Significant difference between two metric variables

[Nanna Brandt]

We have made a pre-test and asked respondents to evaluate the direction and strength of 7 arguments (on 7-point scales ranging from definitely opposed to definitely supportive, and from definitely not effective to definitely effective). Now we want to test if there is significant different between strength and direction of the different arguments. We ere thinking about making a ttest (like example 4 in this Stata-paper here: https://www.stata.com/manuals13/rtte...pArmczP2RHuN20 ) - but we're learned that ttest can only run one ordinal and one metric variable, and both our variables are metric.

How do you test if there is significant difference between two metric variables?

Thankyou!

[Carlo Lazzaro]

Nanna:
welcome to this forum.
I do not think that -ttest- would be the way to go.
You may want to consider Example 1, -pca- entry, Stata .pdf manual, pages 608-609.

[Nanna Brandt]

Thank you very much for replying on this. How do I find the example and the pages, you are referring to?

[ericmelse]

You use help:

h pca

and when the viewer window is displayed, mouse click on the line below the title (pca -- Principal component analysis):

(View complete PDF manual entry)

When your system provides access to the PDF file, you can page through to pages 608-609 for Example 1 as Carlo suggested.

[Marcos Almeida]

If you just wish to check the correlation between each pair, you could use the nonparametric - spearman - for that matter.

[Rich Goldstein]

I am a bit confused - you can certainly use a t-test with two metric variables - see the third syntax in the help file:

help ttest

[Carlo Lazzaro]

Rich:
my doubt was that the original poster seemed to deal with ordinal, not continuous variables.
Obviously, I could have well misunderstood something.