Visual-spatial skills have been an integral modality both for assessment aspects of intelligence and for investigating its components. The search for processes underlying visual-spatial ability has generated various proposals of basic components explaining individual differences.
In this study, we were motivated by a recent proposal made by Hegarty and Waller (2004), and Kozhevnikov and Hegarty (2001). They proposed a two-factor model with perspective taking and rotation as the two underlying factors. However, because solving visual-spatial tasks may be approached with more than just one specific strategy, it seems that characterizing a spatial skill by using a single, primarily operational, terminology such as rotation or perspective taking is insufficient to account for individual differences in spatial abilities.
Accordingly, we propose to add another underlying process to the two-factor model, which we term transformation. The proposed factor refers to another level of visual-spatial task performance, involved in a much wider set of operations and even in a combination of multiple procedures. Thus, transformation refers to more than just one dominant operation such as rotation or changing perspectives. As such, this factor may not be limited to the visual-spatial domain and might be implicated in wider aspects of human cognition.
To examine this issue, 133 participants completed 7 tasks assumed to represent well the basic components involved in visual-spatial ability.
CFA of the resulting data was utilized to model the results and compare the fitness of one-, two- and three-factor models. After establishing the measurement model, we performed a second-order structural model analysis to assess the casual paths and clarify the correlations among the three first-order factors.
The overall statistics for the three-factor model indicated an excellent fit to the data (Model Fit: χ2 (11) = 6.57, p = .83, χ2/df = 0.6, SRMR =.029, RMSEA = 0.00, CFI =1.00). A χ2 difference test indicated significantly better fit of the three-factor model over the single-factor model (χ2 (3) = 9.12, p = .03). The second-order model provided a very good fit to the data (Model Fit: χ2 (13) = 6.99, p = .90, χ2/df = 0.54, SRMR =.030, RMSEA = 0.00, CFI =1.00), and a good account for the correlations among the three first-order factors.
The analyses produced very favorable evidence supporting three well differentiated, yet related, components of visual-spatial skills: rotation, perspective, and transformation. Second-order structural model analysis supported the three-factor model, indicating some degree of a common, underlying cognitive process of the three components. This second-order factor could possibly reflect a building block of general intelligence.
