Emerging reforms in health professions education such as competency-based education, mastery learning, entrustable professional activities, and adaptive learning environments underscore the need for valid assessments of learning outcomes. The currently standard framework for thinking about assessment validity, first proposed by Messick1 in 1989, defines validity as “the degree to which evidence and theory support the interpretations of test scores for proposed uses of tests.”2 Validity can be viewed as a hypothesis about the meaning (interpretations) and application (uses) of test scores. Like any hypothesis, the validity hypothesis can be tested by collecting evidence, which is then summarized in a coherent narrative or validity argument that identifies strengths, weaknesses, and residual gaps (i.e., the degree of support).3,4 Evidence targeting key assumptions is vital to crafting a strong validity argument.
In this framework, evidence derives from five different sources: content, internal structure, relationships with other variables, response process, and consequences (see Table 1).5,6 The first three sources map to prior notions of content validity; reliability; and criterion, correlational, and construct validity, respectively,7 and as such have been readily understood by educators. However, the concepts of response process and consequences have no counterpart in the older framework, and in our experience it has been challenging for educators to understand these concepts and visualize how these might be implemented in practice. Perhaps for these reasons consequences evidence is rarely reported in health professions education research,6,8 and when reported it tends to be limited in scope.6 Yet, authors have repeatedly emphasized the critical significance of consequences evidence in presenting a compelling validity argument.3,5,6,9
Although the origin of this disparity between what experts request and what investigators report is not fully known, a detailed discussion of consequences evidence would enhance both awareness of the issue and understanding of how to collect needed evidence. The purpose of this article is to explain consequences evidence in easily understood terms and propose a framework for organizing the collection and interpretation of such evidence along with several examples.
In approaching this topic, we first reviewed seminal works on validity in general1–3,5,6,9–11 and consequences evidence specifically.12–17 We also reviewed each article in three systematic reviews of validity evidence in health professions education assessments6,8,18 to identify the frequency and type of consequences evidence presented therein. We then synthesized these theories and exemplars to create a novel framework for planning and organizing consequences evidence, and to propose specific hypothetical examples of how this evidence might be collected in practice.
What Do We Mean by “Consequences”?
Consequences evidence looks at the impact, beneficial or harmful and intended or unintended, of assessment.2,13 In this sense, assessment can be viewed as an intervention. The act of administering or taking a test, the analysis and interpretation of scores, and the ensuing decisions and actions (such as remediation, feedback, promotion, or board certification) all have direct impacts on those being assessed and on other people and systems (e.g., teachers, patients, schools). These impacts should ideally be evaluated to determine whether actual benefits align with anticipated benefits and outweigh costs and adverse effects.
An analogy with clinical medicine may help to illustrate the concept of assessments as interventions. Mammograms are assessments (diagnostic tests) used to screen for breast cancer. Current evidence suggests that they are less useful in younger women because interpretation is more difficult, that comparison with old films is often required before a judgment can be made, and that false positives are common and subject women to unnecessary biopsies and emotional stress.19–23 Yet most experts agree that for women aged 50 to 74, annual screening mammograms are beneficial because they substantially reduce the adverse consequences of breast cancer.24,25 Despite the imperfections of the test and unintended negative consequences of false-positive results, the test has an overall beneficial impact. However, for younger women (for whom the false-positive rate is higher20) and for older women (who might die of other causes before they die of breast cancer) screening mammograms should not be automatic according to some guidelines,24 although this is a matter of controversy.26 Other clinical examples include the use of brain natriuretic peptide for diagnosing heart failure,27 flexible sigmoidoscopy for colon cancer screening,28 and computed tomographic angiography for detection of coronary artery disease29—each of which has been evaluated using randomized trials comparing the long-term impact of testing (and its associated clinical decisions) vs no testing. In each case, the act of testing is in fact an intervention with costs, benefits, and potential harms.
