice does not [ 40 ],[ 41 ]. Is there evidence to support these assertions?
The revised Bloom's taxonomy of learning [ 9 ]-[ 12 ] is helpful in evaluating the level of cognition drawn upon by an assessment (Figure 1 ). By convention, assessment questions targeting the first two levels, are regarded as low-level questions, the third level as intermediate, and the fourth to sixth levels as high-level.
Those who understand the principles underlying the setting of high-quality multiple-choice items have no difficulty in accepting that multiple-choice is capable of assessing high-order cognition [ 10 ],[ 13 ],[ 14 ]. The shift from true-false questions, (which in order to avoid ambiguity frequently test factual information only) to the one-best-answer and EMI formats have facilitated this [ 29 ]. Indeed, there exist well-validated instruments specifically designed to assess critical thinking skills and to measure their development with progress through college-level educational programs, which are entirely multiple-choice based, such as the California Critical Thinking Skills Test [ 51 ],[ 52 ]. Schuwirth and Van der Vleuten [ 48 ] make a distinction between context-rich and context-free questions. In clinical assessment, a context-rich question is typically presented as a case vignette. Information within the vignette is presented to candidates in its original raw format, and they must then analyse, interpret and evaluate this information in order to provide the answer. The stimulus reflects the question which the candidate must answer and is therefore relevant to the content of the question. An example of a final-year question in Internal Medicine is shown in the following example. Such a question requires analysis ( What is the underlying problem? ), application ( How do I apply what I know to the treatment of this patient? ) and evaluation ( Which of several possible treatments is the most appropriate? ), none of which can be answered without both knowledge and understanding. Thus 5 of Bloom’s 6 levels have been tested.
Example of a context-rich multiple-choice item in internal medicine
A 24-year-old woman is admitted to a local hospital with a short history of epistaxis. On examination she is found to have a temperature of 36.9°C. She is wasted, has significant generalised lymphadenopathy and mild oral candidiasis but no dysphagia. A diffuse skin rash is noticed, characterised by numerous small purple punctate lesions. A full blood count shows a haemoglobin value of 110 g/L, a white cell count of 3.8×10 9 per litre and platelet count of 8.3×10 9 per litre. Which therapeutic intervention is most urgently indicated in this patient?
Antiretroviral therapy
Fluconazole
Platelet concentrate infusion
None of the options offered are obviously unreasonable or easily excluded by the candidate who attempts to shortcut the cognitive processes required in answering it by searching for clues in the options themselves. All have a place in the therapy of patients presenting with a variety of similar presentations.
Answering this item requires:
Analysis . In order to answer this item successfully, the candidate will have to recognise (1) that this patient is highly likely to be HIV-positive (given the lymphadenopathy, evidence of oral candidiasis and the high local prevalence of HIV), (2) that the presentation is suggestive of immune thrombocytopenic purpura (given the epistaxis, skin manifestations and very low platelet count), (3) that other commonly-seen concomitant features such as severe bacterial infection and extensive esophageal candidiasis are excluded by a number of negative findings.
Evaluation . Further, in order to answer this item successfully, the candidate will have to (1) consider the differential diagnosis for the principal components of the clinical vignette and, by process of evaluation, decide which are the most likely; (2) decide which of the diagnoses require treatment most urgently, (3) decide which form of therapy will be most appropriate for this.
Knowledge, understanding and application . It is utterly impossible to “recognise” the correct answer to this item without having worked through this process of analysis and evaluation, and the knowledge required to answer it must clearly be informed by deep learning, understanding and application. Hence five of the six levels of Bloom’s taxonomy have been tested. Furthermore it would appear an eminently reasonable proposition that the candidate who correctly answers this question will indeed be able to manage such a patient in practice, hence implying structural validity.
Though guessing has a 20% chance of providing the correct answer, this will be eliminated as a factor by assessing performance across multiple such items and applying negative marking to incorrect answers.
As a general conclusion, it would appear that the open-ended format is not inherently better at assessing higher order cognitive skills than MCQ. The fundamental determinant is the way in which the question is phrased in order to stimulate higher order thinking; if phrased inappropriately, the open-ended format will not perform any better than MCQ. A crucial corollary is that in comparing formats, it is essential to ensure that MCQ questions crafted to elicit high order thinking (particularly those which are context-rich) are compared with open-ended questions crafted to the same level; it is inappropriate to compare high-order items in one format with low order items in the other. Several studies have investigated the effect of the stimulus on thought processes in the open questions and have shown that the stimulus format is more important than the response format . Scores on questions in open-ended format and multiple-choice format correlate highly (approaching 100%) for context-rich questions testing the same material. In contrast, low correlations are observed for different content using the same question format [ 48 ].
In response to the low objectivity and reliability of the classic essay-type questions, modified open-ended formats have evolved which typically combine short answers, carefully crafted questions and rigid marking templates. Yet this increase in reliability appears to come at a significant cost to the presumed advantage of the open-ended format over the multiple-choice format in testing higher orders of cognition. Feletti and Smith have shown that as the number of items in the open-ended examination increases, questions probing high-order cognitive skills tend to be replaced by questions requiring factual recall alone [ 46 ]. Hence as accuracy and reliability increase, any difference between such an assessment and a multiple-choice assessment in terms of other indicators tends to disappear; ultimately they converge on an essentially identical assessment [ 47 ],[ 49 ].
Palmer and Devitt [ 45 ] analysed a large number of multiple-choice and MEQ questions used for summative assessment in a clinical undergraduate exam. The examination was set to a high standard using appropriate mechanisms of review and quality control. Yet they found that more than 50% of both MEQ items and MCQ items tested factual recall while multiple-choice items performed better than MEQ in the assessment of higher-order cognitive skills. They reported that "the modified essay question failed in its role of consistently assessing higher cognitive skills whereas the multiple-choice frequently tested more than mere recall of knowledge”.
In a subsequent study of a rigorously prepared and controlled set of exit examinations, they reported that the proportion of questions testing higher-level cognitive skills was lower in the MEQ paper then in the MCQ paper. More than 50% of the multiple-choice items assessed higher level cognition, as opposed to just 25% of the MEQ items. The problem was compounded by a higher frequency of item-writing flaws in the MEQ paper, and flaws were found in the marking scheme in 60% of the MEQ's. The authors conclude that “The MEQ paper failed to achieve its primary purpose of assessing higher cognitive skills” [ 47 ].
We therefore appear to be dealing with a general rule: the more highly open-ended questions are structured with the intention of increasing reliability, the more closely they converge on an equivalent multiple-choice question in terms of performance, thus negating any potential advantage of the open-ended format over the closed-ended [ 53 ]; indeed they appear frequently to underperform MCQ items in the very area in which they are believed to hold the advantage. Thus the shift to these newer forms of assessment may actually have had a perverse effect in diminishing the potential for the open-ended assessment to evaluate complex cognitive processes. This does not imply that open-ended items such as SAQ, MEQ and key-feature assessments, particularly those designed to assess clinical reasoning, are inherently inferior to MCQ; rather it is a warning that there is a very real risk in practice of “dumbing-down” such questions in an attempt to improve reliability, and empiric observations suggest that this is indeed a consequence frequently encountered even in carefully crafted assessments.
Combining multiple-choice and open-ended tests in the same assessment, in the belief that one is improving the strength of the assessment, leads to an overall less reliable assessment than is constituted by the multiple-choice section on its own [ 49 ], thus causing harm rather than adding benefit [ 50 ].
The second argument, frequently advanced in support of the open-ended format, is that it has greater validity; that spontaneously recalling and reproducing knowledge is a better predictor of the student’s eventual ability to handle complex problems in real-life then is the ability to select an answer from a list [ 54 ]. Indeed, this argument is intuitively highly appealing. The case for the retention of open-ended questions in medical undergraduate and postgraduate assessment largely rests on validity, with the assumption that asking the candidate to describe how they would diagnose, investigate and treat a patient predicts future clinical competence more accurately than does the ability to select the right response from a number of options [ 55 ],[ 56 ]. The question of validity is central. If the open-ended format is genuinely of higher validity than the multiple-choice format, then there is a strong case for retaining essay-type questions, SAQ and MEQ in the assessment protocol. If this contention cannot be supported, then the justification for retaining open-ended items in summative assessment may be questioned.
Is the contention true? Essentially, this may be explored at two levels. The first is to correlate outcomes between the two formats. The second is to perform appropriate statistical analysis to determine whether these formats are indeed testing different dimensions or “factors”.
Validity is an indicator of how closely the assessment actually measures the quality it purportedly sets out to test. It is self-evident that proficiency in many domains, including clinical practice, requires not only the ability to recall factual knowledge, but also the ability to generate and test hypotheses, integrate knowledge and apply it appropriately as required.
Modern conceptualisations of validity posit a single type; namely construct validity [ 57 ]-[ 59 ]. This is based on the premise that ultimately all validity rests on the fidelity with which a particular assessment reflects the underlying construct, “intangible collections of abstract concepts and principles which are inferred from behaviour and explained by educational or psychological theory” [ 60 ]. Construct validity is then defined as a process of investigation in which the constructs are carefully delineated, and evidence at multiple levels is sought which supports a valid association between scores on that assessment and the candidate's proficiency in terms of that construct. For example, five types of evidence have been proposed which may provide support for such an association [ 60 ],[ 61 ], namely content, the response process, internal structure, relationship to other variables and consequences. In this discussion we highlight the relevant to the last two methods; convergent correlations between the two forms of assessment, and the impact of test scores on later performance, particularly that requiring problem-solving under conditions encountered in the work situation. This “is particularly important to those employers more interested in hiring competent workers than good test takers” [ 62 ].
Correlation.
Numerous studies have assessed the correlation of scores between the two formats. If scores are highly correlated, the two formats are essentially measuring the same thing in which case, in terms of validity, there is no advantage of one over the other. With few exceptions, studies indicate that scores on the two forms of assessment are highly correlated. Norman et al. compared the two formats prospectively and showed a strong correlation between the two sets of scores [ 63 ]. A similar result was found by Palmer et al. who suggested that the two types of examination were essentially testing similar characteristics [ 47 ]. Similarly Norcini et al. found that written patient management problems and multiple choice items appeared to be measuring essentially the same aspects of clinical competence, though the multiple-choice items did so more efficiently and with greater reliability [ 17 ]. Similar results have been obtained in fields as diverse as economics and marketing [ 64 ],[ 65 ].
In general correlations between the two formats are higher when the questions in each format are specifically designed to be similar (stem-equivalent), and lower where the items in the two formats differ. However, the difference is not great: in a meta-analysis, Rodriguez found a correlation across 21 studies of 0.92 for stem-equivalent items and 0.85 across 35 studies for non-stem-equivalent items. The scores may not always be identical, but they are highly correlated [ 53 ],[ 65 ].
Identification of the actual constructs measured in an assessment has proved challenging given the lack of congruence between the simple cognitive assumptions on which testing is often based and the very complex cognitive nature of the constructs underlying understanding [ 66 ]. A number of studies have used confirmatory factor analysis and principal component analysis to determine whether the constructs tested by the two formats lie along a single dimension or along two or more divergent dimensions. Bennett et al . compared a one factor model with a two factor model to examine the relationship of the open-ended and closed-ended formats and found that in general the single factor provided a better fit. This suggests that essentially the two formats are testing the same thing [ 67 ]. Similarly Bridgeman and Rock found, using a principal components model, that both formats appeared to load on the same factor, implying that the open-ended format was not providing information on a different dimension [ 68 ]. Thissen and Wainer found that both formats could largely be ascribed to a single shared factor but did find some specific open-ended factors for which only the open-ended items contributed [ 69 ]. Though Lissitz et al . [ 70 ] quote a study by JJ Manhart, which found a two-factor model generally more appropriate than a one factor model, this study has not been published and the significance of the divergence cannot be assessed.
In a study of high school assessments using confirmatory factor analysis, Lissitz et al. showed a correlation of 0.94 between the two formats in the domains of algebra and biology; a two-factor model provided a very slight increment over a one-factor model in terms of fit. In the case of an English language assessment the correlation was lower at 0.74 and a two-factor model provided a better fit. In a test of US government, intermediate results were found with the correlation of 0.83 and a slight superiority of a two-factor model. This suggests that the addition of open-ended items in biology and algebra provided little further information beyond the multiple-choice items, whereas in other domains—English and government—the two formats are to some degree measuring different constructs [ 70 ]. Indeed, the literature in general suggests that differences in format appeared to be of little significance in the precise sciences such as biology and mathematics, but may have some relevance in fields such as history and languages, as suggested by Traub and Fisher [ 71 ]. In summary, there is little evidence to support the belief that the open-ended format is testing dimensions which the multiple-choice format cannot [ 53 ],[ 70 ],[ 72 ].
Construct validity was specifically assessed by Hee-Sun et al . [ 73 ], who attempted to measure the depth of understanding among school-level science students revealed by multiple-choice and short written explanatory answers respectively. They reported that students who showed higher degrees of knowledge integration were more likely to score highly on multiple-choice, though the reverse did not hold true. They suggested that the multiple-choice items were less effective in distinguishing adjacent grades of understanding as opposed to distinguishing high-performance from low performance, a finding similar to that of Wilson and Wang [ 74 ] and Ercikan et al . [ 75 ]. Unfortunately the generalisability of these results is limited since the multiple-choice items were poorly standardised, both in format and in difficulty, and the circumstances under which the testing was conducted were essentially uncontrolled.
Lukhele et al . performed a rigorous analysis of high-quality university placement exams taken by thousands of candidates [ 49 ]. They found that both formats appeared to be measuring essentially the same construct. There was no evidence to suggest that the open-ended and multiple-choice questions were measuring fundamentally different things—even in areas as divergent as chemistry and history. Factorial analysis suggested that there were two variant dimensions reflected in the scores of the multiple-choice and open-ended sections, one slightly more related to multiple-choice and the other to the open-ended format. However these were highly correlated, whatever the factor is that is specifically measured by the open-ended format, multiple-choice would measure it almost as well. Thus for all practical purposes, in such summative assessments, multiple-choice assessments can satisfactorily replace open-ended assessments.
An important principle is that the variance introduced by measuring “the wrong thing” in the multiple-choice is small in comparison with the error variance associated with the open-ended format given its low reliability. This effectively cancels out any slight advantage in validity [ 49 ] (Figure 3 ). Indeed, Wainer and Thissen state that “measuring something that is not quite right accurately may yield far better measurement than measuring the right thing poorly” [ 50 ].
Stylized depiction of the contrasting ability of the presumed open-ended and multiple-choice formats to assess recognition and recall as opposed to higher forms of cognitive learning. Ideally, multiple-choice and open-ended questions would measure two different abilities (such as recall/recognition versus reasoning/application) – this may be shown as two divergent axes (shown on left). The error variance associated with each type of question is indicated by the shaded blocks, and is much greater for the open-ended question, given its inherent lower reliability. In practice, it appears that the two axes are closely aligned, implying that the two types of questions are measuring essentially the same thing (shown on right). What little additional information the open-ended question might be giving (as shown by a slight divergence in axis) is offset by its wide error variance, which in effect overlaps the information given by the multiple-choice question, thus significantly reducing the value of any additional information it provides.
