If you try out 150 items to get the best 20 items for your test using a relatively small sample of cases (N = 30), what process should be used to obtain the true item statistics for the 20 items you have selected?

Question

Quizplus · Accepted Answer

Cross-validation is the correct process to use in this scenario. It involves using a different sample to validate the item statistics obtained from the initial small sample. This helps to ensure that the selected items truly perform well across different samples, addressing the issue of overfitting that can occur when item selection is based solely on a small sample.

If You Try Out 150 Items to Get the Best