Thesis Statistics - - Outcome Known, but want to categorize Participant responses

#1
Thesis Statistics - - Possibly Kruskal Wallis test?

Hi!

I am beginning the process of determining which statistical tests to use for my thesis research involving high school dropouts and their anxiety levels in mathematics and testing situations. Here are my research questions:

1. What are the levels of test and mathematics anxiety among a sample of individuals who dropped out of high school?
2. Does the presence of test and/or mathematics anxiety influence a student’s decision to drop out of high school?

I know this is probably something super simple that I'm panicking about for nothing, but I could really use some help figuring this out, as my thesis advisor ripped apart my proposed analyses earlier today, shredding my confidence to pieces.

Ideally, I would like to take the responses from participants (gathered from two different Likert-scale surveys - one for attitude toward mathematics and one for test anxiety) and categorize them into low anxiety, moderate anxiety, and high anxiety. All participants are high school dropouts -- not a matched samples set-up.

Thank you for your help!!!

Amy B.
 
Last edited:

CB

Super Moderator
#2
Re: Thesis Statistics - - Possibly Kruskal Wallis test?

Hi there, welcome to TS. :Welcome:

Ideally, I would like to take the responses from participants (gathered from two different Likert-scale surveys - one for attitude toward mathematics and one for test anxiety) and categorize them into low anxiety, moderate anxiety, and high anxiety. All participants are high school dropouts -- not a matched samples set-up.
Could you explain why you want to do this? In most cases, collapsing a continuous variable into a discrete number of categories is a bad idea, so it'd be important to see what your reasoning here is.

And is this the only question you wanted help with, or are there other specific issues?
 
#3
Re: Thesis Statistics - - Possibly Kruskal Wallis test?

Hi, and thanks!! In my thesis, I am trying to see if anxiety causes/influences a student's decision to drop out of high school. However, my population are adult learners (18+) who have gone back to school to get their GED's....and yes, this is my only question -- I cannot figure out which test would be best to analyze the data with!

Thanks again!
 

CB

Super Moderator
#4
Hi again,
Ok. It sounds like maths anxiety and test anxiety are your independent variables. But what's your dependent variable?
 
#5
Re: Thesis Statistics - - Possibly Kruskal Wallis test?

I'm pretty sure this is where I would be stuck...I think that my DV is the status as a dropout, but it could also be the level of anxiety experienced (high - moderate - low).
 
#6
Re: Thesis Statistics - - Possibly Kruskal Wallis test?

Here is my proposal draft...if that helps clear anything up for you!
 

CB

Super Moderator
#7
Re: Thesis Statistics - - Possibly Kruskal Wallis test?

I'm pretty sure this is where I would be stuck...I think that my DV is the status as a dropout
But didn't everyone in the sample drop out? It can't be a dependent variable if it doesn't vary...

but it could also be the level of anxiety experienced (high - moderate - low).
From your above comments it sounds like anxiety level is your independent variable ("Does the presence of test and/or mathematics anxiety influence a student’s decision to drop out of high school?")

I guess the question is... how do you intend to see whether anxiety level affects whether someone drops out of high school? (forgetting about the final details of the statistical analysis - how are you actually hoping to investigate this question?)
 
#8
Right. So, what I'm trying to do is see if test/mathematics anxiety influences the decision to drop out (since they are stable across time, the levels we see now should also be what they were [or very close to] during the participants' high school years. There is also a qualitative component where participants are directly asked if they experienced anxiety during high school...So, maybe what I'm doing here is JUST simply assessing anxiety levels?...in which case I have made this much more difficult than it needs to be...
 

CB

Super Moderator
#9
Right. So, what I'm trying to do is see if test/mathematics anxiety influences the decision to drop out (since they are stable across time, the levels we see now should also be what they were [or very close to] during the participants' high school years. There is also a qualitative component where participants are directly asked if they experienced anxiety during high school...So, maybe what I'm doing here is JUST simply assessing anxiety levels?...in which case I have made this much more difficult than it needs to be...
Yeah... it sounds to me like you do not have the appropriate data to do any kind of investigation of the effect of anxiety levels on whether students drop out. Even if you just wanted to look at the correlation between anxiety levels and dropout status you would also need a sample of non-dropouts (and even then correlation does not equal causation). You may have appropriate data just to look at anxiety levels in dropouts, but you may want to discuss this with your supervisor to check your project will still be substantial enough for whatever its purpose is (honours project?)