# How do I check for normality?

#### statdent

##### New Member
Hello everyone,

I need to compare the test scores between males and females. but, before that, I need to check for normality of the data. Now, Should I check for normality separately for males and females? or together? Also, Should I select only the corrects answers while checking for normality? For example, in my excel, subject no. 1 has given correct answer for 1st question and wrong answer for 2nd question. Should I filter the correct answers and then check for normality since only correct answers carry scores? Thanks in advance.

#### Karabiner

##### TS Contributor
Could you please describe the study first (research question, sample size, independent and dependent variables and how they are measured)?

With kind regards

Karabiner

#### Editor

##### New Member
You need to check for normality for variable included in the analysis and in your case it is test score

#### statdent

##### New Member
Could you please describe the study first (research question, sample size, independent and dependent variables and how they are measured)?

With kind regards

Karabiner
The research question is whether there is a difference in KAP between males and females. The sample size is around 270. I used a questionnaire to assess KAP. Correct answers for knowledge, positive attitude, and correct practice are given 1 mark each. Since there is no intervention here, would these answers be out come/ independent variables?

#### statdent

##### New Member
You need to check for normality for variable included in the analysis and in your case it is test score
True, but I was wondering whether the tests scores should be checked separately for males and females and whether only right answers should be included?

#### Karabiner

##### TS Contributor
For an independent samples t-test the normality is assessed for each of the subsamples separately,
but normality of the two populations is only to be assumed if total sample size is small. With n=270,
sample size is definitely large enough so that you do not need to care about normality. Therefeore,
you could compare the mean number of correct answers between groups using the t-test (with
Welch correction).

With kind regards

Karabiner

#### statdent

##### New Member
For an independent samples t-test the normality is assessed for each of the subsamples separately,
but normality of the two populations is only to be assumed if total sample size is small. With n=270,
sample size is definitely large enough so that you do not need to care about normality. Therefeore,
you could compare the mean number of correct answers between groups using the t-test (with
Welch correction).

With kind regards

Karabiner
Thank you so very much! May I ask what the Welch correction is for?

Also, how do I enter the data into Excel? Should I assign 1 to the correct answers and 0 to the wrong answers?

#### Karabiner

##### TS Contributor
I do not use Excel for statistical analyses, so I cannot tell you how you should structure and enter your data.
The Welch correction is for unequal variances between groups (the t-test assumes equal variances). One can
(and should) use that correction by default.

With kind regards

Karabiner

#### statdent

##### New Member
Okay. Thank you for the explanation. Are you familiar with PSPP or Jamovi?

#### jamesmartinn

##### Member
I'm not familiar with either, but will admit I was curious so I looked them up. It seems Jamovi is built on R, and you are able to execute R code.

Here is a guide that show's how the Welch-correction version of the t-test can be ran in R (one command assuming your data is imported): https://www.statology.org/welch-t-test-in-r/

Hopefully you're able to adapt it in Jamovi (or verify Welch is used if an existing t-test procedure is used). Or, given that your analysis is simple enough, just download R and give that a whirl directly.

#### statdent

##### New Member
I'm not familiar with either, but will admit I was curious so I looked them up. It seems Jamovi is built on R, and you are able to execute R code.

Here is a guide that show's how the Welch-correction version of the t-test can be ran in R (one command assuming your data is imported): https://www.statology.org/welch-t-test-in-r/

Hopefully you're able to adapt it in Jamovi (or verify Welch is used if an existing t-test procedure is used). Or, given that your analysis is simple enough, just download R and give that a whirl directly.
Thank you so much for your help. I have heard about R several times but never got to try it. I will check out the link. Do you happen to know any websites that have data sets, preferably health data, to practice each of the statistical tests with?

#### jamesmartinn

##### Member
Thank you so much for your help. I have heard about R several times but never got to try it. I will check out the link. Do you happen to know any websites that have data sets, preferably health data, to practice each of the statistical tests with?
No problem.

The example in the link that demonstrates the Welch T-Test starts from scratch (i.e. sample raw data are first created, then used, step-by-step).

Another good resource is here:

https://stats.oarc.ucla.edu/stat/data/intro_r/intro_r_interactive_flat.html

This should get you started on working with R. It even has another from scratch data analysis example using the Welch t-test:
https://stats.oarc.ucla.edu/stat/da...ractive_flat.html#independent-samples-t-tests

Here is the link to how you can create the sample data that's used:
https://stats.oarc.ucla.edu/stat/da...scriptive-statistics-for-continuous-variables

That dataset looks like it's used in many examples on that page, so it's a great choice to start playing around.