In summary, where studies have suggested that the open-ended format is measuring something that multiple-choice does not (particularly in older studies), the effect has tended to be minimal, or possibly explicable on methodological grounds, or indefinable in terms of what is actually being measured. In contrast, methodologically sound studies converge on the conclusion that the difference in validity between the two formats is trivial. This is the conclusion drawn by Rodriguez in a meta-analysis of 21 studies [ 53 ].
Demonstrating an essential similarity for the two formats under the conditions of summative assessment does not necessarily mean that they provide identical information. It is possible and indeed likely that open-ended questions may make intermediate steps in thinking and understanding visible, thus serving a useful role in diagnostic as opposed to summative assessment [ 73 ],[ 75 ],[ 76 ]. Such considerations are particularly useful in using assessment to guide learning rather than merely as a judgment of competence [ 77 ]. In summative assessment at a stage prior to final exit from a programme, and particularly in formative assessment, the notion of assessment for learning becomes important; and considerations such as the generation effect and the potentiation of memory recall by testing cannot be ignored. Interestingly, a recent publication suggests that multiple-choice format testing is as effective as SAQ-format testing in potentiating memorisation and recall [ 23 ], thus supporting the contention that well-crafted MCQ and open-ended questions are essentially stimulating the same cognitive processes in the learner.
Some authors have raised the concern that students may constitutionally perform differentially on the two forms of assessment, and might be disadvantaged by a multiple-choice assessment should their strengths lie in the open-ended format. Studies in this area have been reassuring. Bridgeman and Morgan found that discrepant results were not predictive of poor academic performance as assessed by other parameters [ 78 ]. Ercikan et al . reported that discrepancies in the outcome between open-ended and multiple-choice tests were largely due to the low reliability of the open-ended component and inappropriate testing strategies [ 75 ]. A study which correlated the two formats with each other and with other measures of student aptitude showed a high degree of correlation and was unable to identify students who clearly had a propensity to perform consistently better on one format than the other [ 79 ]. Thus the belief that some students are constitutionally more suited to open-ended questions than to multiple-choice would appear to be unfounded.
An important question is whether the format of assessment effects the type of learning students use in preparation for it. As early as 1971, Hakstian suggested that anticipation of a specific form of examination did not result in any change in the amount or type of preparation, or any difference in performance in subsequent testing [ 80 ]. He concluded as follows: “The use of various types of tests to foster various kinds of study and learning, although widely advocated would seem to be a practice based on intuitive appeal, but not convincingly supported by empirical research. In particular, the contention that the superiority of the essay examination is its ability to promote more desirable study methods and higher performance on tasks requiring organisation, and deeper comprehension analysis of information should be re-evaluated in light of the evidence in the present study of no differences between groups in terms of study methods, the essay examination, or items from the higher levels of the cognitive domain”. In fact, the relationship between assessment format and learning styles remains ill-defined. Though some studies have suggested that students tended to make more use of surface learning strategies in preparation for MCQ and deeper learning strategies in preparation for open-ended questions [ 81 ],[ 82 ], other studies have failed to show such an association [ 80 ],[ 83 ]. Some studies have even failed to show that deep learning approaches correlated with better performance in applied MCQ’s and a written course project, both of which required high level cognitive performance [ 84 ],[ 85 ], though, a significant finding was that a surface learning strategy appeared deleterious for both factual and applied MCQ scores [ 85 ].
Indeed, a review of the literature on learning strategies suggests that the notion that one or other assessment format consistently calls forth a particular learning strategy is simplistic, and much of the evidence for this may have been misinterpreted [ 86 ]. The student’s choice of learning style appears to be dependent on multiple interacting and to some extent, confounding factors, most importantly the student’s innate learning motivation and preferred learning strategy. This is however subject to modification by other factors, particularly the student’s own perception of whether the assessment is directed at assessment of factual knowledge or of understanding, a perception which may frequently not coincide with the intentions of the examiner [ 87 ]. Individual differences in learning strategy probably outweigh any other consideration, including the assessment format, though this is not constant and students will adapt their preferred learning strategy according to their perception of the requirement for a particular assessment [ 88 ]. A further study has suggested that the approach to learning the student brings into the course is the strongest predictor of the learning style they will employ subsequently and, irrespective of the instructor’s best efforts, the only factor significantly correlated with the change in learning style is a change in the student’s perception of the cognitive demands of the assessment. Thus students are frequently strategic in their choice of learning strategy, but the strategies may be misplaced [ 87 ]. The student’s academic ability may be relevant; one study has shown that more academically able science students correctly identified the MCQ as requiring deep knowledge and adopted an appropriate learning strategy, whereas less able students interviewed the assessment as principally a test of recall and used a counter-productive surface-learning strategy.
Hadwin et al . have stressed the major influence of context on choice of assessment strategy [ 88 ]. There is for example evidence that students will modify their strategy according to whether the assessment is perceived as a final examination or as an interim assessment, irrespective of format [ 81 ]. So-called construct-irrelevant factors such as female gender and increasing maturity tend to correlate with selection of a deep learning strategy [ 85 ] independent of assessment format, while the association of anxiety and other emotional factors with a particular assessment will impair performance and thus operate as a confounding factor [ 89 ],[ 90 ]. In discussing their results, Smith and Miller stated that “Neither the hypothesis that multiple-choice examination will promote student use of surface strategy nor the hypothesis that essay examination will promote student use of deep strategy were supported” [ 91 ]. As a general conclusion, it would appear valid to say that current evidence is insufficient to suggest that the open-ended format should be preferred over MCQ or vice versa on the grounds that it promotes more effective learning strategies.
It is also important to be aware that open-ended assessments may bring confounding factors into play, for example testing language mastery or skills rather than the intended knowledge domain itself [ 70 ], and hand-written answers also penalise students with poor writing skills, low writing speeds and poor handwriting [ 65 ].
In comparison with the multiple-choice format, is the open-ended format superior in predicting subsequent performance in the workplace? This has been assessed and the answer, surprisingly, is that it may be less predictive. Rabinowitz and Hojat [ 92 ] correlated the single MEQ assessment and five multiple-choice assessments written at the conclusion of a series of six clerkships with performance after graduation. Results in multiple-choice assessment consistently demonstrated the highest correlations with subsequent national examination scores and with objective assessments of performance in the workplace. The MEQ questions showed the lowest correlation. Wilkinson and Frampton directly compared an assessment based on long and short essay-type questions with a subsequent assessment protocol containing short essay questions and two multiple-choice papers [ 56 ], correlating these with performance in the subsequent internship year using robust rating methodologies. They found no significant correlation between the scores of the open-ended question protocol and assessments of performance in the workplace after graduation. In contrast they found that the combination of the SAQ paper and two multiple-choice papers showed a highly significant correlation with subsequent performance. This study showed that the predominant use of multiple-choice in the assessment resulted in a significant improvement in the structural validity of the assessment in comparison with essay-type questions alone. It was unable to answer the question as to whether the open-ended questions are necessary at all since the multiple-choice component was not compared with the performance rating independently of the essay questions. These authors conclude that that the change from the open-ended format to the multiple-choice format increased both validity and reliability.
Wainer and Thissen stated that: “We have found no evidence of any comparison of the efficacy of the two formats (when a particular trait was specified and skilled item writers then constructed items to measure it) in which the multiple-choice item format was not superior” [ 50 ]. Lukhele et al . concluded: “Thus, while we are sympathetic to… the arguments… regarding the advantages of open-ended format, we have yet to see convincing psychometric evidence supporting them. We are awash in evidence of their drawbacks”, and further, “… We are forced to conclude that open-ended items provide this information in more time at greater cost than the multiple-choice items. This conclusion is surely discouraging to those who feel that open-ended items are more authentic and, hence, in some sense, more useful than multiple-choice items. It should be” [ 49 ].
Palmer et al . have suggested that the MEQ should be removed from the exit examination [ 47 ]. Given that MEQ's are difficult to write to a high standard and in such a way that they test high-order cognitive skills, and given the time required and the subjectivity in marking, their use does not represent an efficient use of resources. Indeed, they state “… MEQ's often do little more than test the candidate's ability to recall a list of facts and frustrate the examiner with a large pile of papers to be hand-marked”. They conclude there is no good measurement reason for including open-ended items in the high-stakes assessment, given that the MEQ performed poorly in terms of testing high-order thinking in comparison with the multiple-choice despite considerable effort to produce quality questions.
Schuwirth and Van der Vleuten too have suggested that there is no justification for the use of SAQ in assessment, since the stimulus of most SAQ can also be applied with multiple-choice. They recommend that SAQ should not be used in any situation except where the spontaneous generation of the answer is absolutely essential. Furthermore, they believe that there is little place for context-free questions in medical assessment as the context-rich stimulus approximates clinical practice more closely [ 48 ].
Hence the evidence suggests that in written summative assessment the multiple-choice format is no less able to test high-order thinking than open-ended questions, may have higher validity and is superior in reliability and cost-effectiveness. Remarkably this evidence extends as far back as 1926 [ 53 ],[ 93 ], and the reasons underlying the persistence of the open-ended format in assessment are of some interest. I suggest a number of factors. Studies bear out the common-sense expectation that questions designed to test factual knowledge only—irrespective of whether these are presented as open-ended or in multiple-choice format—do not test the same level of reasoning as more complex questions [ 94 ]. Indeed, a recurring finding in the literature is that the so-called deficiencies of the multiple-choice format lie more with the quality of the individual question item (and by inference, with the question-setter), than with the format per se . This leads to a self-fulfilling prophecy: examiners who do not appreciate the versatility of the multiple-choice format set questions which only test low-order thinking and not surprisingly achieve results which confirm their bias. Palmer et al. state that criticism of multiple-choice as being incapable of testing high-order thinking is in fact criticism of poorly written questions, and that the same criticism can be directed at open-ended assessments [ 45 ]. There is indeed evidence that stem-equivalent items tend to behave similarly, irrespective of whether the item is phrased as an open-ended question or in MCQ format. It is therefore essential that in making comparisons, the items compared are specifically crafted to assess the same order of cognition. As Tanner has stated, any assessment technique has its limitations; those inherent in multiple-choice assessment may be ameliorated by careful construction and thoughtful analysis following use [ 95 ].
Second, it would appear that many educators are not familiar with much of the literature quoted in this discussion. The most persuasive material is found in the broader educational literature, and though there are brief references in the medical education literature to some of the studies to which I have referred [ 47 ],[ 48 ], as well as a few original studies performed in the medical assessment context [ 17 ],[ 45 ],[ 47 ],[ 63 ], the issue does not appear to have enjoyed prominence in debate and has had limited impact on actual assessment practice. In their consensus statement and recommendations on research and assessment, Schuwirth et al. stress the need for reference beyond the existing medical education literature to relevant scientific disciplines, including cognitive psychology [ 27 ]. In the teaching context, it is remarkable how the proposition that the open-ended format is more appropriate in testing the knowledge and skills ultimately required for the workplace has been repeatedly and uncritically restated in the literature in the absence of compelling evidence to support it.
Third is the counter-intuitiveness of this finding. Indeed, the proposition that the open-ended format is more challenging than MCQ is intuitively appealing. Furthermore, there is the “generation effect”; experimental work has shown that spontaneous generation of information, as opposed to reading enhances recall [ 18 ],[ 19 ]. Although this applies to learning rather than to assessment, many teachers implicitly attribute a similar but reversed process to the act of recall, believing that spontaneous recall is more valid than cued recall. However, validity at face value is an unreliable proxy for true validity, and the outcome in practice may contradict what seems intuitively correct [ 48 ]. As the literature on learning increases, it has become apparent that evidenced-based practice frequently fails to coincide with the intuitive appeal of a particular learning methodology. Examples include the observation that interleaved practice is more effective than blocked practice and distributed practice is more effective than massed practice in promoting acquisition of skills and knowledge [ 21 ]. There is a need for assessment to be evidence-based; to an extent assessment would appear to lag behind learning and teaching methodology in this respect. Rohrer and Pashler have suggested that underutilisation of learning strategies shown to be more effective than their traditional counterparts, such as learning through testing, distributed practice and interleaved practice, remain so because of “the widespread (but erroneous) feeling that these strategies are less effective than their alternatives” [ 21 ].
Fourth and perhaps most defensible is concern that there is much that as yet remains unknown about the nature of assessment; particularly seen from the viewpoint of assessment for learning, and given very interesting new insights into the cognitive basis of memorisation, recall and reasoning, a field which is as yet largely unexplored, and may be expected to have a significant impact on the choice of assessment format. For diagnostic purposes, the open-ended format may hold value, since it is better able to expose the students intermediate thinking processes and therefore allow precise identification of learning difficulties [ 72 ]. Newer observations such as the generation effect [ 18 ],[ 19 ], the testing effect [ 20 ],[ 23 ], the preassessment effect, where the act of preparation for an assessment is itself a powerful driver of learning [ 96 ], and the post-assessment effect, such as the effect of feedback [ 96 ] are clearly important; were it to be shown that a particular format of assessment, such as the open-ended question, was superior in driving learning, then this would be important information which might well determine the choice of assessment. At this point however no such reliable information exists. Preliminary work suggests that MCQ items are as effective as open-ended items in promoting the testing effect [ 23 ]. None of these considerations are as yet sufficiently well supported by experimental evidence to argue definitively for the inclusion of open-ended questions on the basis of their effect on learning, though the possibility clearly remains. Furthermore, this debate has concentrated on high-stakes, summative exit assessments where the learning effects of assessment are presumably less important than they are at other stages of learning. Certainly, open-ended assessment remains appropriate for those domains not well-suited to multiple-choice assessment such as data gathering, clinical judgement and professional attitudes [ 92 ] and may have value for a particular question which cannot be presented in any other format [ 48 ]. Though the evidence is less compelling, open-ended items may be superior in distinguishing between performances of candidates occupying the two extremes of performance [ 75 ].
The need for assessment of research to move beyond empiric observations to studies based on a sound theoretical framework has recently been stressed [ 27 ],[ 96 ]. There is as yet little written on the reasons for the counter-intuitive finding that MCQ is as valid as open-ended assessments in predicting clinical performance. I suggest that the observation is highly compatible with cognitive-constructivist and situated learning theory, and in particular the theory of conceptual change [ 97 ]. Fundamental to this theory is the concept of mental models. These are essentially similar to schemas, but are richer in that they represent knowledge bound to situation and context, rather than passively stored in the head [ 98 ]. Mental models may therefore be thought of as cognitive artifacts constructed by an individual based on his or her preconceptions, cognitive skills, linguistic comprehension, and perception of the problem, which evolve as they are modified through experience and instruction [ 99 ]. Conceptual change is postulated to represent the mechanism underlying meaningful learning, and is a process of progressively constructing and organizing a learner’s personal mental models [ 100 ],[ 101 ]. It is suggested that an effective mental model will integrate six different aspects: knowledge appropriately structured for a particular domain (structural knowledge), pathways for solving problems related to the domain (procedural knowledge), mental images of the system, associations (metaphors), the ability to know when to activate mental models (executive knowledge), and assumptions about the problem (beliefs) [ 102 ]. Therefore increasing proficiency in any domain is associated not just with an enlarging of store of knowledge and experience, but also with increasing complexity in the extent to which knowledge is organised and the manner in which it is stored and accessed [ 103 ], particularly as complex mental models which may be applied to problem-solving [ 104 ]. A counterpart in the domain of medical expertise is the hierarchy of constructs proposed by Schmidt et al . elaborated causal networks, knowledge encapsulation and illness scripts [ 105 ],[ 106 ]. Conceptual change theory has a clear relationship to our current understanding of expertise, which is postulated to emerge where knowledge and concepts are linked as mental representations into propositional networks which allow rapid processing of information and the omission of intermediate steps in reasoning [ 107 ],[ 108 ]; typically the expert’s knowledge is grouped into discrete packets or chunks, and manipulation of these equates to the manipulation of a large amount of information simultaneously without conscious attention to any individual component [ 104 ]. In comparison with non-experts, the representations of experts are richer, more organised and abstract and are based on deep knowledge; experts also recognise the conditions under which use of particular knowledge is appropriate [ 109 ]. As Norman has stated, “expert problem-solving in medicine is dependent on (1) prior experiences which can be used in routine solution of problems by pattern recognition processes and (2) elaborated conceptual knowledge applicable to the occasional problematic situation ” [ 110 ]. The processes of building expertise and that of constructing mental models are essentially parallel [ 99 ].
