Abstract:
|
Longitudinal data collection and analysis techniques are becoming increasingly common in many disciplines. A recent search for 'longitudinal modeling' in a single area of application over a five-year period resulted in nearly 9,000 published articles. In this paper we present the results of a systematic review of longitudinal modeling practices observed in one area of application, education research literature. We systematically recorded the data, analysis, and presentation properties of a selection of 200 published articles, reporting on such properties as sampling technique, data properties such as missing data, analyses applied, and presentation of results. Early results show that in this area of application it is rare to report power analyses, describe sampling methods, present model equations, or address missing data, among other issues. In general we observe that most researchers apply either restrictive, classical models such as repeated-measures ANOVA, or rely on currently popular methods in the discipline such as HLM. Through this paper we describe the most common omissions and recommendations for best practices in collecting, analyzing, and reporting on longitudinal data.
|