It is clear that the chi-square test for a 2D array of data does not assume a uniform distribution. Is the chi-square test (for a single variable) valid for a non-uniform expected distribution? E.g. for a cohort of members all of whom joined on a specific date, I have data for the number who dropped out each week over the subsequent weeks - to simplify things until there are no cohort members left. My hypothesis is that this exponential decay (I am looking for deviations form it - e.g. that the drop out rate peaks in the third week). I can do a regression to get the time constant, can I use this to build the "expected" data and chi-square test against it? And then, if the p-value if smaller than something, so I can reject the null hypothesis, look for elements of the chi-square array which are large?
Sorry for my lack of understanding but if I may ask, say I have multiple questions in the survey I'll have to conduct a Chi square on every single one of them?
The Chi Square checks for a relationship between 2 variables. For example, Is “gender” related to “screen time”. Conduct a chi square test for each pair of variables you’re analyzing. Hope this helps!
I also have a video for the Chi Square Test of Independence (with a contingency table) ua-cam.com/video/KW1lR31XpTQ/v-deo.html. Thanks for watching!!
Thank you. very informative steps.
You are so welcome!
It is clear that the chi-square test for a 2D array of data does not assume a uniform distribution. Is the chi-square test (for a single variable) valid for a non-uniform expected distribution? E.g. for a cohort of members all of whom joined on a specific date, I have data for the number who dropped out each week over the subsequent weeks - to simplify things until there are no cohort members left. My hypothesis is that this exponential decay (I am looking for deviations form it - e.g. that the drop out rate peaks in the third week). I can do a regression to get the time constant, can I use this to build the "expected" data and chi-square test against it? And then, if the p-value if smaller than something, so I can reject the null hypothesis, look for elements of the chi-square array which are large?
That's a great question. This video from Lis Wilson may help ua-cam.com/video/Xi6WUaEYj0Y/v-deo.htmlsi=nh83Y1zHaMb4pnvn
Sorry for my lack of understanding but if I may ask, say I have multiple questions in the survey I'll have to conduct a Chi square on every single one of them?
The Chi Square checks for a relationship between 2 variables. For example, Is “gender” related to “screen time”. Conduct a chi square test for each pair of variables you’re analyzing. Hope this helps!
Thank you.
You're welcome!
Thanks!
You are very welcome, Facundo!