Toronto • May 28-29, 2010 • Student Day and Workshops: May 27
Paper: Assessment of work readiness — Rasch analysis of the spinal function sort
Download Presentation PDF:
Session : 15. Sustainable RTW (3/3)
Day/Time: Saturday at 14:00
Room: St. David Room, 3rd Floor
Objectives:
The Spinal Function Sort (SFS)1,developed to quantify workers’ ability to perform work tasks that involve the spine, has been extensively researched in terms of reliability and validity using Classical Test Theory2. The SFS is assumed to measure a single contruct (work ability). Our objective was to utilize Rasch Modelling to evaluate if the SFS items function as part of a unidimensional interval level scale.
Methods:
The SFS contains 50 items; an item consists of a line drawing of a work task with a description below it. The evaluee scores each item from 1 (able) to 5 (unable) on a Likert scale. SFS item data from 260 individuals with low-back pain presenting for Functional Capacity Evaluation were fit to the Rasch Model3. The participants’ ages ranged from 21 to 64 years old. There were 125 females and 173 males. The physical occupational demands of the participants’ jobs ranged from “Sedentary” to “Very Heavy” according to the DOT (1991) classification4. The scoring scheme is predicated on the notion that each item contributes equally to the total score. Since the Spinal Function Sort is a Likert type scale with each item having an equal number of response options and assumed equal intervals between each score, Andrich’s Rating Scale Model5 was used. We measured item fit, threshold ordering, item difficulty (ease of endorsement) and differential item functioning along with unidimensionality of the scale which is prerequisite for calculating summative scores.
Results:
Initially 6 of the 50 items were identified as misfitting i.e. mean squares greater than 1.4 and ZSTD greater than 2.0. Once these were removed, other items demonstrated poor fit. Disordered thresholds within items were common. Patterns were not improved by category reduction. Attempts to rectify these problems were fruitless. Some items in the scale also showed gender differential item functioning. Males reported lower abilities on traditionally female tasks such as dish washing and kitchen floor sweeping despite these tasks being less physically demanding than other tasks that they indicated they were capable of performing.
Conclusions:
The problems with the SFS lack of unidimensionality and lack of local independence could not be overcome by eliminating items or persons or rescaling items. The order of presentation of the items from lighter to heavier tasks and the clustering of similar tasks influences the response for one item with the response on the similar item. A better model might be as a computer administered test where pictures are displayed in a random order and gender neutrality is maintained. Further investigation using factor analysis would be warranted determine if several discrete constructs are included in this instrument.
References:
1.Matheson, L.N., & Matheson, M.L. (1989). Spinal function sort. Rancho Santa Margarita, CA: Performance Assessment and Capacity Testing.
2.Crocker, L, & Algina, J. (1986). Introduction to classical and modern test theory. Fort Worth: Harcourt Brace Jovanovich College Publishers
3.Rasch G. (1960 / 92). Probabilistic models for some intelligence and attainment tests. Chicago:MESA Press
4.U.S. Dept. of Labor. Bureau of Statistics. (1991). Dictionary of occupational titles (Revised 4th ed.). Washington, DC: US Government Printing Office
5.Andrich, D. (1978). A rating formulation for ordered response categories. Psychometrika, 43(4),561-573
