Test Taker Experiences and Beliefs A Case Study: Recent Examinees Krista Breithaupt, MCC Greg Pope, Yardstick Inc. Bruno D. Zumbo, University of British Columbia Snapshot – Experiences and Beliefs from our test takers, February 2011. Evolving notions of social & psychological variables in assessment. TOPICS 2 Questionmark Survey • Case Study: MCC test takers were invited to respond to a survey in the first two weeks of February of 2011. • These responses are the most current snapshot of MCC test taker experiences and beliefs. – MCC sent out an email invitation to 24,000 test takers in our repository who had registered for exams in the last five years. • We obtained a fantastic response: 3296 respondents completed the survey, most in just a few days! • We used Questionmark Perception to author and administer the web-based surveys. – French and English versions of the survey were selectable. • What follows is a summary of results. 3 Demographics The MCC is currently examining our past exam results to confirm no gender bias is evident in performances of candidates. Most took the exam within 5 years. 4 Beliefs of Test Takers - Fees “I believe that exam fees are reasonable for candidates.” The MCC exams are comparable to other certification exams in cost. However, finding ways to operate more cost-effectively will benefit our candidates. Reviewing existing processes for operational efficiency can reduce our costs in future.. 78% Disagree 5 Beliefs of Test Takers – Adaptive Designs and Score Apeals “I believe that computerized adaptive tests (where different sets of questions are administered to each candidates depending on their answers to previous questions on the exam) are fair.” 71% Strongly Agree or Agree 71% Strongly Agree or Agree “I believe that if I need to appeal my exam score that there is a fair and accessible method for me to do so.” 6 Beliefs of Test Takers – Bias and Security “I believe that the questions on the certification and licensing exams are not biased against one gender or a specific racial demographic group.” 91% Strongly Agree or Agree 35% Strongly Agree or Agree “I believe that cheating on certification and licensing exams is a significant problem.” 7 Beliefs of Test Takers – Security… “I believe that it is easy to obtain leaked actual exam questions on the internet if I wanted to find them.” Perhaps the biggest threat to test security is collusion among test takers in high stakes examinations - sharing what they recall after taking the test. More and more testing programs are taking pro-active steps to identify and act when violations occur over copyright protected test content. 19% Strongly Agree or Agree 8 Beliefs of Test Takers – Security… “I believe that using automatic statistical methods to detect people cheating on tests should be done more often.” 62% Strongly Agree or Agree “I believe that biometric authentication (e.g., fingerprint analysis) is a good way to identify candidates.” 64% Strongly Agree or Agree Candidates see increased value in their credential if we’ve been aggressive about security. 9 Beliefs of Test Takers – Security… “I believe that it is acceptable to be video and/or audio recorded during an exam.” 63% Strongly Agree or Agree 36% Strongly Agree or Agree “I believe that remote exam proctoring using web-cameras and audio listening devices is as effective as inperson exam proctoring methods.” At MCC, we administer our exams outside of secure test centers. This has been a convenient way to hold exams in convenient locations (Medical Schools and Hospitals) across Canada. It may be time to closely examine the impact on fees, and potential security risks this convenience creates. 10 Beliefs of Test Takers – Fairness and CBT 63% Strongly Agree or Agree “I believe that scores that I obtain on the exam are an accurate reflection of what I know and can do.” “I prefer taking computer-based exams versus paper-and-pencil exams.” Only 35% Strongly Agree or Agree While only 65% of our test takers see the exams as a strong predictor of practice, this may be appropriate and true! We need to expand our measurement to broader competencies. 11 Clinical Skills Exams (OSCEs) “ The exam where candidates attend a series of clinical situations with a standardized patient is an accurate measure of what a physician would do in practice.” If only 54% of candidates see the OSCE as a valid representation of their practice skills. What is contributing to their belief these are not representative performances? Beliefs of Test Takers – Meaning of Scores “I believe that the exam scores I obtain accurately predict how well I will do in my profession; e.g., higher exam scores = higher pay or greater promotions.” 56% Strongly Agree or Agree “I believe that if I took the same certification or licensing exam again that I would obtain a similar score.” 