Rule
Posts: 10479
Joined: 12/5/2005 Status: offline
|
quote:
ORIGINAL: thornhappy What confidence interval are you using, and what's your population size? This poll ain't gonna generate good statistical data. My local profs would shake a stick at you. If not a finger. I have a feeling it's like your previous polls to prove how true submissives are clinically depressed. I have done previous such polls? I believe you are right. I seem to vaguely recall to have done so, quite some time ago now, I think. Where can I find them? (I have given up on the search function.) Shows that depression is one of my continuing interests. I sometimes do have expectations as regards the result of a poll. But I do not set out to prove something when I start a poll. I am merely gathering data. My polls always are objective, not slanted. How the results turn out is up to the honest voters. This is a fuzzy poll. I do not work with confidence intervals. If all choices get equal votes, I conclude that there is no detectable difference between the voters. If three choices get zero votes and the fourth choice gets all votes, I conclude that there is significant difference between them. Those two extremes form the ends of the statistical spectrum for me. What weight I attach to the results also depends on the number of voters. 4, 4, 6, 5 is a slight indication of difference. 41, 44, 63, 51 is a strong indication of significant difference, even though the proportions are nearly identical compared to the first example. Now if this was a statistical test of such simple things as light bulbs, then one might work with confidence intervals of between 2.5 to five per cent, as is customary I seem to recall from long ago. But this is a fuzzy poll. If your profs manage to make sense of all of the fuzziness, then please present their results and confer to them my appreciation. For now I am just curious whether there is no difference, slight difference or strong difference between the choices.
< Message edited by Rule -- 6/1/2010 5:42:08 PM >
|