discarding small samples from reports?

[erdnase17]

If I run a report for all players using a custom stat 'minr freq from SB' I get a certain value, however if I filter for players that had a number of opportunities to raise >= 100 I get a more meaningful value.
How do small samples affect stats, say lots of players with 2/4 for example (2 actions out of 4 opportunities)?
Should I remove these small samples when running reports?

Thanks.

[kraada]

If you're just looking at one number in total the times and opportunities from the low sample size players just add in as any other numbers. Say you have three people with values of:
17/500, 2/4, 1/1

Then the total would be 20 / 505.

It kind of depends what sort of analysis you want to do as to what kind of filtering you want to apply. Do you want to know how to think about an unknown? Filter to players with smaller samples. Do you want to know how to think about a regular? Filter to players with larger samples, etc.

[erdnase17]

If you're just looking at one number in total the times and opportunities from the low sample size players just add in as any other numbers. Say you have three people with values of:
17/500, 2/4, 1/1

Then the total would be 20 / 505.

It kind of depends what sort of analysis you want to do as to what kind of filtering you want to apply. Do you want to know how to think about an unknown? Filter to players with smaller samples. Do you want to know how to think about a regular? Filter to players with larger samples, etc.

Since my database has around 19000 unknowns and just a couple hundred regs this separate analysis makes sense. The unknowns will most likely skew the results if I do a global analysis.