Therefore any form of assessment intended to measure proficiency must successfully sample the candidate’s organisation of and access to knowledge, and not just content knowledge alone [ 99 ],[ 111 ]. I have reviewed the empirical evidence which suggests that the multiple-choice format is indeed predictive of proficiency, which provides important evidence that it is valid. This is explicable in terms of mental models. An alternative view of a mental model is as an internal representation of a system that the learner brings to bear in a problem-solving situation [ 103 ],[ 104 ],[ 112 ]. The context-rich written assessment [ 48 ] is essentially an exercise in complex problem-solving, and fits the definition of problem-solving as “cognitive processing aimed at accomplishing certain goals when the solution is unknown” [ 103 ],[ 113 ].
Zhang has introduced the concept of a “distributed cognitive task”: a task requiring that information distributed across both the internal mind and the external environment is processed [ 114 ]. If we extend Zhang’s concept of external representation to include a hypothetical patient, the subject of the clinical vignette, who represents the class of all such patients, then answering the context-rich multiple-choice item may be seen as a distributed cognitive task. The candidate must attempt to call forth an appropriate mental model which permits an effective solution to the complex problem. In a sequence of events which parallels that described by Zhang, the candidate must internalise the information provided in the vignette, form an accurate internal representation (an equivalent concept is that of the problem space, a mental representation of the problem requiring solution [ 115 ]); this in turn activates and interacts with the relevant mental models and is followed by externalization: the return of the product of the interaction of internal representation and mental model to the external environment, and the selection of a solution. In effect a relationship has been defined between environmental information, activation of higher level cognition and externalisation of internal representations [ 114 ].
Assessment items which require complex problem-solving call on mental models appropriate to that particular context, and the item can only be answered confidently and correctly if the mental model is present at the level of proficiency. There is therefore no such thing as the student with generic expertise “in answering multiple-choice questions”, which explains the findings of Hakstian [ 80 ], Bridgeman and Morgan [ 78 ], Ercikan et al. [ 75 ] and Bleske-Rechek et al . [ 79 ], none of whom found convincing evidence for the existence of a class of student with a particular skill in answering multiple-choice questions.
Recent observations that retrieval of knowledge improves retention, and may be enhanced in the learning process by frequent testing [ 20 ],[ 21 ], and in particular a recent publication summarising four studies performed in an authentic learning environment which demonstrates that that testing using MCQ format is as effective as SAQ testing [ 23 ], supports the hypothesis that the MCQ format engages with high order cognitive processes, in both learning and retrieval of memory. This is further supported by their finding that high-level test questions stimulate deeper conceptual learning and better learning retention then do low-level test questions [ 23 ].
In summary, the multiple-choice item is testing the integrity and appropriateness of the candidate’s mental models, and in doing so, is in fact assessing proficiency. If the item is designed to test factual recall only then it will fail for this purpose, since it is the solution of a complex problem which tests the strength of the mental model and the cognitive processes which interact with it. Yet even a low-quality assessment based on factual recollection will correlate significantly with proficiency. Firstly, all mental models are based on a foundation of structural knowledge. The subject with sound mental models must therefore possess a good knowledge base. Secondly, possessing effective and appropriate mental models facilitates the retention and recall of knowledge [ 103 ]. Not surprisingly therefore, even on a fact-based assessment, good students will correctly recall the information and excel; students with deficient mental models, are less likely to be able to recall the information when needed. This is supported by the work of Jensen et al . [ 116 ] who found that high order questions stimulated deep conceptual understanding and retention, and correlated with higher performance on both subsequent high order assessment items and low-order assessment items. Indeed, recognition and recall are highly correlated [ 50 ]. There is evidence that the cognitive processes evoked by the multiple-choice format are not influenced by cueing [ 117 ], though the reasons for the frequent observation that MCQ scores are higher than those for equivalent open-ended item assessments raise concern that cueing may yet have a role [ 118 ]. However, where the stem and options have been well-designed―particularly such that the distractors all appear attractive to the candidate without the requisite knowledge― cueing should not be an issue [ 29 ],[ 48 ], and the common argument that it is easier to recognize an answer than it is to generate it spontaneously would appear not to hold true.
Problem-solving skills are poorly generalizable [ 41 ]. This is explicable in that mental models are essentially domain-specific, representing a particular set of knowledge and circumstances, but the actual process of developing them is highly dependent on domain-general processes including metacognition, self-regulation and cognitive flexibility [ 99 ].
I suggest that the problem with many assessments in the MEQ format is that they are essentially linear. By requiring the candidate to think one step at a time, the assessment effectively misses the crux of the problem-solving process, which is to look at and respond to a complex problem in its entirety, and not stepwise. The context-rich vignette-based multiple-choice item by contrast presents a complex problem which must be holistically assessed. Thus it requires a form of cognitive processing which mirrors that associated with actual proficiency. Hybrid formats such as key feature assessments in effect also break down the clinical reasoning process into a sequence of sequential steps; whether this is regarded as a drawback will depend on the relative importance ascribed to decision-making at critical points in the decision tree and global assessment of a problem viewed holistically. This is a critical area for future research in clinical reasoning.
Educators who mistrust the multiple-choice format have tended to concentrate on the final, and cognitively the least important, step in this whole process: the selection of a particular option as the answer, while ignoring the complex cognitive processes which precede the selection. Indeed, in a good assessment, the candidate is not “selecting” an answer at all. They recognise the external representation of a problem, subject the internalised representation to high level cognitive processing, and then externalise the product as a solution [ 119 ], which (almost as if coincidentally) should coincide with one of the options given.
The multiple-choice format is by no means unlimited in its capacity to test higher-order thinking. The literature on problem-solving stresses the importance of highly-structured complex problems, characterised by unknown elements with no clear path to the solution and indeed a potential for there to be many solutions or even no solution at all [ 99 ]. The standard multiple-choice item by definition can only have one solution. Thus, though it may be context-rich, it is limited in its complexity. It is difficult however to imagine how a practically achievable open-ended written assessment might perform better. In order to accommodate complexity, the question would essentially have to be unstructured—thereby eliminating all the structured short-answer progeny of the essay format, such as MEQ. In order to permit the candidate to freely demonstrate the application of all his or her mental resources to a problem more complex than that permitted by a multiple-choice vignette, one would in all probability require that the candidate is afforded the opportunity to develop an extensive, unstructured and essentially free-ranging, essay-length response; marking will be inherently subjective and we are again faced with the problem of narrow sampling, subjectivity and low reliability.
In effect the choice would then lie between an assessment comprising one or two unstructured essay length answers with low objectivity and reliability, and a large number of highly reliable multiple choice items which will effectively test high-order problem-solving, but will stop short of a fully complex situation. Perhaps this is a restatement of the assertion that “measuring something that is not quite right accurately may yield far better measurement than measuring the right thing poorly” [ 50 ], the situation depicted in Figure 3 .
Another way of understanding the validity of the multiple-choice format is by comparing the responses of candidates at different phases of the learning process with the stages of increasing proficiency posited by Dreyfus et al . [ 16 ] (Table 1 ). Here the first column comprises the stages of learning; in this context, we shall regard stage of learning as synonymous with level of proficiency or expertise, which is a measure of the effectiveness of problem-solving skill. The second column contains descriptors for each stage chosen for their relevance to complex problem-solving posed by a well-constructed context-rich multiple-choice item. The third column contains a description of the likely performance on that item of a candidate at that stage of proficiency. The relationship between proficiency and performance in a complex multiple-choice item is in fact remarkably direct. The candidate who has reached the stage of proficiency or expertise will be more likely to select the correct response than candidates at a lower level, and the more widely such proficiency is spread across the domain, the higher the aggregate score in the assessment. Though the score for a standard multiple-choice item is binary (all or nothing), the assessment as a whole is not. Whereas candidates in the top categories are likely to arrive at a correct solution most of the time, and students in the lowest category hardly ever, the middle order candidates with less secure mental models will answer with less confidence, but will in a number of items proportional to their proficiency, come up with the correct solution, their mental models proving to be sufficiently adequate for the purpose. Over a large number of items such a multiple-choice assessment will therefore provide a highly accurate indication of the level of proficiency of the candidate. To avoid all confounding variables however it is absolutely essential that the options are set such that cueing is eliminated.
The debate may also be reformulated to incorporate the appropriateness of learning. Deep learning is characterised by an understanding of the meaning underlying knowledge, reflection on the interrelationships of items of information, understanding of the application of knowledge to everyday experience, integration of information with prior learning, the ability to differentiate between principle and example and the organisation of knowledge into a coherent, synthetic structure [ 99 ],[ 100 ]—essentially an alternative formulation of the mental model. One can thus argue that the candidate who possesses deep knowledge has, by the very fact of that possession, demonstrated that they have the sort of comprehensive and intuitive understanding of the subject—in short, the appropriate mental models as described by Jonassen and Strobel [ 97 ],[ 101 ]—to allow the information to be used for problem-solving. Correspondingly, the weak student lacks deep knowledge, and this will be exposed by a well-constructed multiple-choice assessment, provided that the items are written in a manner which explores the higher cognitive levels of learning.
Therefore, if candidates demonstrate evidence of extensive, deeply-learned knowledge, and the ability to solve complex problems, be it through the medium of multiple-choice assessment or any other form of assessment, then it is safe to assume that they will be able to apply this knowledge in practice. This accounts for the extensive correlation noted between multiple-choice performance, performance in open-ended assessments, and tests of subsequent performance in an authentic environment.
The argument that open-ended questions do not test higher order cognitive skills, and consequently lack validity, is not supported by the evidence. Some studies may have been confounded by the unfair comparison of high-order items in one format with low-order items in another. This cannot be discounted as partly responsible for the discrepancies noted in some of the work I have referenced, such as that of Hee-Sun et al . [ 73 ], yet where the cognitive order of the items have been carefully matched, a number of careful studies suggest that, particularly in science and medicine, the two modalities assess constructs which though probably not identical, overlap to the extent that using both forms of assessment is redundant. Given the advantage of the multiple-choice format in reliability, efficiency and cost-effectiveness, the suggestion that open-ended items may be replaced entirely with multiple-choice items in summative assessment is one which deserves careful consideration. This counter-intuitive finding highlights our lack of understanding of the cognitive processes underlying both clinical competence and its assessment, and suggests that much further work remains to be done. Despite the MCQ format’s long pedigree, it is clear that we understand little about the cognitive architecture invoked by this form of assessment. The need for a greater role for theoretical models in assessment research has been stressed [ 27 ],[ 96 ]. As illustrated in this debate, medical teaching and assessment must be based on a solid theoretical framework, underpinned by reliable evidence. Hard evidence combined with a plausible theoretical model - which must attempt to explain the observations on the basis of cognition - will provide the strongest basis for the identification of effective learning and assessment methodologies.
That the multiple-choice format demonstrates high validity is due in part to the observation that well-constructed, context-rich multiple-choice questions are fully capable of assessing higher orders of cognition, and that they call forth cognitive problem-solving processes which exactly mirror those required in practice. On a theoretical basis it is even conceivable that the multiple-choice format will show superior performance in assessing proficiency in contrast with some versions of the open-ended format; there is indeed empirical evidence to support this in practice [ 56 ],[ 92 ]. Paradoxically, the open-ended format may demonstrate lower validity than well-written multiple-choice items; since attempts to improve reliability and reduce objectivity by writing highly focused questions marked against standardised, prescriptive marking templates frequently “trivialize” the question, resulting in some increase in reproducibility at the expense of a significant loss of validity [ 120 ]. Indeed, I have argued that, based on an understanding of human cognition and problem-solving proficiency, context-rich multiple-choice assessments may be superior in assessing the very characteristics which the proponents of the open-ended format claim as a strength of that format.
Though current evidence supports the notion that in summative assessment open-ended items may well be redundant, this conclusion should not be uncritically extrapolated to situations where assessment for learning is important, such as in formative assessment and in summative assessment at early and intermediate stages of the medical programme given that conclusive evidence with respect to the learning effects of the two formats is as yet awaited.
The author was solely responsible the literature and writing the article.
RJH is currently Dean and Head of the School of Clinical Medicine at the University of KwaZulu-Natal, Durban, South Africa. He studied at the University of Cape Town, specialising in Internal Medicine and subsequently hepatology, before moving to Durban as Professor of Medicine. He has a longstanding interest in medical education, and specifically in the cognitive aspects of clinical reasoning, an area in which he is currently supervising a number of research initiatives.
Modified essay question
Multiple-choice question
Short answer question
Objective structured clinical examination
Siemens G: Connectivism: Learning as Network-Creation. [ http://www.elearnspace.org/Articles/networks.htm ]
Siemens G: Connectivism: A learning theory for the digital age. Int J Instr Technol Distance Learn. 2005, 2: 3-10.
Google Scholar
Perkins DN, Salomon G: Learning transfer. International Encyclopaedia of adult education and training. Edited by: Tuijnman AC. 1996, Pergamon Press, Tarrytown, NY, 422-427. 2
Haskell EH: Transfer of learning: Cognition, Instruction, and Reasoning. 2001, Academic Press, New York
Spelke E: Initial Knowledge: Six Suggestions. Cognition on cognition. Edited by: Mehler J, Franck S. 1995, The MIT Press, Cambridge, MA US, 433-447.
Barnett SM, Ceci SJ: When and where do we apply what we learn? A taxonomy for far transfer. Psychol Bull. 2002, 128: 612-637.
Brown AL: Analogical Learning and Transfer: What Develops?. Similarity and Analogical Reasoning. Edited by: Vosniadou S, Ortony A. 1989, Cambridge University Press, New York, 369-412.