Similarly, educational assessments can be viewed as interventions with potential costs, benefits, and harms. For example, a board certification exam might protect patients from incompetent physicians and encourage physicians to study, but might also force competent physicians with poor test-taking skills to engage in needless remediation. This exam has “intervened” in the lives of physicians and patients and led to both beneficial and harmful consequences. To further illustrate, Table 2 cites published studies in which use of educational assessments improved knowledge and skills, altered study behaviors, enhanced faculty rater skills, or led to curricular change.
Stated another way: Consequences evidence does not address the question, “Are we measuring what we think we are measuring?” (a question answered by the other sources of validity evidence). Rather, it addresses, “Does the activity of measuring and the subsequent interpretation and application of scores achieve our desired results with few negative side effects?”
Investigators occasionally confuse consequences as a source of assessment validity evidence (the focus of this article) with other uses of the word “consequences” (e.g., as a general synonym for impact or outcome). For example, studies often evaluate the consequences of training activities (courses, curricula, online modules, or simulation scenarios) using outcomes measured in a test setting or in real clinical practice; such evaluations of training interventions are conceptually distinct from studies evaluating consequences evidence to support assessment validity. Alternatively, an assessment validation study might evaluate the association between test scores and other concurrent or future measurements of patients, programs, or society (i.e., real-life outcomes or “consequences”). Such associations would inform the validity argument by establishing “relationships with other variables”2,7 but would not reflect consequences validity evidence (i.e., the analysis focuses on the relationships among scores rather than the consequences of the assessment itself). Of course, there are situations in which measures of impact constitute evidence of assessment consequences (assessments are, after all, interventions), and correlational analyses can provide consequences evidence (see Table 2 and Appendix 1 for examples). What matters is not the study design or statistical analysis but, rather, how the evidence is presented in the validity argument: Consequences evidence establishes the impact of interpretations and uses of assessment scores.
The Importance of Consequences Evidence
Clinicians are often taught not to order a test if it won’t improve patient management. The same holds true for educational assessments: If they do not lead to improved learning outcomes sufficient to outweigh costs and potential harms, they should not be used. Messick1(p85) argued that “Evaluation of the consequences and side effects of testing is a key aspect of the validation of test use.” Kane’s3(p54) more recent conceptual reframing of validation, which focuses on key inferences in the validity argument, gives similar priority to evidence supporting the consequences of assessment: “Consequences, or outcomes, are the bottom line in evaluating decision procedures. A decision procedure that does not achieve its goals, or does so at too high a cost, is likely to be abandoned, even if it is based on perfectly accurate information.” Other authors have also supported the primacy of consequences evidence.5,6,9
Just as the ultimate evidence for the value of a diagnostic test is the impact on practice, the ultimate evidence for the value of an educational assessment is the impact on learners, teachers, and the people and systems they influence.12 Like clinical tests, educational assessments may fail to realize their intended benefits or may have costs or unintended negative consequences that outweigh the benefits.12,13,17 In such instances one could argue that the rigor of instrument development, the reliability of scores, and the strength of score correlations with other variables really don’t matter. Such concerns underpin many recent criticisms of high-stakes testing as part of the board recertification process.30 For this reason, we believe that evidence of consequences is ultimately the most important source of validity evidence.
Consequences Evidence in Health Professions Education Research
Consequences evidence is reported only infrequently in health professions education. A systematic review of 22 clinical teaching assessments found only 2 studies (9%) that reported consequences evidence, and in neither case did the original researchers identify the evidence as such.8 One study found that providing formative feedback to teachers enhanced their teaching scores,31 whereas the other study found that the assessment raised awareness of effective teaching behaviors.32 A systematic review of 417 articles examining simulation-based assessment6 found only 20 studies (5%) reporting consequences evidence. The majority of this evidence comprised establishing a pass/fail cut point (n = 14). Two studies explored an anticipated impact on students or patients,33,34 3 contrasted the number of actual vs. expected passing grades,35–37 and 1 study noted differential item functioning as a possible source of invalidity.38 No study reported an unanticipated impact. Finally, a systematic review of 55 studies evaluating assessment tools for direct observation18 found 11 studies (20%) reporting consequences evidence other than satisfaction with the assessment activity. All of these evaluated the impact of assessment, documenting outcomes including curricular changes based on common deficiencies,39 improved feedback,40–43 poor recall of feedback provided (i.e., failure to achieve intended consequence),44 improved objectively measured skills,45,46 and increased test preparation activities.47Table 2 contains illustrative quotes from several of these published studies.
