Learning SAS? Welcome to the exclusive online community for all SAS learners.

How to start

Reply
New Contributor
Posts: 2

How to start

Hello,

I am going to analyze some surveys of students referred to employer branding. This is my first time using this software and I don't know what devices would bei useful for me and how I can present my results in an easy way.

 

Thanks for your help!

Best regards

Daniel

Super User
Posts: 22,823

Re: How to start

SAS is a tool. First figure out what you want to do in terms of statistics and analytics. Look at reports from survey results and figure out what you have and what you want to achieve. Then we can help get you there, but right now that's a fairly generic topic. 

 

In general, you can search lexjansen.com for survey analysis and you'll find many papers written on the topic to help you get started.

 

If your survey has weights be sure to explore PROC SURVEYFREQ and SURVEYMEANS. In fact, I'd start off working through the base examples for each. 

 

Then I suspect your next challenge is going to be importing your data. 

Here's some video's that cover basic questions such as how to import an excel or csv file and how to create new variables. 

http://video.sas.com/#category/videos/how-to-tutorials

 

 

New Contributor
Posts: 2

Re: How to start

Thanks for your help so far. I will have a look and get some information in the guide.

 

If there is a question it would be kind to keep on offer solutions, also it would be easy for most of you. Next time it try to give more details.

 

Thanks a lot.

 

Have a nice day!

Best regards

 

Daniel

Super User
Posts: 13,008

Re: How to start

The first thing I always do with data is to examine the data. A basic tool is Proc Freq because 1) the syntax is simple for an entire data set and 2) it reports counts and percentages of values.

 

With surveys I look at things like unexpected values for a variable: If a question only supposed to have values of 1,2,3,4,5 and I find other values then there may be an issue with reading the data or the collection instrument actually used doesn't match your expectations. If there is a date of response is it in the expected range? Are respondent ID values duplicated?

 

If you have skip patterns in a survey you may want to check using Proc freq syntax similar to: Tables var1*var2 ; If Var2 should only have responses when Var1=Yes (or similar) you can see quickly if there are responses when Var1 has other answers.

 

What to do if any of the data checks reveals any issue is another more complicated problem.

 

You may also discover that some of your categorical type questions get turned into a host of yes/no response depending on collection software or data entry software behavior. The data checking may tell you that if you want a single response for a category that you need to recode some of the answers. And some of the software if you use multiple categorical responses may have "category x" responses recorded in different variables. Which would indicate that you may want to recode the data so you have a single "category X"=yes variable.

 

 

Ask a Question
Discussion stats
  • 3 replies
  • 378 views
  • 1 like
  • 3 in conversation