Gick ML, Holyoak KJ: Schema Induction and Analogical Transfer. 2004, Psychology Press, New York, NY US
Bloom BS: The Cognitive Domain. Taxonomy of Educational Objectives, Handbook I. 1956, David McKay Co Inc, New York
Anderson LW, Krathwohl DR, Airasian PW, Cruikshank KA, Mayer RE, Pintrich PR, Raths J, Wittrock MC: A Taxonomy for Learning, Teaching, and Assessing: a revision of Bloom's Taxonomy of Educational Objectives. 2001, Longman, New York
Anderson LW, Sosniak LA: Bloom's Taxonomy: A Forty-year Retrospective. Ninety-third yearbook of the National Society for the Study of Education: Part II. Edited by: Anderson LW, Sosniak LA. 1994, University of Chicago Press, Chicago IL
Conklin J: A taxonomy for learning, teaching, and assessing: a revision of Bloom's taxonomy of educational objectives. Educ Horiz. 2005, 83: 154-159.
Haladyna TM, Downing SM: A taxonomy of multiple-choice item-writing rules. Appl Meas Educ. 1989, 2: 37-51.
Haladyna TM: Developing and Validating Multiple-choice Test Items . Mahwah NJ: L. Erlbaum Associates; 1999.
Miller GE: The assessment of clinical skills/competence/performance. Acad Med. 1990, 65: S63-S67.
Dreyfus HL, Dreyfus SE, Athanasiou T: Mind over Machine: The Power of Human Intuition and Expertise in the Era of the Computer. 1986, Free Press, New York
Norcini JJ, Swanson DB, Grosso LJ, Webster GD: Reliability, validity and efficiency of multiple choice question and patient management problem item formats in assessment of clinical competence. Med Educ. 1985, 19: 238-247.
Taconnat L, Froger C, Sacher M, Isingrini M: Generation and associative encoding in young and old adults: The effect of the strength of association between cues and targets on a cued recall task. Exp Psychol. 2008, 55: 23-30.
Baddeley AD, Eysenck MW, Anderson M: Memory. 2010, Psychology Press, New York
Karpicke J, Grimaldi P: Retrieval-based learning: a perspective for enhancing meaningful learning. Educ Psychol Rev. 2012, 24: 401-418.
Rohrer D, Pashler H: Recent research on human learning challenges conventional instructional strategies. Educ Res. 2010, 39: 406-412.
Smith MA, Roediger HL, Karpicke JD: Covert retrieval practice benefits retention as much as overt retrieval practice. J Exp Psychol Learn Mem Cogn. 2013, 39: 1712-1725.
McDermott KB, Agarwal PK, D’Antonio L, Roediger HL, McDaniel MA: Both multiple-choice and short-answer quizzes enhance later exam performance in middle and high school classes. J Exp Psychol Appl. 2014, 20: 3-21.
Cutting MF, Saks NS: Twelve tips for utilizing principles of learning to support medical education. Med Teach. 2012, 34: 20-24.
Schuwirth LWT, Van der Vleuten CPM: General overview of the theories used in assessment: AMEE Guide No. 57. Med Teach. 2011, 33: 783-797.
Van der Vleuten CP, Schuwirth LW: Assessing professional competence: from methods to programmes. Med Educ. 2005, 39: 309-317.
Schuwirth L, Colliver J, Gruppen L, Kreiter C, Mennin S, Onishi H, Pangaro L, Ringsted C, Swanson D, Van der Vleuten C, Wagner-Menghin M: Research in assessment: Consensus statement and recommendations from the Ottawa 2010 Conference. Med Teach. 2011, 33: 224-233.
Schuwirth LWT, Van der Vleuten CPM: Programmatic assessment and Kane's validity perspective. Med Educ. 2012, 46: 38-48.
Case SM, Swanson DB: Constructing Written Test Questions for the Basic and Clinical Sciences. 2002, National Board of Medical Examiners, Philadelphia, 3
Norcini J, Anderson B, Bollela V, Burch V, Costa MJ, Duvivier R, Galbraith R, Hays R, Kent A, Perrott V, Roberts T: Criteria for good assessment: Consensus statement and recommendations from the Ottawa 2010 Conference. Med Teach. 2011, 33: 206-214.
Shepard LA: The role of assessment in a learning culture. Educ Res. 2000, 29: 4-14.
Coburn CA, Yerkes RM: A study of the behavior of the crow corvus americanus Aud. By the multiple choice method. J Anim Behav. 1915, 5: 75-114.
Yerkes RM, Coburn CA: A study of the behavior of the pig Sus Scrofa by the multiple choice method. J Anim Behav. 1915, 5: 185-225.
Brown W, Whittell F: Yerkes' multiple choice method with human adults. J Comp Psychol. 1923, 3: 305-318.
Yerkes RM: A New method of studying the ideational behavior of mentally defective and deranged as compared with normal individuals. J Comp Psychol. 1921, 1: 369-394.
Davidson C: Davidson CN: Now You See It: How the Brain Science of Attention Will Transform the Way We Live, Work, and Learn. 2011, Viking Press, New York
Frederiksen JR, Collins A: A Systems Approach to Educational Testing. Technical Report No. 2. 1990, Center for Technology in Education, New York
Guthrie JT: Testing higher level skills. J Read. 1984, 28: 188-190.
Nickerson RS: New directions in educational assessment. Educ Res. 1989, 18: 3-7.
Stratford P, Pierce-Fenn H: Modified essay question. Phys Ther. 1985, 65: 1075-1079.
Wass V, Van der Vleuten C, Shatzer J, Jones R: Assessment of clinical competence. Lancet. 2001, 357: 945.
Rotfield H: Are we teachers or job trainers?. Acad Mark Sci Q. 1998, 2: 2.
Crocker L, Algina J: Introduction to Classical & Modern Test Theory. 1986, Holt, Rinehart and Winston, Inc., Fort Worth, TX
Angoff W: Test reliability and effective test length. Psychometrika. 1953, 18: 1-14.
Palmer EJ, Devitt PG: Assessment of higher order cognitive skills in undergraduate education: modified essay or multiple choice questions? Research paper. BMC Med Educ. 2007, 7: 49-49.
Feletti GI, Smith EK: Modified essay questions: Are they worth the effort?. Med Educ. 1986, 20: 126-132.
Palmer EJ, Duggan P, Devitt PG, Russell R: The modified essay question: its exit from the exit examination?. Med Teach. 2010, 32: e300-e307.
Schuwirth LW, Van der Vleuten CPM: Different written assessment methods: what can be said about their strengths and weaknesses?. Med Educ. 2004, 38: 974-979.
Lukhele R, Thissen D, Wainer H: On the relative value of multiple-choice, constructed response, and examinee-selected items on two achievement tests. J Educ Meas. 1994, 31: 234-250.
Wainer H, Thissen D: Combining multiple-choice and constructed-response test scores: toward a Marxist theory of test construction. Appl Meas Educ. 1993, 6: 103-118.
Facione PA: The California Critical Thinking Skills Test--College Level. Technical Report #1. Experimental Validation and Content Validity. 1990, California Academic Press, Millbrae CA
Facione PA, Facione NC, Blohm SW, Giancarlo CAF: The California Critical Thinking Skills Test [Revised]. In Millbrae CA: California Academic Press; 2007.
Rodriguez MC: Construct equivalence of multiple-choice and constructed-response items: A random effects synthesis of correlations. J Educ Meas. 2003, 40: 163-184.
Falk B, Ancess J, Darling-Hammond L: Authentic Assessment in Action: Studies of Schools and Students at Work. 1995, Teachers College Press, United States of America
Rethans JJ, Norcini JJ, Baron-Maldonado M, Blackmore D, Jolly BC, LaDuca T, Lew S, Page GG, Southgate LH: The relationship between competence and performance: implications for assessing practice performance. Med Educ. 2002, 36: 901-909.
Wilkinson TJ, Frampton CM: Comprehensive undergraduate medical assessments improve prediction of clinical performance. Med Educ. 2004, 38: 1111-1116.
Baker EL: Standards for Educational and Psychological Testing. Sage Publications, Inc; 2012.
Eignor DR: The Standards for Educational and Psychological Testing. APA Handbook of Testing and Assessment in Psychology, Vol 1: Test Theory and Testing and Assessment in Industrial and Organizational Psychology. Edited by: Geisinger KF, Bracken BA, Carlson JF, Hansen J-IC, Kuncel NR, Reise SP, Rodriguez MC. 2013, American Psychological Association, Washington, DC, US, 245-250.
Eignor DR: Standards for the development and use of tests: The Standards for Educational and Psychological Testing. Eur J Psychol Assess. 2001, 17: 157-163.
Downing SM: Validity: on the meaningful interpretation of assessment data. Med Educ. 2003, 37: 830.
Messick S: The interplay of evidence and consequences in the validation of performance assessments. Educ Res. 1994, 23: 13-23.
Kuechler WL, Simkin MG: Why is performance on multiple-choice tests and constructed-response tests Not more closely related? theory and an empirical test. Decis Sci J Innov Educ. 2010, 8: 55-73.
Norman GR, Smith EK, Powles AC, Rooney PJ: Factors underlying performance on written tests of knowledge. Med Educ. 1987, 21: 297-304.
Bacon DR: Assessing learning outcomes: a comparison of multiple-choice and short-answer questions in a marketing context. J Mark Educ. 2003, 25: 31-36.
Kastner M, Stangla B: Multiple choice and constructed response tests: Do test format and scoring matter?. Procedia - Social and Behav Sci. 2011, 12: 263-273.
Nichols P, Sugrue B: The lack of fidelity between cognitively complex constructs and conventional test development practice. Educ Measurement: Issues Pract. 1999, 18: 18-29.
Bennett RE, Rock DA, Wang M: Equivalence of free-response and multiple-choice items. J Educ Meas. 1991, 28: 77-92.
Bridgeman B, Rock DA: Relationships among multiple-choice and open-ended analytical questions. J Educ Meas. 1993, 30: 313-329.
Thissen D, Wainer H: Are tests comprising both multiple-choice and free-response items necessarily less unidimensional. J Educ Meas. 1994, 31: 113.
Lissitz RW, Xiaodong H, Slater SC: The contribution of constructed response items to large scale assessment: measuring and understanding their impact. J Appl Testing Technol. 2012, 13: 1-52.
Traub RE, Fisher CW: On the equivalence of constructed- response and multiple-choice tests. Appl Psychol Meas. 1977, 1: 355-369.
Martinez ME: Cognition and the question of test item format. Educ Psychol. 1999, 34: 207-218.
Hee-Sun L, Liu OL, Linn MC: Validating measurement of knowledge integration in science using multiple-choice and explanation items. Appl Meas Educ. 2011, 24: 115-136.
Wilson M, Wang W-C: Complex composites: Issues that arise in combining different modes of assessment. Appl Psychol Meas. 1995, 19: 51-71.
Ercikan K, Schwartz RD, Julian MW, Burket GR, Weber MM, Link V: Calibration and scoring of tests with multiple-choice and constructed-response item types. J Educ Meas. 1998, 35: 137-154.
Epstein ML, Lazarus AD, Calvano TB, Matthews KA, Hendel RA, Epstein BB, Brosvic GM: Immediate feedback assessment technique promotes learning and corrects inaccurate first responses. Psychological Record. 2002, 52: 187-201.
Schuwirth LWT, Van der Vleuten CPM: Programmatic assessment: From assessment of learning to assessment for learning. Med Teach. 2011, 33: 478-485.
Bridgeman B, Morgan R: Success in college for students with discrepancies between performance on multiple-choice and essay tests. J Educ Psychol. 1996, 88: 333-340.
Bleske-Rechek A, Zeug N, Webb RM: Discrepant performance on multiple-choice and short answer assessments and the relation of performance to general scholastic aptitude. Assessment Eval Higher Educ. 2007, 32: 89-105.
Hakstian AR: The Effects of Type of Examination Anticipated on Test Preparation and Performance. J Educ Res. 1971, 64: 319.
Scouller K: The influence of assessment method on Students' learning approaches: multiple choice question examination versus assignment essay. High Educ. 1998, 35: 453-472.
Thomas PR, Bain JD: Contextual dependence of learning approaches: The effects of assessments. Human Learning: J Pract Res Appl. 1984, 3: 227-240.
Watkins D: Factors influencing the study methods of Australian tertiary students. High Educ. 1982, 11: 369-380.
Minbashian A, Huon GF, Bird KD: Approaches to studying and academic performance in short-essay exams. High Educ. 2004, 47: 161-176.
Yonker JE: The relationship of deep and surface study approaches on factual and applied test-bank multiple-choice question performance. Assess Eval Higher Educ. 2011, 36: 673-686.
Joughin G: The hidden curriculum revisited: a critical review of research into the influence of summative assessment on learning. Assess Eval Higher Educ. 2010, 35: 335-345.
Scouller KM, Prosser M: Students' experiences in studying for multiple choice question examinations. Stud High Educ. 1994, 19: 267.
Hadwin AF, Winne PH, Stockley DB, Nesbit JC, Woszczyna C: Context moderates students' self-reports about how they study. J Educ Psychol. 2001, 93: 477-487.
Birenbaum M: Assessment and instruction preferences and their relationship with test anxiety and learning strategies. High Educ. 2007, 53: 749-768.
Birenbaum M: Assessment preferences and their relationship to learning strategies and orientations. High Educ. 1997, 33: 71-84.
Smith SN, Miller RJ: Learning approaches: examination type, discipline of study, and gender. Educ Psychol. 2005, 25: 43-53.
Rabinowitz HK, Hojat M: A comparison of the modified essay question and multiple choice question formats: their relationship to clinical performance. Fam Med. 1989, 21: 364-367.
Paterson DG: Do new and old type examinations measure different mental functions?. School Soc. 1926, 24: 246-248.
Schuwirth LW, Verheggen MM, Van der Vleuten CPM, Boshuizen HP, Dinant GJ: Do short cases elicit different thinking processes than factual knowledge questions do?. Med Educ. 2001, 35: 348-356.
Tanner DE: Multiple-choice items: Pariah, panacea or neither of the above?. Am Second Educ. 2003, 31: 27.
Cilliers FJ, Schuwirth LW, van der Vleuten CP: Modelling the pre-assessment learning effects of assessment: evidence in the validity chain. Med Educ. 2012, 46: 1087-1098.
Jonassen DH, Strobel J: Modeling for Meaningful Learning. Engaged Learning with Emerging Technologies. Edited by: Hung D. 2006, Springer, Amsterdam, 1-27.
Derry SJ: Cognitive schema theory in the constructivist debate. Educ Psychol. 1996, 31: 163-174.
Kim MK: Theoretically grounded guidelines for assessing learning progress: cognitive changes in Ill-structured complex problem-solving contexts. Educ Technol Res Dev. 2012, 60: 601-622.
Mayer RE: Models for Understanding. Rev Educ Res. 1989, 59: 43-64.
Jonassen D, Strobel J, Gottdenker J: Model building for conceptual change. Interact Learn Environ. 2005, 13: 15-37.
Jonassen DH: Tools for representing problems and the knowledge required to solve them. Edited by Tergan S-O, Keller T. Berlin, Heidelberg: Springer; 2005:82–94.
Bogard T, Liu M, Chiang Y-H: Thresholds of knowledge development in complex problem solving: a multiple-case study of advanced Learners' cognitive processes. Educ Technol Res Dev. 2013, 61: 465-503.
Van Gog T, Ericsson KA, Rikers RMJP: Instructional design for advanced learners: establishing connections between the theoretical frameworks of cognitive load and deliberate practice. Educ Technol Res Dev. 2005, 53: 73-81.
