I have a question on what meta-analysis is best when aggregating means using non-randomized longitudinal data with a control group that significantly differs from the active group at baseline.
The outcome in question is depression, and the question is whether depression scores soar or fall in the transition from work to retirement. From the pool of data one group has retired, with data available for depression before and after retirement. Another group is continuing in work, where depression data was gathered at baseline and and at the last date of data collection. The group continuing in work is often very close in age to the group retiring, but I have noticed that the groups have noticably different depression levels at baseline (often measured by CES-D).
Another question: several of the studies have split retirement into "voluntary" and "involuntary" groups. Is it scientifically sound to combine the means to get a score of depression relative to overall retirement, when trying to answer the question of whether people get more or less depressed in the transition from work to retirement?
The outcome in question is depression, and the question is whether depression scores soar or fall in the transition from work to retirement. From the pool of data one group has retired, with data available for depression before and after retirement. Another group is continuing in work, where depression data was gathered at baseline and and at the last date of data collection. The group continuing in work is often very close in age to the group retiring, but I have noticed that the groups have noticably different depression levels at baseline (often measured by CES-D).
Another question: several of the studies have split retirement into "voluntary" and "involuntary" groups. Is it scientifically sound to combine the means to get a score of depression relative to overall retirement, when trying to answer the question of whether people get more or less depressed in the transition from work to retirement?