topic Re: Bivariate analyses in SAS Programming

Bivariate analyses

stephanicurry — Mon, 13 Jul 2020 15:02:22 GMT

Best way to do a bivariate analysis of one categorical variable (two categories) and one continuous?? One sample.

Looking to compare means, standard deviations, and p-values.

Would appreciate which proc procedure to use.

Thanks in advance!!

Re: Bivariate analyses

PaigeMiller — Mon, 13 Jul 2020 14:59:50 GMT

@stephanicurry wrote:

Best way to do a bivariate analysis of one categorical variable and one continuous??

Looking to compare means, standard deviations, and p-values.

Comparing means sounds like Analysis of Variance, where you compare the mean of the continuous variable across the categories of the categorical variable. There are assumptions I am making. But I'd rather have you explain in a lot more detail what you want, instead of me guessing what you want.

Re: Bivariate analyses

stephanicurry — Mon, 13 Jul 2020 15:01:39 GMT

The categorical variable only has two categories so I was thinking not ANOVA

Re: Bivariate analyses

PaigeMiller — Mon, 13 Jul 2020 15:08:23 GMT

What analysis were you thinking of?

Re: Bivariate analyses

stephanicurry — Mon, 13 Jul 2020 15:19:15 GMT

Maybe a one sample t-test? Not 100% sure

Re: Bivariate analyses

Reeza — Mon, 13 Jul 2020 15:20:28 GMT

You need a hypothesis first usually.

Here's examples of common statistical analysis options:
https://stats.idre.ucla.edu/sas/whatstat/what-statistical-analysis-should-i-usestatistical-analyses-using-sas/

Re: Bivariate analyses

PaigeMiller — Mon, 13 Jul 2020 15:22:17 GMT

Here's a t-test example using PROC TTEST:

https://documentation.sas.com/?cdcId=pgmsascdc&cdcVersion=9.4_3.4&docsetId=statug&docsetTarget=statug_ttest_examples01.htm&locale=en

Re: Bivariate analyses

stephanicurry — Mon, 13 Jul 2020 15:23:27 GMT

Thank you!

Can I use ANOVA with only two categories? and if I do, would I use proc glm?

Re: Bivariate analyses

PaigeMiller — Mon, 13 Jul 2020 15:27:47 GMT

ANOVA with two categories should give the exact same answer as a t-test. You can use PROC GLM.

Re: Bivariate analyses

stephanicurry — Mon, 13 Jul 2020 20:15:42 GMT

Thank you!

Re: Bivariate analyses

Ksharp — Tue, 14 Jul 2020 12:23:41 GMT

T test PROC TTEST
or Non-parameter version of t-test
Wilconx test by PROC N1PARAM .