topic Re: Regression in SAS Programming

Regression

Reader587 — Sat, 08 Jan 2022 16:54:28 GMT

How do I change Beta=0 to Beta=2 for proc reg?

Re: Regression

PaigeMiller — Sat, 08 Jan 2022 16:59:16 GMT

Please explain further. Provide more detail. Provide example data.

Re: Regression

Reader587 — Sat, 08 Jan 2022 17:27:15 GMT

data plastic2;
input hardness minutes;
cards;
199.0 16.0
205.0 16.0
196.0 16.0
200.0 16.0
218.0 24.0
220.0 24.0
215.0 24.0
223.0 24.0
237.0 32.0
234.0 32.0
235.0 32.0
230.0 32.0
250.0 40.0
248.0 40.0
253.0 40.0
246.0 40.0
;run;
proc reg;
model hardness=minutes/clb alpha=.01;run;
I am trying to change the test from Ho:B1=0 vs Ha:B1^=0 to Ho:B1=2 vs. Ha:B1^=2

Re: Regression

PaigeMiller — Sat, 08 Jan 2022 17:32:49 GMT

Use the TEST statement in PROC REG. Examples: https://documentation.sas.com/doc/en/pgmmvacdc/9.4/statug/statug_reg_syntax20.htm

Re: Regression

Reader587 — Sat, 08 Jan 2022 17:39:51 GMT

What if I want to do a one-sided t-test for regression?

Re: Regression

PaigeMiller — Sat, 08 Jan 2022 17:49:10 GMT

The PROC REG output contains two-sided tests. If you were to do a two-sided test at the alpha=0.05 level, then 0.025 chance of a type I error is in the upper tail and 0.025 chance of a type I error is in the lower tail.

You want a one-sided test with alpha=0.05 in one tail and alpha=0 in the other tail. So, if the pvalue from SAS is <0.10, your one-sided test is statistically significant at the alpha=0.05 level, because 0.05 chance of type I error in one tail, and 0 percent chance of type I error in the other tail.