Skip Navigation
Menu

Science

ASCQ-Me measures were evaluated using both classical and modern psychometric theory:

Both exploratory and confirmatory factor analysis were used to examine the unidimensionality assumption for each of the five ASCQ-Me domains (Emotional Impact, Pain Impact, Sleep Impact, Social Functioning Impact, and Stiffness Impact). Results of these analyses supported the unidimensionality of each item bank.PDF icon

After the IRT assumptions had been evaluated and confirmed, we fitted the unidimensional Graded Response Model (Samejima, 1969) to the data to create individual scores based on item calibrations.  Because some items might not be equally valid across different types of respondents leading to bias in measurement, a differential item function (DIF) analysis was conducted for each of the six measures in ASCQ-Me and items showing DIF were removed from the item banks.  

After the full item banks of all five ASCQ-Me measures were defined and all the items had been calibrated, five items were selected from each item bank to create a short form; this would enable users to minimize respondent burden even if they did not have access to the CATs.  Items were chosen to represent the content of the item bank and to represent different levels of severity for each of the five health dimensions.

To examine the discriminate validity, participants were divided into three groups according to their SCD severity scores, representing low, medium, and high level of severity, respectively. Since there were 9 possible severity scores (i.e., 0 to 8), the percentile corresponding to each level of severity in the entire sample  was calculated, and severity scores closest to the 33rd and 66th percentile were regarded as the cut-off values to determine the three severity groups.

For more details about ASCQ-Me’s methodology, see link on the left.


Reference:

Samejima F. (1969). Estimation of latent ability using a response pattern of graded scores. Psychometrika Monograph, 17.

Line graph image.