Using EFA to Determine Factor Structure of a Computer-Based Version of the Purdue Spatial Visualization Test: Rotations (PSVT:R)

Abstract

Literature shows that spatial skills, and in particular, mental rotation skills, are predictors of success in STEM. Students who have strong spatial visualization skills are more likely to demonstrate better academic performance and higher retention rates in STEM. Several instruments are used to measure mental rotation skills, most of which are paper-based; these include the Mental Rotations Test (MRT), Rotated Colour Cube Test (RCCT), and Purdue Spatial Visualization Test: Rotations (PSVT:R). To measure the range of skills typically seen in undergraduate engineering students, the PSVT:R has been historically preferred for its use of a variety of 3-dimensional shapes, which are appropriately challenging to visualize, and for its established reliability and validity. A data-rich computer-based version of the test offers several advantages over the paper-based test; however, its reliability and validity must be established. We present the analysis of the results of a computer-based version of the PSVT:R administered to first-year engineering students at a mid-sized, public university in the United States. We use an exploratory factor analysis (EFA) to determine the number of latent variables being measured by the instrument in our data. We determine the number of latent variables to be one, with good reliability, which is consistent with the paper-based instrument. In future work, we plan to use a confirmatory factor analysis (CFA) to show evidence of validity of the computer-based PSVT:R.