You should calculate p value from your data . If p value is or below 0.05 than your variation is significant

Your question is not clear enough.

What are your expected proportions?

If you just have one group of persons, with no expected proportions nor a comparative group I fear you may not be able to make any test (as there is no hypothesis to test).

Dear Marius Baguma

That is what I want to ask. I just a list of 500 persons with their blood groups only. My null hypothesis is that the blood group does not vary. How can test this by using STATA.

Dear Renu Gupta

If I understand your question, I can recommend you use the command tabi.

For simplicity, I will try to answer you with a “real” example.

You can download the Excel spreadsheet “Example” attached below, to have exactly the same thing. I simulated a distribution of different blood groups in a sample of 500 persons using realistic proportions.

1° First, you have to import this dataset in Stata (I guess you know how). I choose to display all variable names in lower case.

For me, the command is:

. import excel "E:\Exemple.xlsx", sheet("Feuil1") firstrow case(lower)

2° Then, you have to calculate the proportions of the different blood groups in your sample, by creating a one-way table using the command tabulate:

. tabulate bloodgroup

3° Now that you have the distribution of different blood groups in your sample (196 20 46 238 for A, AB, B and O, respectively) you may test your hypothesis, which is that the blood group does not vary. So, we can consider that your hypothesis is that each blood group should represent 25 % (so, 125 subjects). It means that you want to see if your distribution (196, 20, 46, 238) is different from the “theoretical” distribution of (125, 125, 125, 125).

P.S. This is not really biologically meaningful. In any population, the proportions of different blood group are very different. What could be meaningful, would be to compare the distribution in your sample to a known distribution from the population or from another group. But as the procedure is the same, we can proceed as described below.

You have to crate that table manually, using the command tabi as below (I chose to show the proportion/percentage of each blood group, and to use the Pearson’s chi-squared test to test the null hypothesis):

. tabi 196 125\20 125\46 125\238 125, chi2 column

Now you have a table showing proportions of different blood groups in your sample, and comparing it to a sample where each group would be equally distributed and represent 25 %.

The p-value here is < 0.0001, so the blood group in not equally distributed in your sample.

Hope this is helpful.

If you prefer working with menus, you can find the tabi command in:

Statistics Menu --> Summaries, tables and tests --> Frequency tables --> Table calculator (see attached picture).

Then you can choose whatever options you want.

You can also use the Stata user guide for more information.

Best wishes.

Dear Marius Baguma

It is not taking the variable as expected.

Normally it should.

Maybe there was a problem while importing the dataset. Can you check if all options were correct. Make sure you checked the option "Import first row as variable names" and you selected "lower" for variable case. (see arrows on the picture attached below).

It it doesn't work, try to use your own dataset, following the same procedure

Dear Marius Baguma

Will you please send me the pic of command that you used. Thanks

Dear Renu Gupta

Please find the command (and steps) attached.