90% Disagree! Candidates recognize that success in the profession depends on a broad range of skills and expertise that are outside of the domain of the exams. However, many don’t trust the reliability of scores. What is the basis of their view the exam may lack relevance? 13 Beliefs of Test Takers – Score Reports and Practice Tests 93% Strongly Agree or Agree “I believe that the reports I obtain regarding my performance on the exam are clear and concise.” 65% Strongly Agree or Agree “I believe that taking practice tests before I take the actual certification or licensing exam reduced my stress and anxiety when taking the actual exam.” MCC could explore how to improve communicating what scores mean. Focus groups are often useful in determining how the reports are in fact used, and where misinterpretations may occur. Additional practice tools would be well-received too! 14 WHAT DOES ALL THIS MEAN FOR MCC and THE TESTING INDUSTRY? 15 An Evolving Notion: …a social psychology of assessment & testing is developing that forces us to consider the experiential and contextual factors in our exams. 16 Experiential Factors • We are beginning to form a more formal understanding of the social psychological and experiential variables that may be relevant to assessment and testing. – Social psychological (experiential and contextual) variables may have us asking familiar questions about: • The role (effects) of attitudes toward testing. • The role (effects) of emotional factors like stress. • The role (effects) of test taker characteristics and context they live in (e.g., gender). 17 Experiential Factors • Lets us remind ourselves of our core responsibilities in test development: – Providing test scores (and test data) that fosters valid inferences and decisions from the test scores. – Zumbo (2007, 2009) has presented a view of test validity which shines a light on these contextual and social psychological (experiential) variables drawing them out of the background and bringing them up front in the test validation process. 18 Experiential Factors • But what are our responsibilities as test developers and providers for the experiential factors? – With our responsibilities in mind, we can begin to imagine that there are experiential variables that pose challenges to valid score interpretations – Whereas, other factors are, at worst, nuisances but don’t muddy the construct assessed. 19 Experiential Factors • In terms of our responsibilities we can work from advice from Messick and Cronbach. – We can work from the idea of tracing the source of the experiential factors to construct relevant versus construct irrelevant factors. – When making valid inferences from the test scores, we must distinguish between • Deterrents (construct relevant variance) • Nuisances but not deterrents (construct irrelevant variance) 20 Lets take as an example the experiential variable that may require more attention as a potential deterrent: Examinee Stress 21 Experiential Factors • The matter of stress now comes down to tracing back the “source” of the stress to either construct irrelevant variance or construct relevant variance and this requires input from the test design and test score use in the profession. – We can make this matter more complex by looking at the moderating effect of test taker gender. 22 “When I am taking a certification or licensing exam I feel a great deal of stress and anxiety.” Even in a convenience sample of responders, we see important differences by gender. Validation research is needed to understand this factor. Χ2 = 98.2, 3 df, p <.0001 Agree/Strongly Agree Females: 91% Males: 81% In the total sample, 86% strongly agree or agree. 23 An example of a variable that may be nuisance but not necessarily a deterrent to valid score inferences: Test Fees 24 Age Range Analysis: “I believe that exam fees are reasonable for candidates.” We expect new professionals to object more strongly to fees, compared with those more advanced in their practice. This may be useful for costing programs, but is not significant, nor a validity threat. 25 Rating Concordance – Security Web/Audio by OSCE Validity • Test takers who believe the OSCEs valid are supportive of greater security, as are those who see cheating as a problem. • Younger test takers are more likely to see cheating as a problem. • These are statistically significant associations, but what does this mean for how we design future assessments? Future Considerations • In order to drive program and service improvement, it is important to reach out to our test takers and understand their experiences. • A survey or focus group approach is useful in determining/validating program strategy. • These trends are similar to those in other licensure examinations. • What are our responsibilities for test taker attitudes and beliefs? – How can we meet the challenge of better service and improved communication in evolving our programs? 27 Krista Breithaupt kbreithaupt@mcc.ca Greg Pope gregp@getyardstick.com Bruno D. Zumbo bruno.zumbo@ubc.ca 28