Using Rasch Analysis to Examine the Validity and Reliability of the Praxis tests of the Evaluation in Ayres Sensory Integration

Research Objectives: To investigate the evidence for construct validity and internal reliability of the normative data collected using the seven praxis tests of the Evaluation in Ayres Sensory Integration (EASI)

Design: Psychometric evaluation using the Rasch measurement model.

Setting: EASI datasets were compiled from normative data collection in 53 countries. The EASI is appropriate for use in a variety of clinical settings.

Participants: We analyzed EASI data from 2409 children ages 3-12 (1089 male, 1320 female) with no known medical or educational diagnoses relevant to sensory integration and no siblings who had these diagnoses. Participants were recruited using convenience sampling.

Interventions: Not applicable.

Main Outcome Measures: EASI is a suite of 21 individual tests which measure sensory integration functions. Seven tests examine praxis: Praxis: Positions (PrP), Praxis: Sequences (PrS), Ocular Praxis (OP), Praxis: Following Directions (PrFD), Visual Praxis: Designs (VPrD), Visual Praxis: Construction (VPrC), Praxis: Ideation (PrI).

We examined four indicators of construct validity for each test: (1) item fit statistics (mean squares), (2) rating scale thresholds, (3) principal components analysis (contrast eigenvalues and disattenuated correlations), and (4) differential item functioning based on sex (contrast magnitude and significance). We examined two indicators of reliability: (1) Rasch person reliability indices and (2) strata.

Results: Item Fit: For six of seven tests, >90% of items fit the model. For VPrD, only 79.2% fit.

Rating Scale Thresholds: All thresholds progressed monotonically.

Principal Components Analysis: All tests demonstrated evidence for unidimensionality (eigenvalue of largest contrast < 3 and disattenuated correlations > .57).

Differential Item Functioning: All items had no/negligible DIF (contrast magnitudes < .43).

Reliability: PRIs ranged from .70 - .95. Strata ranged from 2.60 - 5.93.

Conclusions: Rasch analyses revealed strong evidence for validity and reliability of normative data collected using the six of the seven praxis tests of the EASI. These results suggest that the EASI is appropriate for measuring sensory integration abilities in typically developing children. However, further investigation is necessary to confirm that the EASI is appropriate for use in clinical populations.

Author(s) Disclosures: Not applicable.