Calcite | Level 5

## TEST OF STATISTICAL SIGNIFICANCE

Hello,

Please, when doing a test of statistical significance judging based on the p-value and intervals, is it possible to get difference results. For example, is it possible that judging from the p-values, the independent variables are not statistically significant but judging from the intervals they are statistically significant?

Thanks a lot.

Regards,

lizzy100

2 REPLIES 2
Opal | Level 21

## Re: TEST OF STATISTICAL SIGNIFICANCE

Yes it is possible that p-values and confidence intervals sit on opposite sides of a of a given significance threshold (such as 0.05). But it is quite rare. In any case, there isn't that much qualitative difference between p=0.049 and p=0.051.  P-values are random quantities, after all.

PG
Diamond | Level 26

## Re: TEST OF STATISTICAL SIGNIFICANCE

@lizzy100 wrote:

Hello,

Please, when doing a test of statistical significance judging based on the p-value and intervals, is it possible to get difference results. For example, is it possible that judging from the p-values, the independent variables are not statistically significant but judging from the intervals they are statistically significant?

I'd prefer not to make any meaningful comment on your results without seeing the results. Please show us the actual SAS output so we can see what you are seeing. I don't even understand the answer from @PGStats without seeing an example.

It is my understanding that (except for round-off error), the confidence interval and the statistical test ought to produce equivalent answers.

--
Paige Miller
Discussion stats
• 2 replies
• 428 views
• 0 likes
• 3 in conversation