The Student Room Group

which anova?

Hi

I must only use either a one-way ANOVA, a two-way ANOVA or a two-factor ANOVA in this exercise.


I'm considering using a one-way ANOVA, but I'm still not quite sure...
(edited 8 years ago)
Original post by SikoSiko
Hi

The question is: "Is the average salary different for persons with different categories of education?"

I must only use either a one-way ANOVA, a two-way ANOVA or a two-factor ANOVA in this exercise.
I will attach the data that I have available, maybe you could look at it and tell me which ANOVA is most fitted?
My data is a sample of full-time occupied persons. The sample is considered representative of the working population.

I'm considering using a one-way ANOVA, but I'm still not quite sure...


The obvious question is "why must you use one-way ANOVA, a two-way ANOVA or a two-factor ANOVA in this exercise"? It's pretty obvious that salary will be a function of at least sex age and experience, so the obvious analysis to do is a linear regression adjusting for these possible confouncers. Whether you use salary or log(salary) as outcome depends on the distribution of residuals.
Reply 2
So I should use linear regression instead of ANOVA?
Original post by SikoSiko
So I should use linear regression instead of ANOVA?


How many subjects do you have in your dataset?
Reply 4
Original post by Gregorius
How many subjects do you have in your dataset?


Okay, so what I have understood of the assignment so far is that to answer this question ("Is the average salary different for persons with different categories of education?") I should use the Salary variable and the Schooling-cat (as seen in the attachment) because in the next question sounds as follows: "Are the conclusions from the previous question changed if gender is included in the analysis?"
That's why I thought to first start out with ANOVA....
Original post by SikoSiko
Hi

The question is: "Is the average salary different for persons with different categories of education?"

I must only use either a one-way ANOVA, a two-way ANOVA or a two-factor ANOVA in this exercise.
I will attach the data that I have available, maybe you could look at it and tell me which ANOVA is most fitted?
My data is a sample of full-time occupied persons. The sample is considered representative of the working population.

I'm considering using a one-way ANOVA, but I'm still not quite sure...


What's the difference between "two-way" and "two-factor" ANOVA?

Original post by SikoSiko
So I should use linear regression instead of ANOVA?


Linear regression and ANOVA are the same - they're just different ways of expressing a general linear model.

Given the question, a one-way ANOVA using schooling-cat would answer it. It doesn't ask you to control for anything else (as far as you've said).
Reply 6
Original post by chazwomaq
What's the difference between "two-way" and "two-factor" ANOVA?



Linear regression and ANOVA are the same - they're just different ways of expressing a general linear model.

Given the question, a one-way ANOVA using schooling-cat would answer it. It doesn't ask you to control for anything else (as far as you've said).


All right, thank you! So in the next question where I include gender, it is then sufficient with a two-way ANOVA?
Original post by SikoSiko
Okay, so what I have understood of the assignment so far is that to answer this question ("Is the average salary different for persons with different categories of education?":wink: I should use the Salary variable and the Schooling-cat (as seen in the attachment) because in the next question sounds as follows: "Are the conclusions from the previous question changed if gender is included in the analysis?"
That's why I thought to first start out with ANOVA....


Ah right, I hadn't twigged that you were answering questions on an assignment. Yes, one-way ANOVA for the first is fine and two-way for the second.

However, with data like this, the proper way to do the analysis is via regression using continuous variables as continuous.
Original post by SikoSiko
All right, thank you! So in the next question where I include gender, it is then sufficient with a two-way ANOVA?


Yes.

In general, please give as much information as you can - "this is an assignment", "this is part 1 of a 3 part question", then give all parts etc. etc.
Reply 9
Original post by chazwomaq
Yes.

In general, please give as much information as you can - "this is an assignment", "this is part 1 of a 3 part question", then give all parts etc. etc.


Yes, I will remember that. Thank you! :-)

Quick Reply

Latest