A Framework for Evaluating Consequences Evidence
Consequences evidence consists of data on the impact of an assessment on diverse parties: learners, educators, and educational institutions; patients, providers, and health care institutions; and even society at large. Such impact can be beneficial or harmful, and it may be intentional or unintentional.13 Intentional benefits are probably the easiest to anticipate and measure; unintentional harms may be the most difficult (because they cannot be easily anticipated or explicitly targeted).48 Experts have also distinguished direct effects of score use (e.g., instructional guidance or advancement decisions) from indirect effects (e.g., influence on student motivation or preparation activities, instructor lesson plans, and public perceptions).17 However, although these classifications are helpful for categorizing, interpreting, and reporting consequences evidence once it has been collected, they are inadequate for helping investigators to consider broadly the potential sources of consequences evidence when planning an assessment validation study. Moreover, the same effect might be considered intended or unintended, beneficial or harmful, and direct or indirect depending on the proposed theory, interpretation, and use of the assessment. For example, an assessment might have unintended effects on learners’ general orientations toward performing well relative to peers vs. mastering content for its own sake (performance vs. mastery goal orientations49). However, promoting stronger mastery goal orientations may be an explicitly intended consequence of assessment when adopting a mastery learning curricular model.50 Similarly, one could imagine educational assessments that lead physicians to be risk averse in beneficial ways (e.g., carefully following protocol for central line placement after a central line assessment) or in detrimental ways (e.g., practicing “defensive medicine” by ordering unnecessary lab tests after a test of medical knowledge).
Previous authors, including ourselves, have included evaluations of the rigor, appropriateness, and consistency of classification cut points and labels as consequences evidence.5–7,50 Although such evidence has direct bearing on the implications and decisions arising from the assessment, on careful reflection we believe it might be more correctly labeled preconsequences evidence because it affects, rather than results from, the actual consequences of assessment. With this caveat, we continue to agree that such evidence fits most appropriately as consequences evidence in Messick’s framework. (As an aside, we note that in Kane’s more recent framework such evidence fits squarely under the inference of “implications and decision.”3,9)
In considering how to help investigators prospectively plan the collection of consequences evidence and help consumers identify evidence gaps, we have integrated the above conceptual elements to create a comprehensive framework for systematically prioritizing and organizing consequences evidence. First, evidence can derive from evaluations of the impact on examinees, educators, and other stakeholders (e.g., patients), and the impact of classifications (“preconsequences,” e.g., different cut scores or labels, and accuracy across examinee subgroups). Second, studies can be distinguished as evaluating the impact of test score use (similar to the “direct” effects noted above) such as the effectiveness of score-guided remediation or advancement decisions; or the impact of the assessment activity itself (independent of scores) such as change in preassessment study behaviors or the effect of test-enhanced learning. To use a clinical example: A woman might get anxious about an upcoming mammogram because she is scared that it might detect cancer (impact of [anticipated] “score” use), or she might be worried about the potential pain or financial cost (impact of the test activity independent of the “score”). Each of these dimensions could include consequences that are intended or unintended, and beneficial or harmful; adding the latter points completes a four-dimensional framework (see Figure 1). Investigators could use this framework to systematically consider the potential consequences of an assessment, prioritize evidence gaps, and select research approaches to fill these gaps. We briefly discuss below how data might be collected to evaluate impact and defensibility, and illustrate this in Appendix 1 with examples spanning all dimensions.
A straightforward approach to evaluate a test’s impact—for both a clinical diagnostic test and for an educational assessment—would be to randomize half of the participants to complete the test and the other half to no test,13,51 and then quantitatively measure relevant anticipated outcomes, or use qualitative methods to observe for anticipated and unanticipated effects. Of course, local needs may make a randomized trial infeasible in many situations. A less robust but still useful approach might use less rigorous study designs (such as nonrandomized cohort, single-group pretest–posttest, or even single-group posttest-only studies) but measure the same outcomes. Those being assessed are not the only ones impacted by an assessment. Appendix 1 illustrates potential impacts on educators, patients, institutions, and nonassessed learners.