Schmidt HG, Norman GR, Boshuizen HP: A cognitive perspective on medical expertise: theory and implication. Acad Med. 1990, 65: 611-621.
Schmidt HG, Rikers RMJP: How expertise develops in medicine: knowledge encapsulation and illness script formation. Med Educ. 2007, 41: 1133-1139.
Norman G, Young M, Brooks L: Non-analytical models of clinical reasoning: the role of experience. Med Educ. 2007, 41: 1140-1145.
Ericsson KA, Prietula MJ, Cokely ET: The Making of an Expert. Harv Bus Rev. 2007, 85: 114-121.
Hoffman RR: How Can Expertise be Defined? Implications of Research From Cognitive Psychology. Exploring Expertise. Edited by: Williams R, Faulkner W, Fleck J. 1996, University of Edinburgh Press, Edinburgh, 81-100.
Norman GR: Problem-solving skills, solving problems and problem-based learning. Med Educ. 1988, 22: 279-286.
Ifenthaler D, Seel NM: Model-based reasoning. Comput Educ. 2013, 64: 131-142.
Jonassen D: Using cognitive tools to represent problems. J Res Technol Educ. 2003, 35: 362-381.
Mayer RE, Wittrock MC: Problem-Solving Transfer. Handbook of Educational Psychology. Edited by: Berliner DC, Calfee RC. 1996, Macmillan Library Reference USA, New York, NY, 47-62.
Zhang J, Norman DA: Representations in distributed cognitive tasks. Cogn Sci. 1994, 18: 87-122.
Simon HA: Information-Processing Theory of Human Problem Solving. Handbook of Learning & Cognitive Processes: V Human Information. Edited by: Estes WK. 1978, Lawrence Erlbaum, Oxford England, 271-295.
Jensen JL, Woodard SM, Kummer TA, McDaniel MA: Teaching to the test…or testing to teach: exams requiring higher order thinking skills encourage greater conceptual understanding. Educ Psychol Rev. 2014, 26: 307-329.
Cohen-Schotanus J, Van der Vleuten CPM: A standard setting method with the best performing students as point of reference: practical and affordable. Med Teach. 2010, 32: 154-160.
Desjardins I, Touchie C, Pugh D, Wood TJ, Humphrey-Murto S: The impact of cueing on written examinations of clinical decision making: a case study. Med Educ. 2014, 48: 255-261.
Pretz JE, Naples AJ, Sternberg RJ: Recognizing, Defining, and Representing Problems. The Psychology of Problem Solving. Edited by: Davidson JE, Sternberg RJ. 2003, Cambridge University Press, New York, NY US, 3-30.
Schuwirth LWT, Schuwirth LWT, Van der Vleuten CPM: ABC of learning and teaching in medicine: written assessment. BMJ: British Med J (International Edition). 2003, 326: 643-645.
Download references
The author would like to thank Dr Veena Singaram for her insightful and challenging appraisal of the manuscript.
Authors and affiliations.
Clinical and Professional Practice Research Group, School of Clinical Medicine, University of KwaZulu-Natal, Durban, 4013, South Africa
Richard J Hift
You can also search for this author in PubMed Google Scholar
Correspondence to Richard J Hift .
Competing interests.
The author declares that he has no competing interests.
Below are the links to the authors’ original submitted files for images.
Authors’ original file for figure 2, authors’ original file for figure 3, rights and permissions.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License ( http://creativecommons.org/licenses/by/4.0 ), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver ( http://creativecommons.org/publicdomain/zero/1.0/ ) applies to the data made available in this article, unless otherwise stated.
Reprints and permissions
Cite this article.
Hift, R.J. Should essays and other “open-ended”-type questions retain a place in written summative assessment in clinical medicine?. BMC Med Educ 14 , 249 (2014). https://doi.org/10.1186/s12909-014-0249-2
Download citation
Received : 08 May 2014
Accepted : 07 November 2014
Published : 28 November 2014
DOI : https://doi.org/10.1186/s12909-014-0249-2
Anyone you share the following link with will be able to read this content:
Sorry, a shareable link is not currently available for this article.
Provided by the Springer Nature SharedIt content-sharing initiative
ISSN: 1472-6920
Open-ended questions are the questions asked that do not give the option of a yes/no answer, instead, they require full sentences. They usually signify the beginning of a dialogue.
If you sincerely want to connect on deeper levels and encourage other people to talk about themselves, you should ideally use open-ended questions to stimulate your conversation and get the ball rolling.
Open-ended questions are those which require more thought and more than a simple one-word answer. An open-ended question is designed to encourage a full, meaningful, and deliberate answer using the subject’s own knowledge and/or feelings.
It is the opposite of a closed-ended question, which encourages a short or single-word answer.
Read More: Close Ended Questions:Definition + [Questionnaire Examples]
Open-ended questions are used for interviews, with the caveat that there are those with no right or wrong answers. An interview question, for example, could be along the lines of a person asking an interviewee about their past work experience. As a requirement, Open-ended questions demand that the applicant offer more detail and demonstrate their ability to communicate effectively.
Open-ended questions can be used in examinations such that the student is required to provide a response. Unlike multiple-choice(close-ended) tests that do not allow much, if any, room for error, open-ended questions give the students room to convince the examiner. Usually, a test is composed of a few open-ended questions compared to the 50 to 100-question multiple-choice(close-ended) assessment.
For a Survey, open-ended questions are ideal for a number of reasons. First, they allow an infinite number of possible answers. They also collect more detail and the person administering the questions might even learn something they didn’t expect. For complex issues, open-ended questions ensure you get adequate answers. Lastly, open-ended questions for a survey encourage creative answers and self-expression and help you understand how your respondents think.
In transacting business, open-ended questions are essential for sales success. They allow reps to get inside the head of prospects and better understand their pain points. The right open-ended questions help ensure that reps are building rapport, uncovering pain points, establishing needs, and clearly articulating the value of their offering.
Since open-ended questions are designed to prompt long, detailed answers, here are a few tips that can help you ask open-ended questions better.
Open-ended questions allow you to better understand the respondent’s true feelings and attitudes about the survey subject. Close-ended questions, due to their limitations, do not give respondents the choice to truly give their opinions.
Open-ended questions allow respondents taking your survey to include more information, giving you, the researcher, more useful, contextual feedback. Close-ended questions provide none of those. The answers are short, concise, and very direct.
Open-ended questions in surveys solicit additional information to be contributed by respondents. They are sometimes also called infinite-response questions or unsaturated-type questions. Generally, close-ended questions require respondents to answer in just one or two words.
Open-ended questions cut down on two types of response error; respondents are not likely to forget the answers they have to choose from if they are given the chance to respond freely, and open-ended questions simply do not allow respondents to disregard reading the questions and just “fill in” the survey with all the same answers.
Since open-ended questions allow for obtaining extra information from the respondent, such as demographic information, surveys that use open-ended questions can be used more readily for secondary analysis by other researchers than can surveys that do not provide contextual information about the survey population.
If you’re looking for questions that allow someone to give a free-form answer, the open-ended questions are the choice. Even though close-ended questions are often good for surveys, because you get higher response rates because users don’t have to type so much, they don’t accomplish this.
A key benefit of open-ended questions is that they allow you to find more than you anticipate. People are more likely to share motivations that you didn’t expect and mention behaviors and concerns that you knew nothing about. When you ask people to explain things to you, they often reveal surprising mental models, problem-solving strategies, hopes, fears, and much more. Closed-ended questions stop the conversation and eliminate any surprises.
As much as Open-ended questions provide the most feedback, it is important to note that they are a lot harder to analyze. This is because, unlike close-ended questions that provide quantitative data, open-ended questions provide qualitative data.
There are a number of things you should note when interpreting Open-Ended Questionnaire Data, here are a few guides to help you on your way.
With the Short text field, best used for receiving short/single-line text-based answers such as names, location, and statements, you can set a minimum and maximum length of characters your users can input.
The Long text field, as provided by Formplus, is ideal for long answers such as addresses, comments, additional ideas, suggestions, messages, and short essay answers.
You can use Formplus to send form responses to Google Sheets instantly. The Google Sheets integration makes it easy to collaborate on documents and keep your team members up-to-date.
Formplus lets you store tabular data, such as a spreadsheet or database. Also in the Formplus Responses setting you can customize the email notification message, including the user’s response in the notification email sent. You can also receive submissions as a PDF/Doc attachment in your emails as well as enable the option to display images on your attachments.
Formplus has an unlimited file upload storage, you can submit files, photos, or videos via your online forms without any restriction to the size or number of files that can be uploaded. You can choose to store your received data in your cloud storage of choice. There is also a native Google Sheets integration, which lets you get survey responses updated into spreadsheets automatically created for each form.
Formplus’ easy-to-use form builder allows you to create powerful forms within minutes. Simply click or drag and drop your desired form fields into the builder. You can build any type of online form ranging from Contact Forms to Inventory Forms, Formplus has the tools to help you collect data seamlessly.
You can also monitor your survey performance and identify your traffic source and location with Formplus Analytics. With online form builder analytics, you can determine the number of times the open-ended survey was filled and the number of respondents who reached the abandonment rate. You can also find out the location of respondents as well as the type of device used by the respondent to complete the survey.
Start creating your Open-Ended Survey by signing up with either your Google, Facebook, or Email account. There’s a free plan with limited features you can use to get started.
Sign up to design your Open-Ended Survey with Formplus.
Input your Open-Ended Survey title and use the form builder choice options to start creating your Survey.
Beautify your Open-Ended Survey with Formplus Customisation features.
Edit your Open-ended survey settings for your specific needs
Set an introductory message to respondents before they begin the survey
Share links to your Open-ended survey page with respondents
View Responses to the Open-ended Survey
Toggle with the presentation of your summary from the options. Whether as a single, table or cards. In addition, you can make graphs from received responses, and translate these into charts and key metrics.
Let Formplus Analytics interpret your data from your Open-ended survey
You can also monitor your form performance and identify your traffic source and location with Formplus Analytics.
With online form builder analytics, you can determine:
In any circumstance, open-ended questions guarantee a much more effective result in communication. In the search for a complete and meaningful answer, you need to employ open-ended questions. The best part of Open-ended questions is that they prompt respondents to provide answers using their own words.
For a survey, open-ended questions provide a researcher with qualitative data that they can draw inferences from. On the whole, Open-ended questions make respondents include more information, including feelings, attitudes, and understanding of the subject matter.
Connect to Formplus, Get Started Now - It's Free!
You may also like:
Introduction Matrix questions are a type of survey question that allows respondents to answer multiple statements using rates in rows...
Simple guide on the difference between close and open ended questions. Where and how to use them.
This article outlines 25 great NPS survey questions to help you gather feedback from your customers
Ultimate guide to understanding close ended questions, examples, advantages and questionnaire examples in surveys
Collect data the right way with a versatile data collection tool. try formplus and transform your work productivity today..
Editorial team.
How to Answer Open-Ended Essay Questions. It's test time, and this one isn't multiple choice. Your teacher gives you a sheet of paper with a question on it. The only problem is, you can't immediately see a definite answer. It's time to pull it together and, at the very least, be able to sound like you know what you're talking about.
Read the question carefully. Make sure you understand what is being asked of you. Think of the different meanings of the specific words within the question.
Mull the question over before attempting to answer. Absorb it. Think about what you know or have learned about the topic. Taking this time can have a calming effect, which will help you to write a more cogent response.
Develop an opinion, if you haven't already. Your argument will be more convincing if you believe what you are writing. This decision will be your thesis. You don't have to take an extreme stance. If you are ambivalent about the topic, be prepared to address this in your essay. Being able to cite arguments for and against either side will make you appear to have a better understanding of the material.
Jot down an outline. A disorganized essay, despite its content, will not get your point across. Make sure to address any possible objections to your thesis early in the essay, and save your strongest arguments for the end.
Write your essay, referring to your outline. Hopefully, after having taken time to develop a thesis, mull over the topic, and sketch an outline, your thoughts will flow from you into clear writing.
Read your essay over, if you have time. Focus on high-order content, such as ideas and themes. Make sure you have thoroughly answered the question asked of you. If you are not under strict time constraints, take the time to check for proper grammar and spelling.
This article was written by the CareerTrend team, copy edited and fact checked through a multi-point auditing system, in efforts to ensure our readers only receive the best information. To submit your questions or ideas, or to simply learn more about CareerTrend, contact us [here](http://careertrend.com/about-us).
Regardless of how old we are, we never stop learning. Classroom is the educational resource for people of all ages. Whether you’re studying times tables or applying to college, Classroom has the answers.
© 2020 Leaf Group Ltd. / Leaf Group Media, All Rights Reserved. Based on the Word Net lexical database for the English Language. See disclaimer .
Skip navigation
Open-ended vs. closed questions in user research.
January 26, 2024 2024-01-26
When conducting user research, asking questions helps you uncover insights. However, how you ask questions impacts what and how much you can discover .
Open-ended vs. closed questions, why asking open-ended questions is important, how to ask open-ended questions.
There are two types of questions we can use in research studies: open-ended and closed.
Open-ended questions allow participants to give a free-form text answer. Closed questions (or closed-ended questions) restrict participants to one of a limited set of possible answers.
Open-ended questions encourage exploration of a topic; a participant can choose what to share and in how much detail. Participants are encouraged to give a reasoned response rather than a one-word answer or a short phrase.
Examples of open-ended questions include:
Note that the first two open-ended questions are commands but act as questions. These are common questions asked in user interviews to get participants to share stories. Questions 3 and 4 are common questions that a usability-test facilitator may ask during and after a user attempts a task, respectively.
Closed questions have a short and limited response. Examples of closed questions include:
Strictly speaking, questions 3 and 4 would only be considered “closed” if they were accompanied by answer options, such as (a) never, (b) once, (c) two times or more. This is because the number of times and days could be infinite. That being said, in UX, we treat questions like these as closed questions.
In the dialog between a facilitator and a user below, closed questions provide a short, clarifying response, while open-ended questions result in the user describing an experience.
T | |
Closed questions are heavily utilized in surveys because the responses can be analyzed statistically (and surveys are usually a quantitative exercise). When used in surveys, they often take the form of multiple-choice questions or rating-scale items , rather than open-text questions. This way, the respondent has the answer options provided, and researchers can easily quantify how popular certain responses are. That being said, some closed questions could be answered through an open-text field to provide a better experience for the respondent. Consider the following closed questions:
Both questions could be presented as multiple-choice questions in a survey. However, the respondent might find it more comfortable to share their industry and gender in a free-text field if they feel the survey does not provide an option that directly aligns with their situation or if there are too many options to review.
Another reason closed questions are used in surveys is that they are much easier to answer than open-ended ones. A survey with many open-ended questions will usually have a lower completion rate than one with more closed questions.
Closed questions are used occasionally in interviews and usability tests to get clarification and extra details. They are often used when asking followup questions. For example, a facilitator might ask:
Closed questions help facilitators gather important details. However, they should be used sparingly in qualitative research as they can limit what you can learn.
The greatest benefit of open-ended questions is that they allow you to find more than you anticipate. You don’t know what you don’t know. People may share motivations you didn’t expect and mention behaviors and concerns you knew nothing about. When you ask people to explain things, they often reveal surprising mental models , problem-solving strategies, hopes, and fears.
