Page 49 - Secondary school students’ university readiness and their transition to university Els van Rooij

Chapter 2
factors, were built and tested. In Chapter 4, the aim was to uncover which factors were related to the outcome factor academic self-e cacy. Background factors in the model were gender, coursework (science or social sciences and humanities), and level of parental education. Other factors expected to be related to self-e cacy were need for cognition, out-of-school academic activities, academic interest, and behavioural engagement. In Chapter 7, we investigated how four factors – academic self-e cacy, self-regulated learning, academic motivation, and degree programme satisfaction – were related to the outcomes university GPA, number of obtained credits in university, and intention to persist. e question of interest was whether these factors would be directly related to the outcomes, or via academic adjustment. In both studies we used the following established t statistics to assess whether the conceptual models tted the data (Chen, Curran, Bollen, Kirby, & Paxton, 2008; Hu & Bentler, 1999; Kline, 2005; Steiger, 2007; Tucker & Lewis, 1973): the ratio of the chi-square to its degrees of freedom (χ2/df), which should be less than 3; the root mean square error of approximation (RMSEA), which should be less than .07; the standardised root mean square residual (SRMR), which should be less than .08; and the CFI and TLI, which should at least be greater than .90, but preferably above .95. Furthermore, we looked at the sizes of the standardized path coe cients to assess the relative in uence of the factors.
2.3.2 Latent pro le analysis
In Chapter 5 the aim was to identify student engagement pro les in the last grade of secondary education and to see how these pro les related to academic adjustment and achievement one year later in university. To identify di erent pro les, we used latent pro le analysis in Mplus. Latent pro le analysis is a model-based type of cluster analysis, used to identify if there are hidden groups in the data based on the means of several continuous variables (indicator variables) and, if so, the number of groups that provide an optimal t with the data. We used nine indicator variables, under three headings: (1) behavioural engagement (behavioural engagement and self-e cacy: e ort); (2) cognitive engagement (surface learning; deep learning; metacognitive learning; self-regulated learning; (3) intellectual engagement (need for cognition; academic interest; self-e cacy: understanding). Two- to six-class solutions were tested. e optimal number of groups was identi ed by looking at the following t statistics: Akaike’s information criterion (AIC; Akaike, 1987); the Bayesion information criterion (BIC; Schwartz, 1978); the Adjusted BIC (ABIC); the Vuong-Lo-Mendell-Rubin likelihood ratio test (VLMRT; Vuong, 1989); and
48