Figure 2
Item analysis. (Left panel) For each stimulus, we plot the correlation between performance on this item and LDT accuracy (x-axis) as well as the correlation between performance on this item and WJ-Word-ID raw score (y-axis). Items that were not correlated (threshold r ≤ 0.10, Step 1) with overall test performance or WJ-Word-ID were removed (plotted in red). Items that passed the r > 0.10 threshold are plotted in blue. Pseudowords are depicted in darker colors and real words are depicted in lighter colors. (Right panel) Difficulty (x-axis) and discriminability (y-axis, based on the 2 parameter logistic model) for each item that survived the correlation threshold. Red items were those that were removed in step 2 or 3, based on poor fit statistics of the Rasch (Step 2) or 2 parameter logistic (Step 3) models.