On the other hand, closed questions stop the conversation. If an interviewer or usability-test facilitator were to ask only closed questions, the conversation would be stilted and surface-level. The facilitator might not learn important things they didn’t think to ask because closed questions eliminate surprises: what you expect is what you get.
When you ask closed questions, you may accidentally reveal what you’re interested in and prime participants to volunteer only specific information. This is why researchers use the funnel technique , where the session or followup questions begin with broad, open-ended questions before introducing specific, closed questions.
Not all closed questions are leading. That being said, it’s easy for a closed question to become leading if it suggests an answer.
The table below shows examples of leading closed questions . Reworking a question so it’s not leading often involves making it open-ended, as shown in column 2 of the table below.
One way to spot a leading, closed question is to look at how the question begins. Leading closed questions often start with the words “did,” “was,” or “is.” Open-ended questions often begin with “how” or “what.”
New interviewers and usability-test facilitators often struggle to ask enough open-ended questions. A new interviewer might be tempted to ask many factual, closed questions in quick succession, such as the following:
However, these questions could be answered in response to a broad, open-ended question like Tell me a bit about yourself .
When constructing an interview guide for a user interview, try to think of a broad, open-ended version of a closed question that might get the participant talking about the question you want answered, like in the example above.
When asking questions in a usability test, try to favor questions that begin with “how,” or “what,” over “do,” or “did” like in the table below.
|
|
Another tip to help you ask open-ended questions is to use one of the following question stems :
Finally, you can ask open-ended questions when probing. Probing questions are open-ended and are used in response to what a participant shares. They are designed to solicit more information. You can use the following probing questions in interviews and usability tests.
Ask open-ended questions in conversations with users to discover unanticipated answers and important insights. Use closed questions to gather additional small details, gain clarification, or when you want to analyze responses quantitatively.
Please accept marketing cookies to view the embedded video. https://www.youtube.com/watch?v=LpV3tMy_WZ0
Open vs. Closed Questions in User Research
Always Pilot Test User Research Studies
Kim Salazar · 3 min
Level Up Your Focus Groups
Therese Fessenden · 5 min
Inductively Analyzing Qualitative Data
Tanner Kohler · 3 min
Field Studies Done Right: Fast and Observational
Jakob Nielsen · 3 min
Should You Run a Survey?
Maddie Brown · 6 min
The Funnel Technique in Qualitative User Research
Maria Rosala and Kate Moran · 7 min
Card Sorting: Pushing Users Beyond Terminology Matches
Samhita Tankala and Jakob Nielsen · 5 min
Card Sorting: Uncover Users' Mental Models for Better Information Architecture
Samhita Tankala and Katie Sherwin · 11 min
The Diverge-and-Converge Technique for UX Workshops
Therese Fessenden · 6 min
surveys | December 18, 2019
Daniel Ndukwu
Customer research is a large discipline with multiple methods to get the right information from your audience or customer base.
Surveys are among the most effective ways to get deep insights from your most engaged users. It helps you understand how they feel about specific topics and give you perspectives – through open-ended questions – you might have otherwise missed.
These insights, also known as the voice of the customer , can expand your marketing , improve your products, and cancel out objections. In a world of choice, this is becoming even more important.
In this article, you’ll learn what open-ended questions are, their advantages, how to use them, and solid examples to make them easier to implement.
Table of Contents
Open-ended questions are a type of unstructured survey question that allows the respondent more room to reply in an open text format thereby providing the opportunity to give more detailed answers. The only limitation usually imposed is a character limit so open-ended questions can be divided into long answer and short answer questions.
Put another way, a respondent can draw on their knowledge, feelings, and understanding of the question and topic to give more insightful answers. They’re not limited by preset question options.
An example of an open-ended question could be “how do you feel about your new job?”
Research from Vision Critical found that 87% of consumers want to have a say in a company’s products and services. Open-ended questions give them the opportunity to share information in a way that close ended questions don’t.
As shared in the last section, open-ended questions are free-form and allow respondents to use an open text format to give replies. They’re able to say whatever they want in response to your questions.
Close ended questions, on the other hand, are structured and have a preset group of questions a respondent can answer. Though they can still help you, you’re not able to use the voice of the customer to inform your decisions.
Each one has merits and demerits. For example, an open-ended question allows you to probe much deeper but a close ended question allows you to get concise information that can be quantified. It’s much easier to quantify yes or nos than a paragraph of text.
A relatable example comes from the standardized tests most of us took in school. They usually had two parts – the multiple-choice questions which are equivalent to close ended questions and the essay questions which are equal to open-ended questions.
A close ended question: Yes or No – Was George Washington was the first president of the United States?
An open-ended question: From the perspective of the British, what was the cause of the Revolutionary War?
As you can see from the examples, the open-ended survey questions will give you a look into the thought process of your customers.
Some advantages are obvious while others aren’t but they tend to be more important than the disadvantages because the responses you get have so many direct tangible uses for your business. A few of the advantages include:
With a series of multiple-choice questions, respondents can quickly scan and choose an answer. That answer may or may not be indicative of how they feel. Open-ended questions force your respondents to slow down long enough to consider the question and give a thoughtful answer.
Even if they give a short answer that doesn’t shed much light on the situation, it’s still helpful. It tells you the respondent’s answers shouldn’t have as much weight.
The internet is crowded. The lifespan of a Tweet is about 30 minutes and that of a Facebook post is roughly 1 hour. Anything you say is lost in a short amount of time.
With surveys and open-ended questions, you give your customers an opportunity to voice their opinions and create in a way that can create change in your organization. If they like what you’re doing and care about your products then they’ll take the time to give you useful feedback.
Close ended questions are notorious for only giving you half the answer you need. For example, if you ask a customer “how was your experience with us today?” and they answer “disappointing”, there’s no room to ask them why. This can leave you wondering if you have a real problem.
An open-ended question gives them the opportunity to tell you it was disappointing and lay out the reasons why. With that information, you can determine if it was an isolated incident or something that demands immediate attention.
Open-ended questions were built to deliver qualitative information and, like we talk about in our free course , the more detailed the information you get from a respondent the more engaged they are.
That information is indicative of your hottest buyer segment and the details they reveal will help you create better messages, identify your ideal target market , and otherwise make the right decisions in your business.
The key to using the extra information these types of survey questions can give you is to look for patterns in the data. If one person says something then you may or may not be on the verge of a breakthrough. If five or ten people say something similar then there’s promise.
These questions lend themselves well to qualitative research. That means they should be used when quality is more important than the quantity of data.
In other words, it’s used when you want to use the answers to find deep insights into the mind of your target audience. For example, you’d use them in the following situations:
Unfortunately, you can’t always use open-ended questions. Sometimes, a quick answer is ideal. For example, you want to know if someone has heard of your brand before. There’s no need to wax poetic about the possible reasons why they’ve not heard of it. Yes or no will do.
There are multiple situations in which free-form questions would do more harm than good. A few of them include:
There is a right way and a wrong way to ask questions – especially open-ended questions. Like all surveys, you want to collect unbiased data so you can make decisions that move the needle in the right direction. The wording of your questions can have a big impact on how its perceived by your respondent.
Surveys are not the time to convince someone of your view or to purposely elicit a positive response. Avoid wording that would predispose someone to answer positively or negatively.
For example, a question like “we’re considered a market leader and have over 10,000 customers, what do you think about our company?” is biased. It predisposes the respondent to give you positive feedback.
Consider talking to a team member or an impartial third party and showing them your questions to ensure they’re not biased. Put yourself in the shoes of the respondent and ask yourself if the question makes you feel positively or negatively towards the person asking.
This method is a staple of consumer research. The most effective surveys ask a close-ended question and, depending on how the respondent answers, an open-ended question is used as a follow-up. It helps focus the respondent and bring out insights that would otherwise be missed by a close-ended question.
Another benefit of using these two questions together stems from getting qualitative and quantitative answers. You’re able to say X people were dissatisfied with the product and X people were satisfied. For the ones who were dissatisfied, these are the reasons and places where we can improve.
At times, it can be difficult to determine if a closed or open question will be better for your needs. There’s a quick way to determine the best type. If you want the reasons behind an answer then use open-ended. If you want the raw answer without explanations then use the close ended questions.
Of course, this should be determined on a case by case basis. When in doubt, it may be a better idea to change the question or exclude it altogether. It’s more important to get clean data.
Phrase your questions so they’ll help you understand the reasons and emotions behind an answer. Instead of “How would you describe your support experience today?” Ask “how do you feel about your support experience today?”
The difference is subtle but it can help you understand the emotions associated with an experience or product. If it’s a negative sentiment then you can take steps to change that. If it was a positive sentiment then you can focus on doubling down on what’s working.
1. how does x make you feel.
This question leans towards an emotional response instead of a purely objective one. It’s helpful when finding marketing copy that incorporates the voice of the customer.
This question is useful because it helps reveal psychographic information and can also help you uncover different ways to position your products. For example, you can be the perfect widget for bike enthusiasts.
This works on two separate levels. You can find out which advertising channels are working and the reason why people are seeking you out. This will help refine your messaging.
The question above reveals unbiased information about how your products are perceived. You’re asking the customer to say what they think is good (or bad) about your products.
This question is direct and assumes that there’s room for improvement in your products and services. Use with caution because it may force your respondent to find problems where none exist.
This, again, is a direct question that may force users to mention things they don’t truly like. Use with caution.
Questions like these help you derive insights that make your products fit seamlessly into the lives of your target market. If your people like to work out at home, you can create products that cater to that preference. If they like to travel by road, you can create relevant products.
The X here can be general or specific depending on what you chose to focus on. For example, “what do you like about our customer service or what do you like about our company?” One of the questions gives a broad answer and the other is focused.
The opposite perspective of the previous question open-ended question.
These questions focus on the direct improvement of a product or service. For example, what can we do to make your support experience better?
This question may not be ideal for a standard questionnaire because it works best in real-time. Instead, you can use it in your live chat or chatbots to engage people at the point of purchase.
There are multiple ways to go about customer research. One of the most powerful and inexpensive is surveys.
They can give you deep insights from a large number of people in a relatively short amount of time. This article has gone through everything you need to know to make effective open-ended questions to improve your business and grow your audience.
Let me know what you think in the comments and don’t forget to share it.
What is an open-ended question.
Open-ended questions are a type of unstructured survey question that allows the respondent more room to reply in an open text format and provides the opportunity to give more detailed answers
This depends on the situation and your goals. Open-ended questions give more insights but close ended can help with quantification of responses.
Good evening, My name is Cristina Raffaghello, Adjunct Professor at Eastern Piedmont University in Vercelli and online e-campus University Novedrate, Italy. In my opinion, your article is useful for teaching, as well. Do you think I may use in my course of Germanic Philology? This discipline regards the origin and development of Germanic languages (English, German, Frisian Gothic and Islandic) from linguistical, juridical, historical and literary viewpoints. I thank you for your attention. Cheers, Cristina
Of course Cristina, you’re free to use it to teach your students.
I want to write research proposal on knowledge and practices on malnutrition amongst women can I use open or closed questionnaire
Both, of course. It depends on the kind of data you’re looking for and how well-versed you are on the topic before conducting the research.
This is so interesting
Inspiring, educative. I am doing a qualitative research study, and I find both open and closed.-ended questions fitting
Your email address will not be published. Required fields are marked *
This site uses Akismet to reduce spam. Learn how your comment data is processed .
Learn / Blog / Article
Back to blog
Unless you’re a mind reader, the only way to find out what your users are thinking is to ask them. That's what surveys are for.
But the way you ask a question often determines the kind of answer you get—and one of the first decisions you have to make is: are you going to ask an open-ended or a closed-ended question?
Reading time.
Understanding the difference between open-ended and close-ended questions helps you ask better, more targeted questions, so you can get actionable answers. The question examples we cover in this article look at open- and closed-ended questions in the context of a website survey, but the principle applies across any type of survey you may want to run.
Start from the top or skip ahead to
What’s the difference between open-ended and closed-ended questions?
4 tips on how to craft your survey questions for a maximum response rate
5 critical open-ended questions to ask customers
When to ask open-ended questions vs. closed-ended questions
Open-ended questions are questions that cannot be answered with a simple ‘yes’ or ‘no’, and instead require the respondent to elaborate on their points.
Open-ended questions help you see things from a customer’s perspective as you get feedback in their own words instead of stock answers. You can analyze open-ended questions using spreadsheets , view qualitative research and data analysis trends, and even spot elements that stand out with word cloud visualizations.
Closed-ended questions are questions that can only be answered by selecting from a limited number of options, usually multiple-choice questions with a single-word answer (‘yes’ or ‘no’) or a rating scale (e.g. from strongly agree to strongly disagree).
Closed-ended questions give limited insight, but can easily be analyzed for quantitative data . For example, one of the most popular closed questions in market research is the Net Promoter Score® (NPS) survey, which asks people “How likely are you to recommend this product/service on a scale from 0 to 10?” and uses numerical answers to calculate overall score trends. Check out our NPS survey template to see this closed-ended question in action.
Let’s take a look at the examples of open-ended questions vs. closed-ended questions above.
All the closed questions in the left column can be responded to with a one-word answer that gives you the general sentiment of each user and a few useful data points about their satisfaction, which help you look at trends and percentages. For example, did the proportion of people who declared themselves happy with your website change in the last three, six, or 12 months?
The open-ended questions in the right column let customers provide detailed responses with additional information so you understand the context behind a problem or learn more about your unique selling points . If you’re after qualitative data like this, the easy way to convert closed-ended into open-ended questions is to consider the range of possible responses and re-word your questions to allow for a free-form answer.
💡 Pro tip : when surveying people on your website with Hotjar Surveys , our Survey Logic feature lets you ask follow-up questions that help you find out the what and the why behind your users’ actions.
For more inspiration, here are 20+ real examples of open- and closed-ended questions you can ask on your website, along with a bunch of free pre-built survey templates and 50+ more survey questions to help you craft a better questionnaire for your users.
Or, take advantage of Hotjar’s AI for Surveys , which generates insightful survey questions based on your research goal in seconds and prepares an automated summary report with key takeaways and suggested next steps once results are in.
Use Hotjar to build your survey and get the customer insights you need to grow your business.
It’s often easy to lead your customers to the answer you want, so make sure you’re following these guidelines:
Some customers may find it hard to leave negative feedback if your questions are worded poorly.
For example, “We hope there wasn’t anything bad about your experience with us, but if so, please let us know” is better phrased neutrally as “Let us know if there was anything you’d like us to do differently.” It might sting a little to hear negative comments, but it’s your biggest opportunity to really empathize with customers and fuel your UX improvements moving forward.
“You bought 300 apples over the past year. What's your favorite fruit?” is an example of a leading question . You just planted the idea of an apple in your customers' mind. Valuable survey questions are open and objective—let people answer them in their own words, from their own perspective, and you’ll get more meaningful answers.
Tacking “and why?” on at the end of a question will only give you simple answers. And, no, adding “and why?” will not turn closed-ended questions into open-ended ones!
Asking “What did you purchase today, and why?” will give you an answer like “3 pairs of socks for a gift” (and that’s if you’re lucky), whereas wording the question as “Why did you choose to make a purchase today?” allows for an open answer like, “I saw your special offer and bought socks for my niece.”
