Under the alternative-hypothesis conditions, the means of rti Q3 and rti MSRW departed from the null expected values in the predicted direction, and the departures were more pronounced as the REs increased.
Second, in the continuous case rti Q3 tended to perform better than rti MSRW under the alternative-hypothesis conditions.
As a summary, the general results suggest that both rti Q3 and rti MSRW, would be good at detecting individuals with strong REs in tests of 40 or more items.
Finally, the rti Q3 and rti MSRW estimates were obtained based on the calibrated item parameters and the EAP scores.
Figure 1 shows the distribution of the simulated rti MSRW values (dashed line) together with the distribution of the real values (solid line).
According to rti MSRW the corresponding percentage would be 88%.