The aim of this study was to investigate the hypothesised unidimensional nine-factor structure of the VPAT in a South African Foundation Phase learner sample from Port Elizabeth, by exploring the reliability and validity of the measure using Cronbach’s alpha and goodness- of-fit indices. In summary, although the results showed marginal model fit and sufficient reliability for certain sub-tests, the VPAT did not meet sufficient levels of reliability and validity to be deemed valid and reliable in its current structure with regard to the sample group of learners based in the Eastern Cape. These results may be due to a variety of factors which will now be discussed.
The first possible reason is that the low levels of internal consistency reliabilities and insufficient model fit obtained in the data analysis may be due to the possibility that specific items within each sub-test that are not able to measure the visual perceptual constructs proposed by the VPAT. On a closer inspection of each specific item, which compose each sub-test within the VPAT, the researcher found that there was a combination of several items that were potentially problematic in each sub-test, and this could account for the questionable
132 psychometric properties of the measure. In the original development of the scale, certain items were omitted from the sub-test and that resulted in slightly more significant levels of Cronbach’s alpha. However, when the same items were removed in the analysis performed in this research, the alpha levels remained below .70, suggesting a low level of internal consistency for the sub-test.
According to neuropsychological models of information processing of visual perception, it is theoretically possible that the Foundation Phase learners that participated in this research had varying levels of visual perception developed, and thus, were unable to consistently complete the items for each sub-test. The development of visual perception, according to the information processing paradigm, is dependent on the receptive fields in the visual cortex and how well the individual is able to respond to the different visual stimuli according to neural networks and the nervous system. If the receptive fields are still unable to properly process the information presented to individuals, they will be unable to differentiate between lines, edges or the orientation of the stimuli. Therefore, they will be unable to complete the tasks set by each sub-test within the VPAT.
The second possible reason for lack of support for the unidimensional hypothesised model of the VPAT could be due to the fact that items from each sub-test did not surpass the factor loading cut-off point of 0.35 with regard to the sub-scales dominant factor. There was not one sub-scale in any of the analyses executed where all the items of each sub-test consistently loaded on their prospective factor. According to Brown et al. (2006), the fact that items for some of the sub-scales consistently loaded on their primary factor, whereas items for other sub-scales loaded on other sub-scales could result in items being multidimensional which would suggest a mixed picture with respect to the factor structure of the VPAT. The point is
133 that the results from the CFA, bifactor CFA, ESEM and bifactor ESEM do not provide support for the unidimensional assumption regarding the VPAT measure of visual perception.
A third possible reason for the VPAT being inconsistent in the South African context and not replicating the results found in the Western Cape by Clutten (2009) is due to cultural differences. It is possible that visual perception, according to neuropsychological models of information processing, specifically related to visual perception, does not merely transfer cross- culturally as seemly as first predicted for the South African context. Here, it is difficult to claim which of these results are problematic because of cultural differences of the Foundation Phase leaners from Port Elizabeth, and which are accounted for by problematic items within each of the sub-tests. Thus, the researcher recommends that future research is required for this to be better understood. It is important that when a measure that has been regarded to be standardised for a whole country, is able to tap into the cultural differences that could possible effect the development of visual perception. An emic understanding of the development of visual perception according to a neuropsychological information processing approach in the South African context from a bottom-up and top-down perspective will assist in the understanding of the development of visual perceptual constructs in diverse settings.
A fourth possibility for the lack of reliability and validity of the VPAT in this context is that the instructions for taking part in the measurement were not clear enough to be accurately interpreted by the Foundation Phase learners in the Eastern Cape. It is also possible that learners did not have the required level of perceptual skills. Thus, Clutten (2009) argues that, according to the neuropsychological theory of visual perception described by Gordon (2004), aspects of visual perception are acquired over different stages of development, and some individuals mature faster in their acquisition of visual perceptual constructs, such as visual form constancy, compared to other individuals. It is possible that this sample group of Foundation Phase leaners
134 may not have developed the specific visual perceptual skills required to complete the VPAT. Therefore, although the factorial validity of the hypothesised nine factor model of the VPAT showed promise according to specific goodness-of-fit indices specified by Byrne (2012) the hypothesised model still exhibited a marginal model fit with insufficient validity reliability.
A final possible reason for the poor measurement results for the VPAT in the South African context is related to the theoretical underpinnings of the test. Clutten (2009) described the construct of visual perception as being complex in nature. She accepted defined that visual perception is multidisciplinary whereas there is actually no real consensus about the definition of the construct. The test was developed with an overarching definition of visual perception based on theories with roots in neuropsychology, developmental psychology and cognitive psychology (Clutten, 2009). Each sub-factor was defined based on definitions from a variety of visual perception tests with no basis in any one individual theory. This research interpreted the results of the test from the broad perspective of information processing and neurophysiological structures in the brain. It is therefore possible that the constructs as used and described in the VPAT do not fit with the ‘true’ theoretical underpinnings of each sub- factor, resulting in poor levels of reliability and validity.