As noted above, preconsequences evidence includes factors that directly influence the defensibility of classifications based on test results (interpretations and decisions), such as the labels applied to the test itself and any subtests1; the definition of the passing score (e.g., at what point is remediation required?)5; and differences in scores among subgroups where performance ought to be similar (e.g., men vs. women), suggesting that decisions may be spurious.52 Finally, investigators could monitor pass/fail rates; for example, a failure rate higher or lower than expected might indicate a test that is either too hard or too easy, respectively.
We distinguish unintended consequences, which can be nonetheless anticipated and prospectively measured, from unforeseeable consequences, which can only be identified after the fact. We further emphasize that data need not be numeric. Qualitative data, properly planned and collected, could provide strong evidence9—especially when seeking to identify unintended or unforeseeable consequences.
The data in many of these examples are highly subjective and open to alternative interpretation. For example, score differences among subgroups could be a sign of invalidity if scores should be the same, but could also be interpreted as supporting validity if scores would be expected to vary. Similarly, the ideal failure rate will vary by situation. It is essential to articulate in advance what findings would support or undermine the validity argument,9,10,53 often guided by a theory of action linking the assessment and its consequences.3,54 Ultimately, it may be difficult if not impossible to establish a clear cause–effect relationship between the assessment and its consequences.14 This should not, however, justify educators in ignoring this important element of the validity argument. Triangulation of different evidence sources and data collection methods will help establish a defensible argument.
Finally, the side effects of intended uses of an assessment should not be confused with the effects of misuse.10 Any application of test scores beyond the scope of existing evidence constitutes, strictly speaking, a misuse. This would include adopting an assessment for new purposes (e.g., using licensure exam scores to inform admissions decisions) or adapting an assessment by changing elements in the instrument, procedures, or learner population. Although it is commonplace and often profitable to adopt or adapt an existing assessment, those doing so should remember that “Test makers are not responsible for negative consequences following from test misuse.… When users appropriate tests for purposes not sanctioned and studied by the test developers, users become responsible for conducting the needed validity investigation.”13(p8)
Identifying and Using Consequences Evidence in Practice
Not all consequences evidence is equally compelling. Simple improvement in test scores from one testing occasion to the next (e.g., “Students did better when they were retested, suggesting that their skills had improved as a result of the first test”) would not, for example, contribute persuasive evidence of consequences because we can imagine plausible alternative explanations for this change (i.e., learning from other experiences). Learner and faculty ratings of satisfaction with the assessment, self-reported improvements in skill attributed to the assessment, and pass/fail rates without a comparison reference point would provide useful but rather weak evidence. Similarly, the establishment of a pass/fail cut point, regardless of how rigorously done, is relatively weak evidence until the consequences of that cut point have been evaluated in practice. Anecdotes without robust quantitative or qualitative data likewise provide only weak support. Stronger evidence will come from studies using a comparison group (randomized, or nonrandomized historical or concurrent control group); objective measures of the desired outcomes that are different from the test itself; or rigorous qualitative data collection and analysis.
Although consequences evidence is the most important source of evidence, test developers, test users, researchers, and journal editors must remember that it constitutes only one of several elements in a comprehensive validity argument. No single source can or should dominate. Moreover, robust consequences “evidence cannot be collected until the test is used as intended for some period of time.”14(p15) As such, a stepwise approach seems reasonable. We propose that during initial instrument evaluation, developers and researchers might prioritize presumably easier and less costly evidence sources (e.g., content, internal structure/reliability, relationships with other variables, response process [see Table 1]) and then progress to rigorous evaluation of consequences if this evidence proves supportive.
