# Interpretation of McNemar test example

In this tutorial

https://learningstatisticswithr.com/book/chisquare.html

in Example 12.8.1 I understand the result of the test

## McNemar's chi-squared = 12.033, df = 1, p-value = 0.0005226


But they conclude:

And in fact, it looks like the ads had a negative effect: people were less likely to vote AGPP after seeing the ads

Why do they conclude that? How can we know if the effect of the ads (in this case) is positive or negative?

• What are the counts of response_before for yes and response_after for no and yes? Commented Jan 30, 2021 at 12:31
• response_before_yes_response_after_yes=5; response_before_yes_response_after_no=25; response_berfore_no_response_after_yes=5; response_before_no_response_after_no=65; therefore after the ads the response is 5 in both cases, there is no negative effect
– Ana
Commented Jan 30, 2021 at 13:14

Your data should be something like this:

tab = as.table(rbind(c(65,5),c(25,5)))

colnames(tab) = c("no","yes")
rownames(tab) = c("no","yes")

names(dimnames(tab)) <- c("before", "after")

tab
after
before no yes
no  65   5
yes 25   5


And the test goes:

mcnemar.test(tab)

McNemar's Chi-squared test with continuity correction

data:  tab
McNemar's chi-squared = 12.033, df = 1, p-value = 0.0005226


You can interpret it as, before seeing the ad, 30% responded yes:

 rowSums(tab)
no yes
70  30


After seeing the ad, only 10% responded yes:

colSums(tab)
no yes
90  10


Hence it has a negative effect.