Combining two variables

[Hanna Hi]

Hello,
i'm just working on a do-file and I'm struggeling to combine to variables.
The first variable is asking for support for democracy (0 no support to 10 strong support).
Than I have a second variable asking if someone sees free elections as a criteria for democracy (0 strongly disagree to 10 agree strongly).

Now I want to combine this two variables, if possible the scale should stay from 0 to 10.
I had in mind to add ore multiply but than I wasn't sure how to get the 0 to 10 scale back.
Does someone has an idea?

Thanks a lot in advance!
Hanna

[Sam Bufe]

If you don't care about weighting the variables at all, couldn't you just multiply the two and then divide by 10?

[Hanna Hi]

Thanks a lot for your advice!

I'm a absolute stata beginner, so sorry if the answer was obvious.
I got two following questions:
1) What do you mean about the weighting? Is that normally necessary?
2) My generated variable is from 0 to 10 now, but with values like 6,3/ 7,2 and so on... Is it possible only to have whole numbers? And if not does that negatively effect my regression analysis?

[Sam Bufe]

Sometimes you would weight responses if you're using responses to a validated scale or something.

But don't put that variable you just created in a regression. I have no idea how you would interpret a one-unit change in this variable. For analysis purposes, I imagine you'd be better off making a few dummy variables and maybe interacting them.

[Hanna Hi]

The newly created variable would be my dependent variable, would that be ok?

[Nick Cox]

You lost me at the beginning with wanting to combine the two variables. Why?

Multiplying the two variables together is totally arbitrary and just a recipe for losing information.

[Hanna Hi]

I want to combine the variables because, there might be people seeing free elections not as a criteria for democracy.
If that is the case they support something what isn't actually democracy.
You know what I mean?
I'm also not sure just multiplying the two variables is what I want but I don't really have a better idea.

[David Benson]

But then it sounds like what you want is to regress one on the other, or to see the correlation between the two variables. That will tell you if people who claim strong support for democracy also claim strong support for free elections. And if they don't, presumably you have other variables to help you determine what they mean by democracy if not free elections (i.e. the ability to criticize the govt without fear of retribution? Freedom of the press? Rule of law? Protecting the rights of a minority group? An orderly transition between govt leaders, etc).

Alternatively, you could treat them as two separate DV's and see if what predicts one also predicts the other.