# Data specification confusion

Printable View

• Apr 28th 2011, 04:23 PM
blueirony
Data specification confusion
I have a research project and I am confused about the data specification.

I asked people to complete a survey. The survey had 10 qualities that people look for in their partners. I asked every person to tell me which of the qualities they think is important. I also asked each person to indicate whether they are in a relationship or not, give me an age range (I gave them 3 ranges to choose from) and which country they come from (I gave them 5 countries to choose from).

The question I want to answer is this:
Which factors determine what people value the most when looking for partners? The factors are: relationship status, age and country.

What I don't know is how to set up the variables. I need to eventually use Eviews to interpret the data but before that, I need to set the data up. I have no idea what to do.

'What people value most' is the depedent variable. Relationship, age and country are the independent variables.

Each variable has a number of 'options' each person could choose when completing the survey:
'What people value most' had 10 options.
'Relationship' had 2 (either a yes or no).
'Age' had 3 options.
'Country had 5 options'.

How do I set these up into variables? For each variable, do I just assign each option with a number and then just fill in the data that way? What I'm most confused with is this: what type of variable do I use if I can ONLY have integers? Is that what count data is, therefore do I use a Poisson regression? The way I understand it, Poisson measures the number of times something occurs... which isn't what I am doing here. Because, if say I wanted to find the sample mean and I find that it is 5.6, it doesn't mean anything because each of the qualities is assigned a number and you can't have anything in between them... if that makes sense.

Please help if you can! You have no idea how much I would appreciate it.
• Apr 28th 2011, 05:23 PM
TKHunny
You'll need an awful lot of data for all those buckets. 2*3*5 = 30!
You'll need even more data if you are going to accept whatever 10 items might be mention on any response. There may be no tnearly so much overlap as you want.

I might start with this:

1) Ignore Age and Country. See if there is a significant difference between "Relationship" only.
2) Ignore Relationship and Country. See if there is a significant difference between "Age" only.
3) You get the picture.

Get that done. Let's see your results.