One of my classes is initiating a research project. We are examining differences (if any) in physical activity among residents of an area after a public recreation area becomes available sometime in 2013. We are using a recall instrument (Godin & Shephard, 1985) to gather baseline data regarding current levels of physical activity and plan to use the same instrument multiple times over the course of the next 18 months (and maybe longer) to get comparison data. I had envisioned this as a simple pre and post test design but it was pointed out to me that this is not a particularly strong design. We will not be able to access the same individuals; we will be collecting random samples (using online and in person methods) at various time points. Use of a comparison group seems very complicated as identifying a similar group (rural residents in similar communities who do not have access to a new recreation area) would be difficult. According to Tabachnick & Fidell, we would need 50 collection points to use time series analysis and, while future students may continue this project to this point, it is not going to happen during the short term - I estimate 5-6 survey times (we would like to do seasonal data collection assuming that there will be seasonal differences in activity).
Any suggestions? We are working on the material for IRB now so this is the time to add to the data we collect and/or alter the survey. Is there any way to use some type of survival analysis? I had minimal exposure to survival analysis in a class but am not that knowledgeable about the requirements. Other longitudinal techniques I am aware of require multiple observations from the same individual(s), which we do not believe we can obtain. If we were to try to use a comparison group, how similar do they need to be? I am afraid that matching demographics, area population density, access to facilities, etc. would be very difficult.
Anything offered, including readings that might be helpful, is greatly appreciated.
Advertise on Talk Stats