My data will compare student digital experiences in three universities. Two categories of students (distance and direct/regular students) would be considered in each university. The data will compare variables (such as access to institutional digital devices, the use of personal devices, experiences of digital technologies in the curriculum and attitudes towards the digital technologies in the digital learning environment).
Christopher S Bova
- It depends how far apart the "time points" are. Longitudinal data provide the opportunity to identify responses that are not consistent or plausible over time because responses from the same individuals can be linked and analysed. For example, arespondent may report smoking at one survey but at a later survey report to the contrary by saying they have never smoked. It would not be possible to identify such errors in cross-sectional data.
- There may also be a ‘learnt component’ within longitudinal surveys whereby participants become more accustomed to the questions over time and are likely to find them easier to answer. In addition, there is evidence in the literature that individuals become ‘conditioned’ by repeated surveys ie responses given in one survey may be influenced by those given in previous surveys or questions may be answered differently when participants know the issues being studied (Anstey & Hofer 2004).
- When there are only two time points of assessment for each subject, such as in a simple pre-/posttreatment or baseline to single follow-up design, ANCOVA, and variations of it, may be suitable to answer important research questions. For example, if the researcher wants to know if two medication groups with different mean symptom levels at baseline change differentially in their symptom levels from baseline to follow-up, ANCOVA indicates if they differ on the follow-up (the dependent variable), holding the baseline (covariate) constant, statistically, i.e., it is assessing differential group change. In some cases, a simple difference score (‘change score’) analyzed as a dependent variable without using the baseline as a covariate might be appropriate, but generally ANCOVA is superior in that the former can be thought of as a specific instance of ANCOVA where the linear regression of follow-up on baseline is assumed to have a slope of one, which may not be true. In fact, the baseline versus follow-up relation may even be nonlinear, which can be accommodated in an ANCOVA model....
- This kind of analysis assumes the time difference between baseline and follow-up is the same (or nearly so) for each subject or that it is irrelevant whether it is or not. If that is not the case, the time difference can be accounted for by introducing it as another covariate and looking for interactions in which the relation of follow-up and baseline differs depending on the time difference (perhaps the baseline to follow-up relation and correlation becomes increasingly attenuated as the time difference becomes greater).
Further Reading
Locascio JJ, Atri A. An Overview of Longitudinal Data Analysis Methods for Neurological Research. Dementia and Geriatric Cognitive Disorders EXTRA. 2011;1(1):330-357. doi:10.1159/000330228.
0 votes
0 thanks
Christopher S Bova
You are welcome.
Feel free to "Recommend" my answer if it helped you.
0 votes
0 thanks
- Badges
- Science topic
More Maysam Toghraee's questions See All
Similar questions and discussions