Open Collections

Preliminary item and subscale analysis of the seated postural control measure

University of British Columbia

The purpose of this research was to conduct a preliminary item and subscale analysis of the Seated Postural Control Measure (SPCM) to determine: 1) whether item function was consistent with the purpose of the measure; and, 2) whether the items were appropriately placed in the theoretically-derived subscales. Statistical review of item properties including item difficulty, item discrimination, item homogeneity and item-subscale congruence was undertaken using data from the two reliability studies conducted in 1991 and 1993. Analysis of these item properties required calculation of the item difficulty index, the individual gain discrimination index and the analysis of inter-item correlations and item-subscale correlations. Inter-item correlations suggested that item homogeneity was appropriate for all but one of the assessed item pairs. Item difficulty in the pre-seating condition was within the anticipated range for 31 of the 34 items. Item sensitivity as assessed by the individual gain discrimination index, was within the anticipated range for 22 of 34 items. Further exploration of the data revealed possible causes for the low discrimination index for 12 of the items including, high item difficulty, lack of effects of the intervention, and the occurrence of negative change which was not detectable by the discrimination index. Results of the assessment of item-subscale congruence supported the grouping of items in the Function scale and two of the four Alignment subscales. Results of the Alignment subscale-section congruence assessment supported the groupings of all subscales except the Head subscale. Limitations of the research which affected the strength of the conclusions were: the retrospective nature of the study; the small sample size for some items; the poor reliability of some of the alignment item data used for two of the hypotheses and the possible effect of reliability on inter-item correlations; the lack of an external measure of change; and, the constraints in the interpretation of ordinal level data. Implications of the study findings for future test development include the need to delete one redundant item, modify some items and possibly make small changes to the subscale structure. Future research should focus on establishing the responsiveness of the measure and the minimal clinically significant score difference.

Thesis/Dissertation

application/pdf

2009-02-06

For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use.

http://hdl.handle.net/2429/4278

Rehabilitation Sciences

University of British Columbia

UBCV

DSpace