Some Black friday info from infoscout

[DonFerrari]

jonhalo said:

DonFerrari said:

People already posted the fine print that only 2000 receipts were game related... no info on how many HW receipts, maybe even less than 500. So I have quite the hard time believing on their precision.

1. that was last year or the year before, when the overall sample was just 150k
2. this is a further proof, that theire sample size is quite decent. if 90% of all receipts would be game related, theire sample size would be screwed towards gamer and way more unreliable.

as said earlier if you want to do a good preliminary election poll, you don't want 100% political interested people. you wan't 40-50%. just like the average country.
it's never about the sample size. 2000 is totally fine for statsitics. it's about the sample itself

don't seem like that... the post was puting the fineprints of these year

How having 2000 of 250000 pooled being relevant something to prove the sample size is quite decent? If they are pooling for gaming, the other 248000 receipts are completely useless.

And 50% of people interested in politics is quite different from 0,2% of the receipts being relevant.. and 2000 receipts for game related, involving, cards, sw, peripheral and HW... how much of that is HW? How can you safely say they are reliable if perhaps their survey only have 200-500 HW surveyed and without better information about their demographic choice?

People already show how far they were on their statistical for last year... they said 63/37 and was 55/45 or something like that... that is 8 percentual points error, that is a very big margin of error.