A.B.
Senior Member
- Messages
- 3,780
To clarify: I don't actually know if PACE factored these dropouts into the calculations correctly or not, but I'd be interested to here the opinions of people better qualified than I am. And does the reanalysis fully account for it?
There is no way to know what the data would look like without the drop outs. To fill in the missing data, the most recent available results for that person can be carried forward, or the average of the available results of that group can be used. There are some other strategies. In the case of PACE the dropouts are probably related to some degree to worsening of health and therefore the average walking distance would be overestimated. I don't remember what PACE did but think it was one of these two strategies, and I'm not a statistician.
If it comes up in a debate, one can point out that the results are probably an overestimate and consistent with poor health in any case.
Last edited: