If one person has more than one vote, and a response to a survey is technically a "vote", then the vote results must be either considered invalid or listed with a disclaimer noting that it was non-scientific, that people could vote more than once, and should list a margin of error. This is on a technical level, not on an emotional level. I can almost guarantee that there are insufficient checks and balances on the survey results that make the statistical sample a biased sample. A quick look at some documentation on the web shows that what I'm talking about would be considered "Selection Bias".
What I'm saying, and I don't for the life of me know why it's so difficult a concept to grasp, is that if the same person takes the survey more than once, and the survey system considers each submission as a unique individual responding, then that is an error, and the results will be skewed. Since I've taken the survey, I can guarantee that someone botting or using multiple alts will answer much differently than I did for some of the questions. If the survey gets sent to Actual Person 1 at their accounts Main 1 and Alt 1, it is all but assured that they are going to answer the survey in the same way for both accounts. This means that the statistical sample will be weighted in a biased manner, but Dave is not going to know how much to weight the results to account for the selection bias.



Bookmarks