Develop a rough and ready theory of party identification. Taking the case of Bulgaria
- Choose variables that could reasonably be measured (though a survey or some other means).
- Consider carefully issues of both conceptual and operational definition. For example, if you posit religion as an independent variable, do you mean that there will be partisan differences between people of different faiths, or do you mean that people who are more religious (you also need to consider what that means and how you might measure it) will differ from those who are less so.
- Specify the nature of the relationship. Do not merely say that party identification is related to age. Are older people more likely than younger people to be Liberal, or are they more likely to be Socialists for instance?
- Provide rationales for your hypotheses. (“Senior citizens are more likely than younger people to be Socialists because they are more worried about social equality” “Senior citizens are more likely than younger people to be Socialists because they are more worried about Social Security.”)
- Don’t worry at this point about whether your hypothesis is valid.