Hello Everyones,
I've seen many posts in your community that seem to answer my questions but I have a tough time figuring out how my situation compares to the one described by the op's and I'm really worried about drawing the wrong parallels... I'm especially ashamed because I studied statistics and worked with SPSS at university, but that was ten years ago and I remember it being a real pool to dip into and I'm having a hard time getting back into it. I've already wasted a lot of time trying to get back into it, pvalue, khi2, my brain is in a loop and I'd appreciate if you could make my life easier by telling me what test to complete so that I can focus on the core of my research...
I take an example similar to mine: we ask multiple people what they have to say about electric scooters. We list each type of remark and count the number of times they recur ( e.g. X people say they save time, Y that they get stolen too easily, etc.).
This same procedure is repeated 2 years later: the responses given are very similar but the proportions that each of them represent are obviously different from the first time. Let's say that here the proportion of those who say that scooters save time is 20% lower than before (we go from 10% of the responses to 8% for example). How can I confirm that this difference is significant and reflects a reality at a very large scale, that of a population as a whole. I have no doubt that going from 2% to 3% on a sample of 10 people and 100'000 people does not have the same degree of significance (one may be related to Hazard and is very prone to different biases while the other may allow to identify the beginning of a trend) how can I understand/reflect my situation there...? Can I, for example, find a way to say that there is an X% chance that the value is 20% (with a tolerance of Y) or to give a "significance" score to this 20% value.
Ps: in the first survey 4049 people gave an answer and 1211 in the second...
thank you!!!
I've seen many posts in your community that seem to answer my questions but I have a tough time figuring out how my situation compares to the one described by the op's and I'm really worried about drawing the wrong parallels... I'm especially ashamed because I studied statistics and worked with SPSS at university, but that was ten years ago and I remember it being a real pool to dip into and I'm having a hard time getting back into it. I've already wasted a lot of time trying to get back into it, pvalue, khi2, my brain is in a loop and I'd appreciate if you could make my life easier by telling me what test to complete so that I can focus on the core of my research...
I take an example similar to mine: we ask multiple people what they have to say about electric scooters. We list each type of remark and count the number of times they recur ( e.g. X people say they save time, Y that they get stolen too easily, etc.).
This same procedure is repeated 2 years later: the responses given are very similar but the proportions that each of them represent are obviously different from the first time. Let's say that here the proportion of those who say that scooters save time is 20% lower than before (we go from 10% of the responses to 8% for example). How can I confirm that this difference is significant and reflects a reality at a very large scale, that of a population as a whole. I have no doubt that going from 2% to 3% on a sample of 10 people and 100'000 people does not have the same degree of significance (one may be related to Hazard and is very prone to different biases while the other may allow to identify the beginning of a trend) how can I understand/reflect my situation there...? Can I, for example, find a way to say that there is an X% chance that the value is 20% (with a tolerance of Y) or to give a "significance" score to this 20% value.
Ps: in the first survey 4049 people gave an answer and 1211 in the second...
thank you!!!
