# Is it correct to use chi-square test here? Quick reply highly appreciated!

#### wellfor

##### New Member
Dear colleagues, very quick and brief advice is highly appreciated! I am about to finish work and have only one hesitation.

Situation:

1) In observational study there are 3 groups (according to diagnosis).
2) All patients were also divided into two groups based on scale score. (Please, see the table).

My question is: Is it correct to use chi-square test to say that between group1 and group2 there is a sighnificant difference when score is above 50, but not between group1 and group3.

Score <50 >51
Group 1 159 127
Group 2 262 10
Group 3 125 57

#### Karabiner

##### TS Contributor
What do you mean by "significant difference" here, difference with respect to what?
And which Chi² test do you mean, Chi² of association? One-sample Chi²?

With kind regards

K.

#### wellfor

##### New Member
Dear Karabiner, many thanks for reply!

To test association between two groups (1 and 3, 1 and 2) and test a hypothesis that - when the score is ">50" there is no difference between 1st and 2nd group.

The main idea is that we can not use that score level (>50) as a tool to distinguish 1st and 2nd group in our sample.

#### Karabiner

##### TS Contributor
test a hypothesis that - when the score is ">50" there is no difference between 1st and 2nd group.
I beg your pardon - difference with respect to what?
When score > 50, then you have n=127 in group 1 and
n=10. Do you want to compare frequencies, relative
frequencies, frequencies retalive to expected frequencies,
or something else?
The main idea is that we can not use that score level (>50) as a tool to distinguish 1st and 2nd group in our sample.
Distinguish with respect to which criterion?

With kind regards

K.