RandomPlayerx
Recruit
I can only reiterate what several others have already said, in that your results are likely to be skewed/biased (to be precise selection bias). I say this because I do not think that the MB subreddit is a accurate representation of the BL total playerbase or even the total MB playerbase.You even acknowledge this in your initial disclaimer.
It is very likely that the MB subreddit has a higher proportion of "hardcore" (for lack of a better word) fans and less casual fans, when compared to the overall BL/MB playerbase. "Hardcore" fans are more likely to have already played Warband or even the original MB before playing Bannerlord. In other words, the percentage of veteran players (= BL was not their first MB game) is probably higher in the MB subreddit than it is in the total BL/MB playerbase. If this is indeed the case, then you can't draw conclusion on the overall BL or MB playerbase based on polls of the MB subreddit, because the sample (MB subreddit) is not representative of the population (MB/BL playerbase) that you want to analyse .The results/estimates from the polls suffer from (selection) bias - they overestimate the proportion of veteran players.
Increasing the sample size also does nothing against bias. To build on the example that Grindelvald gave, if you want to find out who the general population considers to have better sportshoes - Nike or Adidas - and you conduct an sample in the headquarters of Adidas, then your sample size can be 100, 1000 or even 10000 - the number doesn't matter regarding bias, because you are still asking only employees of Adidas. Adidas employees are not representative of the overall population - and the results are biased.
It is very likely that the MB subreddit has a higher proportion of "hardcore" (for lack of a better word) fans and less casual fans, when compared to the overall BL/MB playerbase. "Hardcore" fans are more likely to have already played Warband or even the original MB before playing Bannerlord. In other words, the percentage of veteran players (= BL was not their first MB game) is probably higher in the MB subreddit than it is in the total BL/MB playerbase. If this is indeed the case, then you can't draw conclusion on the overall BL or MB playerbase based on polls of the MB subreddit, because the sample (MB subreddit) is not representative of the population (MB/BL playerbase) that you want to analyse .The results/estimates from the polls suffer from (selection) bias - they overestimate the proportion of veteran players.
Increasing the sample size also does nothing against bias. To build on the example that Grindelvald gave, if you want to find out who the general population considers to have better sportshoes - Nike or Adidas - and you conduct an sample in the headquarters of Adidas, then your sample size can be 100, 1000 or even 10000 - the number doesn't matter regarding bias, because you are still asking only employees of Adidas. Adidas employees are not representative of the overall population - and the results are biased.