# Thread: Which test to use?

1. ## Which test to use?

I have just learned that I chose the wrong test for my data and I am now at a total loss.

Basically I sent out a questionnaire which had a likert type scale measuring attitudes towards safety. There were 64 statements in all.

Respondents were grouped in to 5 categories depending on length of time employed in the industry.

I want to look at whether tenure can predict attitude towards safety.

I figure my DV is tenure and my IV's are the attitude scale scores.

I just cant fathom out which test I should be using. I looked at a 2 way ANOVA but then read on the net you couldn't use this for categorical data so I dismissed it. I know correlation and multiple regression arent right but now I am just totally lost and cant see the wood for the trees.

If anyone can help I would really appreciate it. I dont expect answers handed to me on a plate but I have been struggling with this for over two months and its making me ill. Im just not a stats person.

Thanks!

3. Originally Posted by 2461c
I want to look at whether tenure can predict attitude towards safety.

I figure my DV is tenure and my IV's are the attitude scale scores.
Other way round!

Now, what kind of data comprise your variables?
Is "attitude towards safety" a sum of the 64 Likert items, or just one of the Likert items in specific? How many response options do the Likert items have?
Do you have access to the *actual* length of time employed in the industry, or only the groupings? The actual length of time would probably be a better variable to work with. If you only have the groupings, what are the groupings? Like 0-5 years, 5-10, etc?

If your attitude towards safety variable is a sum of 64 Likert items I see no reason why you couldn't use an ANOVA (although I don't see where your second IV is for a 2-way ANOVA; it sounds more like you're just looking at a one-way ANOVA). ANOVAs are used when you have a continuous DV and categorical independent variable(s). The sum of 64 Likert items isn't technically continuous data, but with that many items you can certainly justify assuming continuous data on the scale level (i.e. a sum across the items), even if assuming continuous data on the item level might be more problematic, depending on how many points there are on your Likert response scale.

Oh, and what's your sample size?

4. Thank you for replying

Sample size is 155.

The 64 items are all related to safety - hope that makes sense?

I will have a look at ANOVA - thanks ever so much for replying

 Tweet

#### Posting Permissions

• You may not post new threads
• You may not post replies
• You may not post attachments
• You may not edit your posts