The type, quantity, and rigor of consequences evidence will vary depending on the assessment—more specifically, on the proposed arguments or claims for benefit. For example, a licensure exam that claims to enhance patient safety (anticipated benefit) will impact the employability of physicians who fail. Such an assessment likely merits greater evidence of consequences (e.g., Are anticipated benefits realized? How was the pass/fail cut point established? How often do competent physicians fail?) than an assessment designed to promote feedback to medical students. However, some supposedly “low-stakes” exams could have potentially significant consequences, especially if implemented on a large scale or repeated over an extended period of time. For example, an assessment intended to promote feedback could have significant cumulative effects across multiple domains of competence, professional identity, self-directed learning, and self-efficacy if administered daily over an entire year of training.
Although our framework is more comprehensive than any other we found, application of this framework will require thoughtful consideration of assessment purposes, procedures, and theory; the level of evidence required; and practical constraints of context.
Unfavorable validity evidence often points to problems elsewhere in the assessment process. Negative consequences can usually be traced back to one of four underlying problems3: the measurement or scoring procedure (e.g., irrelevant, unreliable, or omitted test items); the specific interpretation (e.g., an inappropriate pass/fail cut point); the attribute being measured (i.e., the wrong construct); or the response (e.g., the actions that follow the decision). For example, a test intended to identify students in need of remediation in cardiac auscultation might fail to have intended consequences because it contains flawed items, because too many competent students are labeled as incompetent, because it measures knowledge rather than skill, or because the remediation program is ineffective.
Finally, although the present article focuses on education, the importance of assessment consequences is not limited to educational tests. Indeed, the earlier example of the consequences of mammography illustrates the application of this concept to clinical medicine. Other applications would include (but certainly are not limited to) patient symptom scales, teacher rating scales, employment aptitude inventories, customer satisfaction surveys, and research questionnaires.
In conclusion, we emphasize the following. First, assessments are really diagnostic tests, and both in medicine and in education they can be viewed as interventions. Second, consequences validity evidence looks at the impact of assessments (as interventions) on examinees and other stakeholders, and the defensibility of score classifications (“preconsequences” evidence). Such consequences can arise from score use or the assessment activity itself, and can be intentional or unintended and beneficial or harmful. Third, consequences validity evidence is the most important source of evidence because if the assessment does not have the desired impact, it should not be used. Finally, the type, quantity, and rigor of consequences evidence will vary depending on the assessment and the claims for its use.
As health professions educators increasingly rely on assessments to guide important decisions (e.g., recertification, competency-based promotion), they will need stronger evidence to support the validity of the inferences and decisions made. To date, such evidence of consequences has been infrequently reported. Going forward, our framework, which distinguishes the direct impact of the assessment and the indirect influence of other mediating factors and identifies multiple domains within each classification, can help test developers and users to consider a broad view of potentially relevant consequences evidence.
1. Messick S. Validity.Linn RL. Educational Measurement. 1989:3rd ed. New York, NY: American Council on Education and Macmillan; 13103.
2. American Educational Research Association, American Psychological Association, National Council on Measurement in Education. Validity.Standards for Educational and Psychological Testing. 2014:Washington, DC: American Educational Research Association; 1131.
3. Kane MT. Brennan RL. Validation.Educational Measurement. 2006:4th ed. Westport, Conn: Praeger; 1764.
4. Cook DA. When I say… validity. Med Educ. 2014;48:948949.
5. Downing SM. Validity: On meaningful interpretation of assessment data. Med Educ. 2003;37:830837.
6. Cook DA, Zendejas B, Hamstra SJ, Hatala R, Brydges R. What counts as validity evidence? Examples and prevalence in a systematic review of simulation-based assessment. Adv Health Sci Educ Theory Pract. 2014;19:233250.
7. Cook DA, Beckman TJ. Current concepts in validity and reliability for psychometric instruments: Theory and application. Am J Med. 2006;119:166.e7166.e16.
8. Beckman TJ, Cook DA, Mandrekar JN. What is the validity evidence for assessments of clinical teaching? J Gen Intern Med. 2005;20:11591164.
9. Cook DA, Brydges R, Ginsburg S, Hatala R. A contemporary approach to validity arguments: A practical guide to Kane’s framework. Med Educ. 2015;49:560575.
