Conditional Statistical Inference With Multistage Testing Designs

In this paper it is demonstrated how statistical inference from multistage test designs, where students are administered different modules of items depending on their responses to earlier modules, can be made with the method of conditional maximum likelihood. It is shown how the match between item difficulty and student proficiency may result in a better fit of simple measurement models, owing to the avoidance of undesirable response behavior like slipping and guessing. Attention is given to the assessment of model fit. The results were illustrated with simulated data as well as with real data.