To what degree does the missing-data technique influence the estimated growth in learning strategies over time? A tutorial example of sensitivity analysis for longitudinal data

Abstract

Longitudinal data is almost always burdened with missing data. However, in educational and psychological research, there is a large discrepancy between methodological suggestions and research practice. The former suggests applying sensitivity analysis in order to the robustness of the results in terms of varying assumptions regarding the mechanism generating the missing data. However, in research practice, participants with missing data are usually discarded by relying on listwise deletion. To help bridge the gap between methodological recommendations and applied research in the educational and psychological domain, this study provides a tutorial example of sensitivity analysis for latent growth analysis. The example data concern students’ changes in learning strategies during higher education. One cohort of students in a Belgian university college was asked to complete the Inventory of Learning Styles–Short Version, in three measurement waves. A substantial number of students did not participate on each occasion. Change over time in student learning strategies was assessed using eight missing data techniques, which assume different mechanisms for missingness. The results indicated that, for some learning strategy subscales, growth estimates differed between the models. Guidelines in terms of reporting the results from sensitivity analysis are synthesised and applied to the results from the tutorial example.

Polygenic predictors of age-related decline in cognitive ability

“…The correlation sizes among the PGSs were generally low: the strongest absolute relation between any of the PGSs was that between the education and BMI PGSs, estimated at r = -0.26. Despite its poor phenotypic prediction of the depressive state as described above, the PGS for major depressive disorder correlated in the expected direction with that for neuroticism (r = 0.23) [60,61]. As planned in the preregistration, we ran a parallel analysis of the fourteen PGSs using the psych package for R: this indicated that there were four factors in the data, with no evidence for a strong "general" factor.…”

Appearance Esteem Trajectory According to Three Different Sources of Support Among Adolescents Over a School Year

“…Nevertheless, this study had some limitations that should be considered. Although the sample size was sufficient to test the hypotheses of this study (Coertjens et al, 2017), it was not large enough to provide statistical power to test other moderation hypotheses. For example, a previous study found that gender did not moderate the relationship between social support from family and from peers and body dissatisfaction (Kirsch et al, 2016).…”

“…This type of analysis has been proven to allow the examination of individual differences in within-person variations and co-variations over time (Hoffman, 2007). Given the three measurement waves (Burchinal et al, 2006) and the small amount of missing data (Enders, 2011), the sample size provided sufficient statistical power to detect slope differences between groups even for small effect sizes (Coertjens et al, 2017). Moreover, according to Arend and Schäfer (2019), the power estimation of this two-level model study is considering as sufficient.…”

Domain-general cognitive functions fully explained growth in nonsymbolic magnitude representation but not in symbolic representation in elementary school children

“…This test was nonsignificant (chi-square distance = 272.51, df = 248, p = 0.14), indicating that the MCAR assumption held. Since the MCAR assumption held and the sample size was sufficient, list-wise deletion can be applied to obtain adequate parameter estimates [76].…”