Not many folks love filling in a survey that’s 50 questions long and takes an hour to complete. For the most effective data collection (and decent response rates), you need to keep the respondents’ attention span in mind. Here’s how:
Keep question length short : good questions are one-sentence long and worded as concisely as possible
Limit the number of questions : take your list of planned questions and be ruthless when narrowing them down. Keep the questions you know will lead to direct insight and ditch the rest.
Show survey progress : a simple progress bar, or an indication of how many questions are left, motivates users to finish your survey
Now that you know how to ask good open-ended questions , it’s time to start putting the knowledge into practice.
To survey your website users, use Hotjar's feedback tools to run on-page surveys, collect answers, and visualize results. You can create surveys that run on your entire site, or choose to display them on specific pages (URLs).
Different types of Hotjar surveys
As for what to ask—if you're just getting started, the five open-ended questions below are ideal for any website, whether ecommerce or software-as-a-service:
If you missed the expectations set by a customer, you may have over-promised or under-delivered. Ask users where you missed the mark today, and you’ll know how to properly set, and meet, expectations in the future. An open platform for your customers to tell you their pain points is far more valuable for increasing customer satisfaction than guessing what improvements you should make. Issues could range from technical bugs to lack of product range.
An open “How did you find out about us?” question leaves users to answer freely, without leading them to a stock response, and gives you valuable information that might be harder to track with traditional analytics tools.
We have a traffic attribution survey template ready and waiting for you to get started.
A “What is stopping you?” question can be shown on exit pages ; the open-form answers will help you identify the barriers to conversion that stop people from taking action.
Questions like this can also be triggered in a post-purchase survey on a thank you or order confirmation page. This type of survey only focuses on confirmed customers: after asking what almost stopped them, you can address any potential obstacles they highlight and fix them for the rest of your site visitors.
Finding out the concerns and objections of potential customers on your website helps you address them in future versions of the page they’re on and the products they’ll use. It sounds simple, but you’ll be surprised by how candid and helpful your users will be when answering this one.
Do you want to gather feedback on your product specifically? Learn what to improve and understand what users really think with our product feedback survey template and this expert advice on which product questions to ask when your product isn't selling.
Learning what made a customer click ‘buy now’ or ‘sign up’ helps you identify your levers. Maybe it’s low prices, fast shipping, or excellent customer service—whatever the reason, finding out what draws customers in and convinces them to stay helps you emphasize these benefits to other users and, ultimately, increase conversions.
Whether you’re part of a marketing, product, sales, or user research team, asking the right questions through customer interviews or on-site surveys helps you collect feedback to create better user experiences and increase conversions and sales.
The type of question you choose depends on what you’re trying to achieve:
Ask a closed-ended question when you want answers that can be plotted on a graph and used to show trends and percentages. For example, answers to the closed-ended question “Do you trust the information on [website]?” helps you understand the proportion of people who find your website trustworthy versus those who do not.
Ask an open-ended question when you want in-depth answers to better understand your customers and their needs , get more context behind their actions, and investigate the reasons behind their satisfaction or dissatisfaction with your product. For example, the open-ended question “If you could change anything on this page, what would it be?” allows your customers to express, in their own words, what they think you should be working on next.
Not only is the kind of question you ask important—but the moment you ask it is equally relevant. Hotjar Surveys , our online survey tool , has a user-friendly survey builder that lets you effortlessly craft a survey and embed it anywhere on your web page to ask the right questions at the right time and place.
User research
Whether you’re running focus groups for your pricing strategy or conducting usability testing for a new product, user interviews are one of the most effective research methods to get the needle-moving insights you need. But to discover meaningful data that helps you reach your goals, you need to connect with high-quality participants. This article shares five tips to help you optimize your recruiting efforts and find the right people for any type of research study.
Hotjar team
After the thrill of a successful user interview, the chore of transcribing dialogue can feel like the ultimate anticlimax. Putting spoken words in writing takes several precious hours—time better invested in sharing your findings with your team or boss.
But the fact remains: you need a clear and accurate user interview transcript to analyze and report data effectively. Enter automatic transcription. This process instantly transcribes recorded dialogue in real time without human help. It ensures data integrity (and preserves your sanity), enabling you to unlock valuable insights in your research.
Shadz Loresco
Customer interviews uncover your ideal users’ challenges and needs in their own words, providing in-depth customer experience insights that inform product development, new features, and decision-making. But to get the most out of your interviews, you need to approach them with empathy. This article explains how to conduct accessible, inclusive, and—above all—insightful interviews to create a smooth (and enjoyable!) process for you and your participants.
Created by TestInvite / May, 2024
In open-ended questions, the candidate is asked not to choose an answer, but to create it. How the candidate will create the answer is determined by the “answering type” of that open-ended question. Open-ended questions answered by writing text or numbers can be evaluated automatically by the assessment system with the automatic evaluation rules that can be determined in advance in the system, and the answers given can also be evaluated by the tutors. If preferred, comments can be added to the answers and shared with candidates.
Open ended question types each type of which provides different tools for candidates to construct their answer, can be categorized according to their answering methods. Candidates can compose their answers by typing, speaking, writing code, uploading files, taking photo/screenshots and filling in tables, in accordance with the type of open ended question type you choose.
Short answer type questions.
Candidates are given a one-line text field where they can write their answers.
Candidates are given a multiple-lines text field where they can write their answer.
The given text field can be made into a more easily understandable and useful format for the candidate by using many methods.
If you want candidates to give an answer in a certain language, but you think that they may not have a keyboard compatible with that language, a virtual keyboard can be added to the candidates' screen in the relevant language. In this way, candidates can type their answers using the keys on the virtual keyboard provided.
Virtual keyboard languages supported by the examination system:
If you want candidates to give an answer using the keyboard, but you want to provide a virtual keyboard in order to ensure some characters that you think may be missing on the keyboard, you can activate the auxiliary keypads.
If you want the candidate to type numbers as the answer, you can activate a virtual keypad consisting of numbers, point or comma.
By defining a regular expression (Regular Expression - regex), you can enforce that candidates' response conforms to a rule. If candidates write a text not complying with the rule, you can determine how you will alert them.
The answer given by the candidate by writing to a question can be evaluated automatically by the system through assessing it with a rule defined by the admin.
You can define a special function that will automatically evaluate the candidate's answer. In this case, the answer given by the candidate will be automatically evaluated by the function.
If you want the question to be answered by writing a number, an open ended question answered by writing a number can be preferred. In such case, the candidate can only write one number in the answer space.
The text field of the answer can be made into a more easily understandable and useful format for the candidate by employing several methods.
When the candidate writes a number to answer the question, it can be determined which punctuation marks to be used as thousand and precision separator.
The maximum precision of the numbers can be specified by the admin. In this way, the number that the candidate will write can consist of as many digits after the comma (or after the point) as admin determines.
By defining a regular expression (Regular Expression - regex), you can enforce that candidates2 answers conform to a rule. If a text that does not comply with the rule is written, you can determine how to alert candidates.
The answer can be evaluated automatically through comnparing it with a rule defined by the admin.
You can write a function that will evaluate automatically the candidate's answer. In this case, the answer given by the candidate will be evaluated automatically by the function.
You can create a table, specify the number of rows and columns in the table, and open spaces that the candidate can answer by typing in the desired cells of the table. You can define a single-line text space or a multi-line text space for each cell in the table.
You can add fixed contents to cells in the table, in order to render it more understandable.
You can add a heading to each row and column in the table. In this way, you can make your table more clear.
Candidates gives an answer to the question by creating content. They use the rich content editor to create content.
Following permissions can be granted in the answer creation process:
You can provide the candidate with a virtual keyboard in the question creation process.
Candidates answer the question by recording their voice. For this, they are provided with a voice recording application allowing them to record their voice while speaking.
Candidates answer the question by recording their voice and image together via the webcam on their device. For this, they are provided with a video recording application where they can record by their camera.
Candidates answer the question by writing code in a programming language determined by admin. A code editor compatible with the chosen programming language is provided so that candidates can write their answers.
In the code editor which is provided, you can show an initial content. For this, you can save the question by adding the code you want in the editor.
You can choose among more than 50 language options such as TypeScript, JavaScript, C#, Java, SQL, Python, Ruby, PHP, C, C++ for the code editor.
You can make candidates answer the question by uploading a file. Candidates upload the file they prepared as an answer by using the file upload tool.
You can make candidates answer the question by uploading a screenshot on their computer, or by taking a photo via a camera connected to their devices.
Answers given to open-ended questions can be evaluated by tutors after the exams are completed. When an exam consisting of open-ended questions is completed, the system indicates how many questions need to be evaluated in the relevant exam.
After the answers are evaluated, the exam report is automatically updated.
There are 2 methods for evaluating the answers of open-ended questions. These methods can be determined from the test settings either to effect all questions, or one by one for each question in the result reports.
A commentary content can also be included in the evaluation report of the open-ended questions. The comment may include an explanation of the evaluation, as well as listing the correct and incorrect factors in the answer, clarifying which parts of the answer are incorrect and why they are, and how the correct answer should be.
The results of an exam that is created with open-ended questions can be shared with the candidates after the evaluation process. Candidates can see their answer to each question, read the evaluation score they have received, and the evaluator's comment if it is wanted to be shown.
Want to learn more about TestInvite
Cookie notice.
We use cookies to improve your experience on our website and the performance of our marketing activities. for more info please visit: Privacy policy
Vice President, Innovation, Profiles Division
Open-ended questions are questions that encourage survey respondents to provide answers with depth and nuance rather than a simple “yes” or “no”. There are various goals to keep in mind when writing open-ended questions. In most cases, however, goals are not about the volume of feedback – they are about the quality. Whether you are trying to get respondents to be analytical, creative, or spontaneous, the biggest challenge you face is encouraging people to think and think in the right sort of ways.
Most respondents spend 15 seconds answering the average open-ended question and provide an average of five words. Five words might be enough if it constitutes meaningful feedback, but often it’s humdrum verbiage, which is a bit of a nightmare to analyse.
If you were to ask people to watch an ad and write down what they thought of it, the most common response would be "it was OK." The second- and third-most common remarks would be, "I liked it" and "I didn't like it," and the fourth would be "I don't know." These responses are clearly not a great deal of value, so you might as well have asked a yes/no question, “did you like it or not.”
Open-ended questions are crucial in information gathering. They elicit responses that might not surface with multiple choice, scale-based, or short answer questions. These questions allow for a range of answers, giving us a broader perspective.
By encouraging respondents to narrate their experiences, we gain insight into individual viewpoints. The variations, commonalities, themes, and unexpected insights that surface help create a well-rounded understanding of the subject matter.
In a data-driven world, recognizing the role of open-ended questions is vital. They infuse life into hard facts, offering context, depth, and, ultimately, a richer meaning to our research.
Here are 9 of our best practices for eliciting more thoughtful feedback with open-ended questions. Find the complete guide, with all 23 tips, here .
Bring respondents into the problem you are trying to solve if possible. Instead of asking "Why do you like this brand?" you might say, "This brand is a lot more popular amongst some people than others and we are trying to understand why."
Using language that challenges respondents to do something can be an extremely powerful weapon to encourage feedback. Instead of asking "What brands come to mind?" you might re-phrase that to say, “How many brands can you guess?" Or, instead of saying "What words come to mind when you think of this brand," you might say, "can you think of the most popular words people associate with this brand?"
Kantar has discovered that this type of approach can sometimes double the level of feedback as well.
Asking people to write not more than “X” number of words on a topic and using a word counter so they can see how much they have written is a great way to extend the volume of open-ended feedback. The “X” limit might be 3 words or 300, depending on the circumstances.
One of the reasons people can be reluctant to give open-ended feedback in surveys is because they are worried about their spelling and how well they can write. There are a couple of ways of dealing with this. First, make it clear that you are not worried about it. Secondly, we have found that showing a range of informal feedback, with not necessarily perfect spelling, encourages respondents to be less concerned about their personal style of delivering feedback.
People put far less effort into answering open-ended questions at the end of a survey. There are often trigger points you will notice wherein people tend to give up, and their desire to get to the end of a survey cuts in so strongly that their responses to open-ended questions dry up.
One of the best tricks to get people to put more thought into answering an open-ended question is to ask a question first that seeds a thinking process. For example, asking “Do you think your bank is ‘perfect?” This then seeds a question about what a bank could do to improve its service.
Adding a rule to a question can make an open-ended question a whole lot more interesting and fun for respondents to answer, thus triggering more thoughtful feedback. The easiest way to do this is by placing word limits - e.g., “review this film in 3 words.”
Always remember you are one human being talking to another, and most people's favourite topic is themselves. Try, wherever possible, to frame the question in a personal context. Instead of asking “Why do you like this brand?” you might ask, "If a friend asked you what you thought of this brand, what would you say?"
The word “imagine” is a magical word to use as a means of getting people to think. Kantar Profiles did an experiment in which we asked one group of people to make a list of their favourite shops. We asked another group, “Imagine you could design your perfect shopping centre. What shops would you have in it?" This increased the feedback 4-fold.
What can you achieve if you get it all right?
In experiments where we redesign the way we ask an open-ended question, the biggest uplift in feedback we have ever achieved has been a 6-fold increase from 17 words to 103 worlds.
This used a combination of techniques listed above. With the right thought and approach, you can easily achieve double or treble the volume of feedback by re-engineering the wording of a question.
But note that, as we said at the start, it is often not about volume of words so much as about the clarity of thinking these words stimulate. We find that with increases in volume feedback, is also linked to more thoughtful feedback.
All this begs the question about what you do with all this open-ended feedback. And that can be overwhelming. Some clever ways of automatically processing open ended feedback have emerged with social media text analytic software and natural language computational techniques. You can also try Kantar Profiles’ Taxonomy Builder , a free excel tool, here.
To read all 23 tips for collecting more thoughtful responses from open-ended questions in online surveys, use the form below to download the complete guide. For more on this topic, watch the on-demand webinar with Jon Puleston .
If the form below doesn't appear click here.
Circumstances that require open-ended questions, a follow-up question that would add value to a counseling session, an example of a closed-ended question, circumstances that require closed-ended questions, adding value to a counseling session through a follow-up question.
It should be noted that open-ended questions should be asked after considerable deliberation since they tend to reveal excessive information about the respondent. In addition, the question posed should bear relevance to the topic being pursued. In a bid, to establish a rapport with a client, I would ask him, or her to brief me about the goings-on in his or her life.
Open-ended questions are beneficial to counselors whenever clients are required to provide an elaborate response to a question they have been asked. This is because they provide complicated and in-depth answers to queries that are raised during interactive sessions. Most importantly, this variety of questions comes in handy whenever there is a need to create an insight into a predicament by probing deeper into client concerns.
Consequently, such questions are best used during therapies and other reflective sessions, whenever a persuasion is a viable option during the process. In addition, clients will be attended to properly only if the service providers understand the issues affecting them in their daily lives. It is commonplace that asking open-ended questions is the best way to establish these truths (Schultz, 2010).
