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
- /
- SAS Programming
- /
- SAS Procedures
- /
- Regression analysis

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

11-02-2012 03:33 PM

Hi Team,

I have something like shown below,'

Type of Case

CaseA CaseB CaseC

Diabetes N= N= N=

Y

N

AIDS

Y

N

Headache

Y

N

Fever

Y

N

I did a contingency table of :

Diabetes with Cases

Aids with Cases

Headache with Cases

Fever with Cases

Found that the p-values are significant for a couple of them(less than 0.05)

What is the meaning of that statistically?

Also how to do a regression analysis....How do I proceed from there???

Any help is highly appreciated

Regards

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

11-02-2012 03:56 PM

Whenever a p value is less than 0.05, this indicates that the quantity being tested is statistically significant (at the alpha=0.05 level). In layman's terms, this particular arrangement of data that is tested was unlikely to have happened by random chance, it is probably a "real" effect.

You cannot do regression on this data. You need to have continuous X and continuous Y to do a regression.

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

11-02-2012 04:04 PM

Thanks for the reply.

Could you also provide me with the basics for understanding Logistic Regression ??

Thanks

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

11-02-2012 04:12 PM

Logistic regression is appropriate when you have a continuous X variable, and a categorical Y variable.

For example, if X is temperature, and Y is pass/fail status.

Among other things, logistic regression will compute the probability that you get a pass (or a fail) at a given temperature.

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

11-02-2012 04:22 PM

You've completed what's known as univariate analysis on your data, checking the category of one variable against others one at a time.

It sounds like you're looking to move on to the multivariate stage, looking at multiple variables together, and relationships between them.

The appropriate multivariate model depends on what your research question/hypothesis is and the structure of your data.

As you've shown all categorical data, my guess is a log-linear model, but that's a GUESS.

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

11-02-2012 04:29 PM

Hi,

I am trying to study the **problems associated** with the** type of surgery done.**

**when i use a particular type of surgery what effect/problem it is creating???**

**Regards**

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

11-02-2012 04:49 PM

So your outcome is an indicator if a patient has a problem?

For me at least I'd need more of an explanation of your data and what you were looking to answer.

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

11-02-2012 05:39 PM

I'd say consult with a statistician (possibly a biostatistician) now .There really are too many things that would need to be considered and I've seen too many bad medical publications to help produce one.

Your local university probably has a consulting services that will be free of charge or low cost.

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

11-04-2012 09:21 PM

P value less than 0.05 is statistical significant if your possibility of committing the first mistake is .05 .

That means your H0 hypothesis (Type of Case and Headache are independent ) is not right.

.05 discover there are some correlation between Type of Case and Headache .

I recommend you also use corresponding analysis to check this relation more .

Ksharp