# What statistic test to use?

I am new to statistics!

I have 2 groups of city population based on education level (group A: educated, group B: without education)

I try to figure out if there is a difference between the number of people who live in the city, based on education

What statistic tests to use to prove the following hypothesis, or what steps to follow?

I already calculated mean and standard deviation for both groups, and I also know N for both, but I am stuck from here, what test to use.

H0: no difference between education level

Ha: more "educated" people than "without education" in the city

Ha2: more "without education" people than "educated" people in the city

Also, is it possible to have 2 alternative hypothesis, or I should think to state the Ha as a single hypothesis? How can I make it into 1 alternative hypothesis?

Thanks!

• You have a sample from your population of interest, and your result is count data? That is you have a count of how many people in your sample are educated/non-educated and live in a given city? Are there multiple cities? Can you show us a sample of your data? Commented Apr 20, 2022 at 7:51
• Yes, I upload the data. A brief explanaition is that for each ID (1,2,3, etc), it represents a 1sq km cell, on which interview of people were taken. The number under "EDUCATED" represents the number of respondents who claimed have education, in that particular 1sq km cell, same for "non educated". Commented Apr 20, 2022 at 8:25
• Are these cells all from the same city? Are these cells of interest to you or do you want to compare only based on all the cells? Commented Apr 20, 2022 at 8:27
• Cell 26 has relatively a lot educated people, cell 6 has relatively a lot uneducated people. This shows that there might be a strong non-homogeneous distribution of the education levels. This means that you need to be careful in creating an aggregated statistic. How did you perform the sampling? This is gonna have an influence on the result. For instance a consideration is a question like: Does cell 26 have more respondents because the population density is larger, or because you sampled more effectively? Commented Apr 20, 2022 at 8:34
• @user2974951 : yes, same city ; @ Sextus Empiricus : the sample is given to me for my homework, ; no text given about population density, but I assume that is the reason. The main task is to either prove one of the hypothesis, but my biggest problem is that they ask me to either prove: more educated in city or more uneducated in city. If it was with only one null and alternative hypothesis, I would have done an independent t test Commented Apr 20, 2022 at 8:46

Regarding the part about the two hypotheses:

Normally you just report the effect that you measured. For instance:

• An estimate for the number of educated people in the city is 60%, this is more than half the population.

The statistical test, e.g. a t-test, is performed in order to find out whether the null hypothesis is correct or not and to characterise the significance of the estimated effect size. The test is used to give an indication of the probability of a false positive result. If the p-value is high, then this means that the measured effect (e.g. that 60%) is not a strong evidence against the null hypothesis (since a deviation of this size could likely happen even when the null hypothesis is true of the p-value is high).

The idea about a one-sided t-test is to adjust the power of the test and this is useful when you are more interested in effects of a particular direction. (btw I write t-test but you might consider a different test, e.g. a test for a binomial proportion)

Usually, the t-test comes into play for this (check its assumptions). Also, have a look at a general procedure.

Welcome to the world of statistics !

As a beginner myself, I always found references on "how to select the good statistical test" useful. This paper by might be a good place to start understanding your data and how to properly analyze the results. You can follow the step by step tree to select the most appropriate method.

Hope it helps !