# Statistically significant changes in proportions - comparing 3 populations

#### sev1g14

##### New Member
I have data from three populations, 2 observations each populations (opinion poll data from different countries). I am trying to find out if any changes in opinion on the issue i am investigating are statistically significant.

So, for one topic

In 2010, Group 1 answered 20% yes, Group 2 answered 16% yes, and group three answered 46% yes

In 2016, Group 1 answered 53% yes, Group 2 answered 61% yes and Group 3 answered 47%

I therefore know that Group 1 increased by 33, Group 2 by 45 and Group 3 by 1.

So I want to know if the changes in opinion between the groups are statistically significant but do not know what test to use. How do I test for statistically significant change between three populations?

#### sev1g14

##### New Member
Yes the first observations were taken from a sample of 1000 in each country, and then the second observations were taken from a sample of 2000 in each country.

#### katxt

##### Active Member
The simplest thing to do is a test for proportions on each population separately. You can find a calculator on the net for the difference in two proportions.
Here is the first one Google found - https://www.medcalc.org/calc/comparison_of_proportions.php
It uses the chi squared test, but there are others which do it a different way with the same results. With three tests you should perhaps use a more stringent significance level than <0.05 but this situation is clear cut.