Confused about what test to run for my data ! please help :(

#1
Hi there,

I'm doing my dissertation for uni and my hypothesis is that the 'lack of training affects confidence in performing CPR' and so I conducted a questionnaire which asked "How often does you practice ofter CPR training?" and they had options to answer 'monthly' 'every 3 months' 'every 6 months' 'once a year' or 'less than once a year'. I also had another question which was 'How confident are you at performing CPR?' and it was on a scale of 1 to 5 with 1 representing 'very confident' and 5 representing 'not very confident'. I've attached a screenshot of the excel sheet to show how the data looks, I had 252 responses. I have no idea what sort of stats test to use to show if theres a correlation between the amount of training someone receives with how confident they are at performing CPR. Any ideas would be much appreciated!
 

Attachments

noetsi

Fortran must die
#2
What does your advisor suggest? Generally it is a good idea to ask them what they prefer. Also in the literature what methods do they use.

I am assuming your dependent variable is a 5 point likert scale variable. Ordered (aka ordinal) logistic regression is commonly used to address that although if you are new to statistics that is not an easy place to start.
 
#3
Thanks for your reply :) My supervisor has said to use a chi squared test but I'm not sure how to use it when I'm not comparing two variables? I did a little bit of statistics last year but I really struggled so I'm definitely still a novice. Yes it was a likert scale variable, I haven't used any ordered logistic regression before so I've no idea where to start...
 
#4
Couldn't this be a bivariate correlation? The training variable would be entered as a value between 1 and 5 and the confidence variable would also be entered as a value between 1 and 5. Then you'd predict a positive correlation. As training sessions get farther and farther apart (higher score) confident gets lower and lower (high score).
 
#5
Thanks ebowden! would that be using a Spearman's rank order correlation? I've never performed that test before so I'm not sure how to input the data, which one is my independant variable and which one is my dependant? When you say the training variable would be entered as a value between 1 and 5, would 1 represent 'monthly' and then so and so forth until 5 respresented 'less than once a year'?...I hope I'm on the right track lol!
 

Karabiner

TS Contributor
#6
Perhaps you could read something about Spearman correlation, this could solve the problem with the idea that there's an independent or dependent varible.
 
#7
I would think that a Pearson's would be fine. You are predicting that time between training predicts confidence so training is the independent (predictor) variable and confidence is the dependent (criterion) variable.
 
#8
Regarding coding the data, I suppose you could make an argument for coding the training as 1, 3, 6, 12 representing months between sessions but then I'm not sure what you would do with the more than one year because that could mean anything >12.
 
#10
Thank you ebowden and Karabiner for your responses, that really helps :) I've used the Spearman test and it shows there's a weak correlation so thats my hypothesis out the window lol! but at least I've something to discuss now, thank you!