Operationalizations & Alternative Rival Hypotheses

Last week’s scope & methods notes described a political science research design. This week first jumps back by looking at the correct way to operationalizeze variables, and then goes back even further into the literature review, discussinlternativelternative rival hypotheses.

This class is very much a “nuts and bolts” introduction to political science. The last literature review and research design I wrote, for international politics, was written without the benefit of knowing how to do it “right.” Next time, I will know what to do. This class is well worth it.

And now, the boring notes…

Operationalizing Variables

– define a variable in a way that is empirically measurable
– easiest mistake to make
– example: how do you determine democracy-ness?

Whenever you measure, your measure must be
– exhaustive: all categories should sum to 100%
– exclusive: overlap should be 0%

Levels of Measurement (from lowest to highest)
rule of thumb: measure at highest (level you can
– interval better than ordinal better than nominal
– simple categorizations
– just exhaustive and mutually exclusive
– examples: sex, religion, etc.
– categories that can be logically ranked
– examples: education level (less than hs, hs, some college, …. )
– categories with meaningful standard distances between attributes
– also includes Ratios, where there is an absolute zero
– examples: age, heights
– non-ratio example: net income

Single Indicators v. Multiple Indicators
– Single Indicator: measuring income with “income”
– Multiple Indicators: measuring voter participation with voting in federal, state, and local elections

Measurement Error
– the mismatch between the measurement of the concept and the concept
– two kinds of measurement error
— systematic errors:
— errors introduced that are consistent and constant across cases
— example: if questions are difficult to answer, then the survey will be biased toward intelligent people
— relatively easy to find
— random errors:
— “any random error introduced into the model for any other reason”
— “white noise”
— example: someone mistypes a number into a computer

Validity and Reliability
– validity
— you are measuring what you think you are measuring
— systematic measurement errors lead to invalidity
— incorrect operationalizations also cause invalidity
– reliability
— getting the same results consistently
— reproducibility

Alternate Rival Hypothesis

“A hypothesis is an alternative rival hypothesis if it is ‘mutually exclusive’ to the original hypothesis.” (Mannheim and Rich)
– may be better to say “sheds reasonable doubt”

include, or at least, acknowledge, alternative rival hypotheses in the research design

– compare your group to a control group
– make sure the comparison is between similar group (no hidden variables)
– no end to all possible ARHs
– a research design may help discover ARHs
– should control the independent variable of the ARH to prove the hypothesis

Leave a Reply

Your email address will not be published. Required fields are marked *