10. Kane MT. Validating the interpretations and uses of test scores. J Educ Meas. 2013;50:173.
11. Kane MT. Validation as a pragmatic, scientific activity. J Educ Meas. 2013;50:115122.
12. Linn RL. Evaluating the validity of assessments: The consequences of use. Educ Meas Issues Pract. 1997;16:1416.
13. Shepard LA. The centrality of test use and consequences for test validity. Educ Meas Issues Pract. 1997;16:524.
14. Reckase MD. Consequential validity from the test developer’s perspective. Educ Meas Issues Pract. 1998;17:1316.
15. Lane S, Stone CA. Strategies for examining the consequences of assessment and accountability programs. Educ Meas Issues Pract. 2002;21:2330.
16. Moss PA. Validity in action: Lessons from studies of data use. J Educ Meas. 2013;50:9198.
17. Haertel E. How is testing supposed to improve schooling? Measurement. 2013;11:118.
18. Kogan JR, Holmboe ES, Hauer KE. Tools for direct observation and assessment of clinical skills of medical trainees: A systematic review. JAMA. 2009;302:13161326.
19. Armstrong K, Moye E, Williams S, Berlin JA, Reynolds EE. Screening mammography in women 40 to 49 years of age: A systematic review for the American College of Physicians. Ann Intern Med. 2007;146:516526.
20. Nelson HD, Tyne K, Naik A, Bougatsos C, Chan BK, Humphrey L; U.S. Preventive Services Task Force. Screening for breast cancer: An update for the U.S. Preventive Services Task Force. Ann Intern Med. 2009;151:727737, W237.
21. Hubbard RA, Kerlikowske K, Flowers CI, Yankaskas BC, Zhu W, Miglioretti DL. Cumulative probability of false-positive recall or biopsy recommendation after 10 years of screening mammography: A cohort study. Ann Intern Med. 2011;155:481492.
22. Welch HG, Passow HJ. Quantifying the benefits and harms of screening mammography. JAMA Intern Med. 2014;174:448454.
23. Roelofs AA, Karssemeijer N, Wedekind N, et al. Importance of comparison of current and prior mammograms in breast cancer screening. Radiology. 2007;242:7077.
24. U.S. Preventive Services Task Force. Screening for breast cancer: U.S. Preventive Services Task Force recommendation statement. Ann Intern Med. 2009;151:716726, W236.
25. American Cancer Society. American Cancer Society recommendations for early breast cancer detection in women without breast symptoms. http://www.cancer.org/cancer/breastcancer/moreinformation/breastcancerearlydetection/breast-cancer-early-detection-acs-recs
. Accessed December 15, 2015.
26. Hendrick RE, Helvie MA. United States Preventive Services Task Force screening mammography recommendations: Science ignored. AJR Am J Roentgenol. 2011;196:W112W116.
27. Lam LL, Cameron PA, Schneider HG, Abramson MJ, Müller C, Krum H. Meta-analysis: Effect of B-type natriuretic peptide testing on clinical outcomes in patients with acute dyspnea in the emergency setting. Ann Intern Med. 2010;153:728735.
28. Schoen RE, Pinsky PF, Weissfeld JL, et al.; PLCO Project Team. Colorectal-cancer incidence and mortality with screening flexible sigmoidoscopy. N Engl J Med. 2012;366:23452357.
29. Muhlestein JB, Lappé DL, Lima JA, et al. Effect of screening for coronary artery disease using CT angiography on mortality and cardiac events in high-risk patients with diabetes: The FACTOR-64 randomized clinical trial. JAMA. 2014;312:22342243.
30. Teirstein PS. Boarded to death—why maintenance of certification is bad for doctors and patients. N Engl J Med. 2015;372:106108.
31. Cohen R, MacRae H, Jamieson C. Teaching effectiveness of surgeons. Am J Surg. 1996;171:612614.
32. Copeland HL, Hewson MG. Developing and testing an instrument to measure the effectiveness of clinical teaching in an academic medical center. Acad Med. 2000;75:161166.
