- Posts: 75
There is not an error in CAT scoring. If you administered a Neuro-QoL CAT, the number of items a respondent answers is expected to be variable. The minimum number of items administered is 4 and the maximum number of items administered is 12. In this case, a person with really high social function is answering with the most positive response for each item. The item bank that is used for the CAT has a limited number of items that do a good job distinguishing people at that high end of social function. This means that the estimated error for that person's score stays above the threshold needed to stop the CAT and the maximum number of items is administered (12). This is a measurement ceiling. All measures have a ceiling - that is, a limit to how much of the range (here, of social function) that can be accurately assessed. The Neuro-QoL CAT is hitting a ceiling at T=60, or one standard deviation above the mean. This means the CAT doesn't do a great job at distinguishing people with really excellent ability to participate in social roles and activities to those with really really exceptional ability to participate in social roles and activities.
You should utilize the final T-score produced by the CAT, not a look-up table. You should not use raw sum scores. The look-up tables are only to be used for their corresponding short form. Quite often, the short form does not have the ability to measure as full of the range of something when compared to the CAT.
We used the Neuro-QOL CATs for Ability to Participate in SRA and Satisfaction with SRA but in the score exports some cases stood out by reaching up to 12 items and a raw score of 60. In the manual, only t-score conversion tables are provided for the short-form versions of these measures and their maximum raw scores do not go as high as 60. In order for a subject to have reached 60, they would have needed to respond with 5's to 12 consecutive items. Why would the CAT not move on after only a few of these responses to calculate the score? We are worried that the scores were not computed properly and the system made some subjects respond to more items than were necessary.