Evaluating repeated diary study responses: Latent curve modeling

Abstract

This case study describes how an intensive repeated measures longitudinal diary study was analysed using exploratory and confirmatory factor analyses, invariance testing, and latent curve modeling. Participants were 390 students in a first-year general education course in a research-intensive university who were asked how strongly they experienced a variety of achievement emotions over an assessment period. The emotions presented to students were taken from three previous studies (i.e., Pekrun, Kitayama, and Buchtel). A final sample of 166 students completed on average nine diaries entries; three each in the week prior to the mid-term test, the week of the mid-term test, and the week the test results were released. After concluding that the emotion models provided in the previous studies were not suitable for latent curve model analysis, an exploratory factor analysis concluded that five emotions underlay the student responses. These emotions had metric invariance across the nine times of administration and fit three week-long latent curve models. The final models showed that positive emotions declined, while negative emotions increased, as the test got closer; while after the test, the positive emotions increased and negative emotions declined. Relations to prior GPA and test scores were statistically significant only after the results were known and only in relation to the start level for the emotions. The starting value of positive emotions was positively associated with GPA and test score, while the starting value of negative emotions was negatively associated with GPA and test score. No effect was found for the changeability in emotions.