turn on suggestions

Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

Find a Community

- Home
- /
- Analytics
- /
- Stat Procs
- /
- How to reset alpha and beta values

Topic Options

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

02-18-2011 09:31 AM

I have a project where I need to redefine the level for type I and type II errors (alpha and beta). I want to set alpha to 0.1 and beta to 0.85. Any help is appreciated.

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

02-18-2011 12:07 PM

I think you are mixing up beta and power (1-beta). What kind of testing are you doing? You can always set alpha to whatever you want. Then for that alpha you need to find a sample size that will give you the correct beta/power. How to calculate sample size will depend on what you are trying to do.

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

02-18-2011 01:29 PM

Here's the problem, perhaps in better detail. This study already had a sample size calculated in the study design, from another program, before the data was ever collected. Now I need to do a Wilcoxon signed rank test for 40 subjects, with alpha 0.05 and b (power) 0.85, which are the assumptions from the sample size document. So far what I have figured out is that I can use PROC UNIVARIATE, and ALPHA =, but I can't find how to redefine beta. Everything that I have read which discusses "power" is about sample size and power calculations, not about simply assigning new value for beta.

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Highlight
- Email to a Friend
- Report Inappropriate Content

02-18-2011 01:40 PM

If the data is already collected then you have no control over the power unless you are allowed to collect more data. You don't assign a power when you are doing a Wilcoxon signed rank test (or any test really). You can only influence the power by changing sample size or alpha before you collect your data. That is why all you found about power was related to sample size.