Study of Psychometric Properties of Traffic Test with Emphasis on Identifying the Differential Item Functioning with Accounting Guessing Effect

Being proficient in traffic test has a significant effect on reducing costs due to non-compliance with driving regulations. The traffic test biases put validity, fair evaluation, and correct decision in jeopardy. Therefore, the research objective is to study the psychometric features in order to validate the traffic test.

The present applied study is quantitative in terms of type and categorized as studies called "psychometrics". The research population included participants in the traffic test in 2020 that 297 people were selected by simple random. Data were analyzed by R software, and reliability (α=0.7) as well as validity were verified based on differential action analyses.

The research findings showed that the traffic test is an easy test and participants mostly use guessing the correct answer. The discriminant coefficients of all questions were positive and acceptable. Non-based IRT tests converged, and with the exception of nonlinear regression method, which identified questions 11 and 16 as biased, other methods detected only question 16 as biased. Tests based on IRT failed to converge.

The results of the research showed that the traffic test is valid and reliable. Based on differential functioning analyses, there is generally validity due to homogeneity of measurement in reference and focal groups. Nonlinear regression method is also a useful alternative to methods based on IRT and is has good capabilities in tests where there is a guess effect.