| Many candidates who take certification exams do not pass for various reasons. When groups of candidates are not successful, they are more likely to question the validity of the exam. Extensive research has been conducted on the relationships between nontest factors and candidate pass-fail rates, but research examining the interactions between test-related factors and pass-fail rates is sparse in two important areas: the results cannot be generalized across a specific population of certification exams, and theoretical models of candidate pass rates cannot be developed from the literature. A quantitative, non-experimental survey research study was conducted to explore and confirm the test development (TD) predictors of overall and first-time pass rates using chi-squared automatic interaction detector, categorical regression, and cluster analysis. Participants included 803 representatives from different Credentialing Organizations. Out of 71 TD variables evaluated, four (4) were confirmed as significant predictors of overall pass rates: eligibility requirements (p < .001), equating design (p < .001), computer-based test design (p < .001), and multiple item formats (p < .05). Eligibility requirements produced the greatest changes in overall pass rates. Two of the TD variables were confirmed as significant predictors of first-time pass rates: test purpose (p < .001) and multiple item formats (p < .05). Test purpose yielded the greatest changes in first-time pass rates. Multiple item formats also surfaced as the only predictor in both pass-rate models, providing strong evidence that candidate pass rates are higher when only one item format is used on exams. Using these significant TD predictors, theoretical models were also developed for overall and first-time pass rates as a framework for future research. To improve external validity, recommendations were made to researchers to empirically test these theoretical models or entirely replicate this study using a more representative sample of certification exams. Future research should also be conducted to understand why certain eligibility requirements, exam competency levels, equating and computer-based test designs, and exams with multiple item formats result in lower pass rates. To give candidates every opportunity to be successful, at a minimum, credentialing organizations should design their certification exams with one type of item format. |
