3 ms·
Make it opt in, because most people wouldn't like it here. Even if only a certain percentage of people make up your sample size, you can still get valid data.
by jychang 10y ago
Make it opt in, because most people wouldn't like it here. Even if only a certain percentage of people make up your sample size, you can still get valid data.
- philipov 10y agoI suspect that the sort of person who would opt in to this would not be a random sample of the population of HN, and so the data would be biased by default.
- beagle3 10y ago> you can still get valid data. That would be stretching the definition of "valid". You would get statistics about a group of "I don't care I'm being tracked" people, which is surely statistically different than the group of "I do care about being tracked" people. At the very least, they would have significantly different click rates (and involvement) with the subject like "how to disable web site tracking", but I suspect it's deeper than that -- e.g., they will have statistically significant involvement with security related subjects.
- andybak 10y agoYep. I would count any 'opt-in' analytics as irretrievably tainted even if by some miracle the sample size made it useful.