# What statistical tools should I use?

#### NiRen

##### New Member
Hi, I need help for my thesis. I consulted with 2 statisticians and they gave me different results. So, basically, my problem is to know if independent factors (ex. sex, age, educational attainment, type of employment (direct/agency), years of teaching experience, etc.) influence the salary of teachers. The data collected are in ranges (salary = 15,000 - 20000; 20,001 - 25,000; etc). I want to know if females have higher salaries than males, and so forth, so what statistical tools should be used?

#### Karabiner

##### TS Contributor
How large is your sample size? And, do you want to analyse each factor separately,
or all factors jointly in 1 model?

With kind regards

Karabiner

#### NiRen

##### New Member
Hi Karabiner,

Sample size is only 143 and yes, each factor will be analyzed separately.

#### Karabiner

##### TS Contributor
The scale of your dependent variable "salary ranges" is ordinal. If you want to analyse its
association with an interval scaled variable (such as age, or years of teaching experience), or
with another ordinal scaled variable (such as maybe educational attainment), then you can
use the Spearman rank correlation coefficient rho. If you want to analyse the association with
a binary variable (such as sex, or type of employment), then you can use the Mann-Whitney U-test.
If you have a categorical variable with more than 2 categories, then you can use the Kruskal-Wallis

With kind regards

Karabiner

#### NiRen

##### New Member
The scale of your dependent variable "salary ranges" is ordinal. If you want to analyse its
association with an interval scaled variable (such as age, or years of teaching experience), or
with another ordinal scaled variable (such as maybe educational attainment), then you can
use the Spearman rank correlation coefficient rho. If you want to analyse the association with
a binary variable (such as sex, or type of employment), then you can use the Mann-Whitney U-test.
If you have a categorical variable with more than 2 categories, then you can use the Kruskal-Wallis

With kind regards

Karabiner
Thank you for explaining and simplifying, I now know which one to follow from the two results that I have. I greatly appreciate your help.