Cross tables are used to produce differences between two groups in the chi-square test. For example, if we look at the differences in age groups between the two groups (such as the group A and the group B), and the overall p-values for the under-20, 21-30, 31-40 and 41-50 age groups are less than 0.05, can we use dummy variables to produce p-values for each age group?
5
-
2$\begingroup$ The p value is associated with a test about differences in distribution between two or more groups. What test would one p value per group be associated with? $\endgroup$– Stephan KolassaCommented Jan 22 at 13:11
-
$\begingroup$ Thank you for your message. I initially use cross-table to find out statisitical difference of the age distribution between the group A and the group B. I found out a significant difference between the groups. I then tried to find out the diffirence in each p-value using dammy variables (0 and 1) in age distributions with chi-square test. My question here is whether chi-square test can be used to find out the statistical difference in each categorical variable under these situation. $\endgroup$– 橋本悟Commented Jan 22 at 23:29
-
$\begingroup$ But you will still need to always compare two groups, because you can't find "a difference in only one group". Which would be like the sound of one hand clapping... Also, since the chi squared test looks for differences in group composition, you can't look at just one category. The chi squared test essentially looks for situations where group A has a larger proportion of X than group B, but then B must have a smaller proportion of some Y than A. $\endgroup$– Stephan KolassaCommented Jan 23 at 6:58
-
$\begingroup$ Sorry for my poor explanation. We have two group, say the group A and the group B. We would like to compare that the statistical difference (p-value) between these groups in term of the following age distribution at first. 1. less than 20, 2. 21-30 years, 3. 31-40 years, 4. 41-50 years, 5. more than 50 years. We found a significant difference among the age distribution in the group A and B using Chi-square test. We then tried to find p-values of each category by creating dammy variables in chi-square test. Is this acceptable way to find p-values of each category) using dammy variables? $\endgroup$– 橋本悟Commented Jan 23 at 8:18
-
$\begingroup$ I think I understood your point that we cannot look at just one category. Thank you very much for your help. $\endgroup$– 橋本悟Commented Jan 23 at 8:32
Add a comment
|