33. Berkenstadt H, Ziv A, Gafni N, Sidi A. The validation process of incorporating simulation-based accreditation into the anesthesiology Israeli national board exams. Isr Med Assoc J. 2006;8:728733.
34. Stefanidis D, Scott DJ, Korndorffer JR Jr. Do metrics matter? Time versus motion tracking for performance assessment of proficiency-based laparoscopic skills training. Simul Healthc. 2009;4:104108.
35. Hesselfeldt R, Kristensen MS, Rasmussen LS. Evaluation of the airway of the SimMan full-scale patient simulator. Acta Anaesthesiol Scand. 2005;49:13391345.
36. Hatala R, Issenberg SB, Kassen B, Cole G, Bacchus CM, Scalese RJ. Assessing cardiac physical examination skills using simulation technology and real patients: A comparison study. Med Educ. 2008;42:628636.
37. Hemman EA, Gillingham D, Allison N, Adams R. Evaluation of a combat medic skills validation test. Mil Med. 2007;172:843851.
38. LeBlanc VR, Tabak D, Kneebone R, Nestel D, MacRae H, Moulton CA. Psychometric properties of an integrated assessment of technical and communication skills. Am J Surg. 2009;197:96101.
39. Hastings A, McKinley RK, Fraser RC. Strengths and weaknesses in the consultation skills of senior medical students: Identification, enhancement and curricular change. Med Educ. 2006;40:437443.
40. Paukert JL, Richards ML, Olney C. An encounter card system for increasing feedback to students. Am J Surg. 2002;183:300304.
41. Links PS, Colton T, Norman GR. Evaluating a direct observation exercise in a psychiatric clerkship. Med Educ. 1984;18:4651.
42. Lane JL, Gottlieb RP. Structured clinical observations: A method to teach clinical skills with limited time and financial resources. Pediatrics. 2000;105(4 pt 2):973977.
43. Ross R. A clinical-performance biopsy instrument. Acad Med. 2002;77:268.
44. Kroboth FJ, Hanusa BH, Parker SC. Didactic value of the clinical evaluation exercise. Missed opportunities. J Gen Intern Med. 1996;11:551553.
45. Scheidt PC, Lazoritz S, Ebbeling WL, Figelman AR, Moessner HF, Singer JE. Evaluation of system providing feedback to students on videotaped patient encounters. J Med Educ. 1986;61:585590.
46. Stone H, Angevine M, Sivertson S. A model for evaluating the history taking and physical examination skills of medical students. Med Teach. 1989;11:7580.
47. Burch VC, Seggie JL, Gary NE. Formative assessment promotes learning in undergraduate clinical clerkships. S Afr Med J. 2006;96:430433.
48. Haertel E. Getting the help we need. J Educ Meas. 2013;50:8490.
49. Dweck CS. Motivational processes affecting learning. Am Psychol. 1986;41:10401048.
50. Lineberry M, Soo Park Y, Cook DA, Yudkowsky R. Making the case for mastery learning assessments: Key issues in validation and justification. Acad Med. 2015;90:14451450.
51. Lord SJ, Irwig L, Simes RJ. When is measuring sensitivity and specificity sufficient to evaluate a diagnostic test, and when do we need randomized trials? Ann Intern Med. 2006;144:850855.
52. American Board of Medical Specialties. Standards for the ABMS program for maintenance of certification (MOC) for implementation in January 2015. http://www.abms.org/pdf/Standards%20for%20the%20ABMS%20Program%20for%20MOC%20FINAL.pdf
. Accessed December 15, 2015.
53. Cronbach LJ. Wainer H, Braun HI. Five perspectives on validity argumentTest Validity. 1988:Hillsdale, NJ: Routledge; 317.
54. Lane S. Validity evidence based on testing consequences. Psicothema. 2014;26:127135.
Reference cited in Table 2 only
55. Huang GC, Newman LR, Schwartzstein RM, et al. Procedural competence in internal medicine residents: Validity of a central venous catheter insertion assessment instrument. Acad Med. 2009;84:11271134.