After running a global chisq.test() analysis on the following data, I would like to test for significant differences between pairs.
merged.table
no yes
Central 1735 412
Northeast 419 92
Northwest 419 64
out of state 63 10
South 491 89
> chisq.test(merged.table)
Pearson's Chi-squared test
data: merged.table
X-squared = 12.909, df = 4, p-value = 0.01173
I'm using the chisq.multcomp() function from the RVAidMemoire package but I'm not sure how to interpret the output:
data: merged.table
10 63 64 89 92 412 419 419 491
63 7.5e-10 - - - - - - - -
64 4.8e-10 0.950 - - - - - - -
89 2.9e-15 0.040 0.049 - - - - - -
92 7.0e-16 0.024 0.030 0.862 - - - - -
412 < 2e-16 < 2e-16 < 2e-16 < 2e-16 < 2e-16 - - - -
419 < 2e-16 < 2e-16 < 2e-16 < 2e-16 < 2e-16 0.862 - - -
419 < 2e-16 < 2e-16 < 2e-16 < 2e-16 < 2e-16 0.862 1.000 - -
491 < 2e-16 < 2e-16 < 2e-16 < 2e-16 < 2e-16 0.011 0.021 0.021 -
1735 < 2e-16 < 2e-16 < 2e-16 < 2e-16 < 2e-16 < 2e-16 < 2e-16 < 2e-16 < 2e-16
I'm trying to see which regions (Central, South, NE or NW) differ from one another in terms of proportion who responded "yes"; in other words, if response yes/no is independent of region. Thank you for your help.