Time 2 Sample Size

I was hoping for some advice. I'm looking at the same sample at 2 timepoints.
At the first time point I had 95 participants and at follow up I only got 18 replies back.

Is it worth doing any analysis between the two timepoints, or is the sample at time 2, too small for any reliable results.
Having a quick look at the data, there is a significant positive correlation between Time 1 anxiety, depression with time 2.

Many advice would be great


TS Contributor
I think that the problem is not so much the sample size but the fact that you might have a biased second sample.It is quite possible that there is some pattern in the respondents to the aecond query - i.e, if mostly those answered who had depression while those who did not just ignored the questions, it will look as if therebwas a realtionship even though there wouldn't be any in the whole population.



Omega Contributor
Yeah, this is a common issue. Most likely you are out of luck. Perhaps there is a pattern in characteristics of those who responded to both. If so, you may be able to solely generalize and interpret results solely to those patients.

Ideally lost to follow-up is random, but typically empirically it is not.