It is noteworthy that a person’s life entails diverse and varied aspects that they have to deal with daily. As a result, establishing the root of their afflictions would require additional, further probing into the same matter. Assuming the client had issues at his or her place of work, the ideal follow-up question would require him or her how they feel whenever they discharge their duties at the workplace. This will limit the scope of their answer; while equally allowing them a free hand within the confines of the question to talk about the pressing matters.
These questions are used whenever the questioner intends to narrow down a protracted conversation, to get down to a verdict or finale. They may be asked during the initial stages of research sessions to ascertain the credibility and astuteness of respondents, before the commencement of the fact-finding. In case a survey is being carried out; an ideal question would entail establishing if the client required further clarification on the topic. This would influence the action taken by the researcher, whether he will proceed with the session or take some time to clarify matters to the respondent.
It has been established that closed-ended questions are often used whenever a ‘yes’ or ’no’ response is required. This is because the nature of the question is restrictive, hence prohibiting the respondent from providing further information. It should be noted that most respondents opt to abstain from issuing additional information whenever they are asked such a question (James, 2008).
In addition, these questions often appear to lead the respondent, since one of the possible answers is always mentioned in the query (Bradburn, Sudman & Wansik, 2004). This may affect the integrity of the response issued, subsequently interfering with the entire process. This serves to highlight the importance of these questions when verifying facts that have been reported with regards to a topic of concern.
Assuming the client required clarification about the research topic and the interviewer had responded accordingly, establishing whether the respondent could field the questions is mandatory. In a counseling session, a proper follow-up question could seek to establish, if the respondent was in a position, to proceed with the session, or not. This will enable the counselor to determine a suitable course of action to pursue. If the client is at ease, the session may proceed as intended. In case the client is still uncomfortable; the counselor may find suitable methods of making them relax before proceeding with that session.
Bradburn, N., Sudman, S & Wansik, B. (2004). Asking Questions: the definitive guide to questionnaire design: for market research, political polls, and social and health questionnaires . California, CA: John Wiley and Sons.
James, R. (2008). Crisis intervention strategies . New Jersey, NJ: Cengage Learning.
Schultz, M. (2010). Open Ended Questions for Your Prospects and Customers . Sales marks. Web.
IvyPanda. (2022, January 6). Open and Closed Questions: Circumstances Reconciliation. https://ivypanda.com/essays/open-and-closed-questions-circumstances-reconciliation/
"Open and Closed Questions: Circumstances Reconciliation." IvyPanda , 6 Jan. 2022, ivypanda.com/essays/open-and-closed-questions-circumstances-reconciliation/.
IvyPanda . (2022) 'Open and Closed Questions: Circumstances Reconciliation'. 6 January.
IvyPanda . 2022. "Open and Closed Questions: Circumstances Reconciliation." January 6, 2022. https://ivypanda.com/essays/open-and-closed-questions-circumstances-reconciliation/.
1. IvyPanda . "Open and Closed Questions: Circumstances Reconciliation." January 6, 2022. https://ivypanda.com/essays/open-and-closed-questions-circumstances-reconciliation/.
Bibliography
IvyPanda . "Open and Closed Questions: Circumstances Reconciliation." January 6, 2022. https://ivypanda.com/essays/open-and-closed-questions-circumstances-reconciliation/.
Feb 7, 2023 | Yusuf Mehdi - Corporate Vice President & Consumer Chief Marketing Officer
To empower people to unlock the joy of discovery, feel the wonder of creation and better harness the world’s knowledge, today we’re improving how the world benefits from the web by reinventing the tools billions of people use every day, the search engine and the browser.
Today, we’re launching an all new, AI-powered Bing search engine and Edge browser, available in preview now at Bing.com , to deliver better search, more complete answers, a new chat experience and the ability to generate content. We think of these tools as an AI copilot for the web.
“AI will fundamentally change every software category, starting with the largest category of all – search,” said Satya Nadella, Chairman and CEO, Microsoft. “Today, we’re launching Bing and Edge powered by AI copilot and chat, to help people get more from search and the web.”
There are 10 billion search queries a day, but we estimate half of them go unanswered. That’s because people are using search to do things it wasn’t originally designed to do. It’s great for finding a website, but for more complex questions or tasks too often it falls short.
The new Bing and Edge – Your copilot for the web
We have brought together search, browsing and chat into one unified experience you can invoke from anywhere on the web, delivering:
My anniversary is coming up in September, help me plan a trip somewhere fun in Europe, leaving from London.
Will the Ikea Klippan loveseat fit into my 2019 Honda Odyssey?
Reinventing search with AI
The new Bing experience is a culmination of four technical breakthroughs:
These groundbreaking new search experiences are possible because Microsoft has committed to building Azure into an AI supercomputer for the world, and OpenAI has used this infrastructure to train the breakthrough models that are now being optimized for Bing.
Microsoft and OpenAI – innovating together, responsibly
Together with OpenAI, we’ve also been intentional in implementing safeguards to defend against harmful content. Our teams are working to address issues such as misinformation and disinformation, content blocking, data safety and preventing the promotion of harmful or discriminatory content in line with our AI principles .
The work we are doing with OpenAI builds on our company’s yearslong effort to ensure that our AI systems are responsible by design. We will continue to apply the full strength of our responsible AI ecosystem – including researchers, engineers and policy experts – to develop new approaches to mitigate risk.
Live today in limited preview, expanding to millions soon
The new Bing is available today in a limited preview on desktop, and everyone can visit Bing.com today to try sample queries and sign up for the waitlist. We’re going to scale the preview to millions in the coming weeks. A mobile experience will also be in preview soon.
We’re excited to put the new Bing and Edge into the real world to get the critical feedback required to improve our models as we scale.
Related links:
Amy Hood, Microsoft executive vice president and chief financial officer, will host a conference call with investors at 2:30 p.m. PT.
Brad Smith, Microsoft vice chair and president: Meeting the moment: advancing the future through responsible AI
Learn more about advertising on the new Bing
More information about the announcement
Tags: AI , Bing , Microsoft Edge
For more audio journalism and storytelling, download New York Times Audio , a new iOS app available for news subscribers.
Students are using artificial intelligence to create sexually explicit images of their classmates..
Hosted by Sabrina Tavernise
Featuring Natasha Singer
Produced by Sydney Harper and Shannon M. Lin
Edited by Marc Georges
Original music by Marion Lozano , Elisheba Ittoop and Dan Powell
Engineered by Chris Wood
Warning: this episode contains strong language, descriptions of explicit content and sexual harassment
A disturbing new problem is sweeping American schools: Students are using artificial intelligence to create sexually explicit images of their classmates and then share them without the person depicted even knowing.
Natasha Singer, who covers technology, business and society for The Times, discusses the rise of deepfake nudes and one girl’s fight to stop them.
Natasha Singer , a reporter covering technology, business and society for The New York Times.
Using artificial intelligence, middle and high school students have fabricated explicit images of female classmates and shared the doctored pictures.
Spurred by teenage girls, states have moved to ban deepfake nudes .
There are a lot of ways to listen to The Daily. Here’s how.
We aim to make transcripts available the next workday after an episode’s publication. You can find them at the top of the page.
The Daily is made by Rachel Quester, Lynsea Garrison, Clare Toeniskoetter, Paige Cowett, Michael Simon Johnson, Brad Fisher, Chris Wood, Jessica Cheung, Stella Tan, Alexandra Leigh Young, Lisa Chow, Eric Krupke, Marc Georges, Luke Vander Ploeg, M.J. Davis Lin, Dan Powell, Sydney Harper, Mike Benoist, Liz O. Baylen, Asthaa Chaturvedi, Rachelle Bonja, Diana Nguyen, Marion Lozano, Corey Schreppel, Rob Szypko, Elisheba Ittoop, Mooj Zadie, Patricia Willens, Rowan Niemisto, Jody Becker, Rikki Novetsky, John Ketchum, Nina Feldman, Will Reid, Carlos Prieto, Ben Calhoun, Susan Lee, Lexie Diao, Mary Wilson, Alex Stern, Sophia Lanman, Shannon Lin, Diane Wong, Devon Taylor, Alyssa Moxley, Summer Thomad, Olivia Natt, Daniel Ramirez and Brendan Klinkenberg.
Our theme music is by Jim Brunberg and Ben Landsverk of Wonderly. Special thanks to Sam Dolnick, Paula Szuchman, Lisa Tobin, Larissa Anderson, Julia Simon, Sofia Milan, Mahima Chablani, Elizabeth Davis-Moorer, Jeffrey Miranda, Maddy Masiello, Isabella Anderson, Nina Lassam and Nick Pitman.
Natasha Singer writes about technology, business and society. She is currently reporting on the far-reaching ways that tech companies and their tools are reshaping public schools, higher education and job opportunities. More about Natasha Singer
Advertisement
COMMENTS
2. Facilitating self-expression. Open-ended questions allow us to express ourselves. Imagine only living life being able to say "yes" or "no" to questions. We'd struggle to get across our own personalities! Only with fully-expressed sentences and monologues can we share our full thoughts, feelings, and experiences.
Comparison: Open-ended vs closed-ended questions. Open-ended and closed-ended questions serve as the two sides of the inquiry coin, each with its unique advantages. Open-ended questions: Kickstart with "How", "Why", and "What". No set answers, sparking more thought. Encourage detailed responses, explaining the 'why' or 'how'.
Open-ended questions are those that do not define the scope you should take (i.e., how many and what kinds of experiences to discuss). Like personal statements for other types of applications, open-ended essays have more room for creativity, as you must make the decision on issues such as how expansive or narrow your topic should be.
Download Article. 1. Begin your question with "how," "why," or "what.". As you begin writing your questions, start them with words that could prompt multiple possible answers. Questions that open with more specific words, such as "which" or "when," often have a single correct answer. [6]
MMW2, W2009. Developing Effective Open-Ended Questions and Arguable, Research-Based Claims for Academic Essays. Asking Open-Ended, Arguable Questions. In academic papers, the thesis is typically an answer to a question about a significant issue that has more than one possible answer and requires research to provide evidence.
Define closed-ended question: a close-ended question is a question that expects a specific answer and does not give leeway outside of that answer. In summary, Open-ended questions are broad and do not expect a specific answer. Close-ended questions are limiting and expect a specific answer. Answers. Examples of open questions. Learn the ...
The key is to ask open-ended questions, rather than closed ones. Closed questions usually produce a short, yes or no response; they tend to limit the conversation. On the other hand, open questions produce a longer, fuller response; they expand the conversation. The most important benefit of open-ended questions is that they let you find out ...
An open-ended question opens up a topic for exploration and discussion while a closed-ended question leads to a closed-off conversational path. After "Yes" or "No" or the specific one-word answer to the question, the thread is done. Open-ended questions lead to qualitative answers while closed-ended questions lead to quantitative answers.
1. Encourage deep reflection. Open-ended questions prompt respondents to think critically and reflect on their thoughts, feelings, and experiences. By inviting individuals to express themselves in their own words, these questions encourage deeper introspection and provide better response quality. 2.
The three main goals of this article are: to describe how students need to approach the close reading of the questions, or tasks on the assessments; to identify the kinds of skills and knowledge students need in writing clear, comprehensible responses; and. to examine issues related to fluency in writing and stamina that arise as students work ...
To put it as simply as possible, open-ended questions are questions that require more than a short, fixed response. Open-ended questions try to avoid answers like "Yes.", "No.", "The Battle of Midway.", or "Onions.". Open-ended questions attempt to make the person who is answering the question give a more detailed and elaborate ...
Written assessments fall into two classes: constructed-response or open-ended questions, such as the essay and a number of variants of the short-answer question, and selected-response or closed-ended questions; typically in the form of multiple-choice. It is widely believed that constructed response written questions test higher order cognitive processes in a manner that multiple-choice ...
What is Open Ended Question. Open-ended questions are those which require more thought and more than a simple one-word answer. An open-ended question is designed to encourage a full, meaningful, and deliberate answer using the subject's own knowledge and/or feelings. It is the opposite of a closed-ended question, which encourages a short or ...
It's test time, and this one isn't multiple choice. Your teacher gives you a sheet of paper with a question on it. The only problem is, you can't immediately see a definite answer. ... COLLEGE ; TESTS ; VOCAB ; LIFE ; TECH ; How to Answer Open-Ended Essay Questions. EDITORIAL TEAM CLASS. How to Answer Open-Ended Essay Questions. It's test time ...
Open-Ended vs. Closed Questions. There are two types of questions we can use in research studies: open-ended and closed. Open-ended questions allow participants to give a free-form text answer. Closed questions (or closed-ended questions) restrict participants to one of a limited set of possible answers.. Open-ended questions encourage exploration of a topic; a participant can choose what to ...
18 Tough Open-Ended Questions (And How To Answer Them) Interviewers ask open-ended interview questions during the hiring process to learn more about a candidate's experience and relevant abilities. The ability to answer open-ended interview questions in a detailed and thoughtful manner can show your problem-solving and critical-thinking skills.
Open-ended questions can be a little hard to spot sometimes. How can you know if a question is open-ended or closed-ended? Browse these examples to find out. ... while open-ended questions are more like subjective short responses and essay questions. Now that you know the difference between these question types, ...
For example, an open-ended question allows you to probe much deeper but a close ended question allows you to get concise information that can be quantified. It's much easier to quantify yes or nos than a paragraph of text. A relatable example comes from the standardized tests most of us took in school.
Closed-ended questions are questions that can only be answered by selecting from a limited number of options, usually multiple-choice questions with a single-word answer ('yes' or 'no') or a rating scale (e.g. from strongly agree to strongly disagree). Closed-ended questions give limited insight, but can easily be analyzed for ...
Open-ended questions that the candidate can answer by writing, speaking, filling in tables, taking pictures, writing code, uploading files and using similar methods are essential components in many exams,tests, and assessment processes. In open-ended questions, the candidate is asked not to choose an answer, but to create it.
In a data-driven world, recognizing the role of open-ended questions is vital. They infuse life into hard facts, offering context, depth, and, ultimately, a richer meaning to our research. 9 Best Practices for Writing Open-Ended Questions Here are 9 of our best practices for eliciting more thoughtful feedback with open-ended questions. Find the ...
Such a request demonstrates that the practitioner is actually interested in the answer and helps to escape cliché questions, which can receive closed-ended answers. Another example of rephrasing open-ended questions was Question 4 provided above. Children often come back with "Fine!" after being asked how school was today.
An example of an open-ended question. It should be noted that open-ended questions should be asked after considerable deliberation since they tend to reveal excessive information about the respondent. In addition, the question posed should bear relevance to the topic being pursued. In a bid, to establish a rapport with a client, I would ask him ...
There are 10 billion search queries a day, but we estimate half of them go unanswered. That's because people are using search to do things it wasn't originally designed to do. It's great for finding a website, but for more complex questions or tasks too often it falls short. The new Bing and Edge - Your copilot for the web
ing results, extracurricular involvement, essay quality, per-sonal factors, and student background. Id., at 600. Readers are responsible for providing numerical ratings for the aca-demic, extracurricular, personal, and essay categories. Ibid. During the years at issue in this litigation, un-derrepresented minority students were "more likely to
The Southern Baptist Convention, long a bellwether for American evangelicalism, voted to oppose the use of in vitro fertilization.
Warning: this episode contains strong language, descriptions of explicit content and sexual harassment. A disturbing new problem is sweeping American schools: Students are using artificial ...