Method
Relationships Between Variables
In scholarly work, we often set out to show that there is a relationship between X and Y for one or more cases.
- We may want to show that when X is false/true, Y is false/true;
- when X is 0 or 1, Y is 0 or 1;
- when X increases, Y increases –
these would all be examples of relationships between an independent and a dependent variable.
The case of no relationship would be X changes but Y remains constant.
| Y | X = 0 | X = 1 |
|---|---|---|
| Y = 0 | Case | |
| Y = 1 | Case | |
| X | ||
The case of no relationship would be X changes but Y remains constant.
A research design would require a minimum of two cases, and can have many more. We would need a case of X=0 and of X=1, or X low and of X high. For each of these, we will set out to find out the values of Y which we expect to be related to X.
In a case-study design, we have information on X and Y from two or more cases.
In a large-N design, we have many more – say, more than 50.
A medium-N design would be something in-between.
In a large-N design, we may take advantage of statistical techniques, to demonstrate that X and Y are related.
A case could be a country,
- if you wanted to know why different countries go to war, or run a budget deficit.
A case could be individuals,
- in surveys, we ask people how they feel about things, positing a relationship between say a person’s sense of job security and that person’s tendency to vote left or right.
Data
You can simply read to infer the values of your data
- books, government docs, and newspaper articles. You can simply cite those and declare that X and Y for a particular case have certain specific values.
You can also generate your data by asking people things
- by running your own survey.
You can download existing data
- usually in the form of something like a spreadsheet.
Bottomline
All research designs are:
- comparative in nature,
- all inference from your data is based on empirical variation between cases.
Whether you have two cases, or thousands of cases
- you are comparing how the dependent variable varies as your independent variable changes.
You can attach many other descriptions to your method:
- most-similar or most-different design, interviews, observational or experimental, over time, panel, etc.,
- you can find some methods more or less compelling.
The above statements will continue to describe the basic logic of all research.