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
- /
- weighted data vs unweighted data

- 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

05-11-2016 02:02 PM

Hello to everyone

What does weighted data and unweighted data mean? What are the importance of weighted data?

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

05-11-2016 09:47 PM

weight will change the influence of obs in the model .

More detail information , Check Rick's blog :

http://blogs.sas.com/content/iml/2013/09/13/frequencies-vs-weights-in-regression.html

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

05-16-2016 01:47 PM

Thanks **Ksharp**.

If the rate parameter is estimated using both weighted and unweighted and with unweighted the rate parameter is 0.00128 ± 0.00011 (CI) and 0.00137 ± 0.00018 (CI) with weighted.

So first what does it mean such change in rate parameters? Which one should I use (rate parameter obtained with weighted or unweighted data)??

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

05-16-2016 08:52 PM - edited 05-17-2016 01:53 AM

I think you mixed up parameter estimator and weight of obs . It depends on how much important for a special obs in model and what do you want. by default , very obs has weight 1, if you think some one of obs is very important for the model, you can increase its weight, vice verse.

Did you check Rick's blog ? weight would not change other statistical like DF, STD ERROR . therefore it would not change parameter estimator.

As Rick's blog showed, if you want robust parameter estimator(there are some outliers in model) ,you should use weight of obs.

otherwise unweight.

Maybe you should ask Rick , who might give you more information.

P.S. I would choose 0.00128 ± 0.00011 (CI) , since STD ERROR is smaller than another.