Proposing Explanations, Framing Hypotheses, and Making Comparisons

Harriet Goers

What We’ve Learned So Far

Variables are measurable traits derived from concepts
Types:
- Nominal (e.g., party ID)
- Ordinal (e.g., trust levels)
- Interval (e.g., vote share)

We describe them using central tendency and dispersion.

Today’s Big Question

Why do political behaviors and attitudes vary?

We’re moving from what to why.

To answer why, we need causal explanations.

Building a Causal Explanation

Two core ingredients:

Dependent variable (DV): What we’re trying to explain
Independent variable (IV): What we think causes the change

From Theory to Hypothesis

Theory: Education increases turnout by building awareness and civic confidence.

→ This becomes:

Hypothesis:
In a comparison of individuals, those with more education will be more likely to vote than those with less.

What Makes a Hypothesis Good?

✅ Identifies IV and DV
✅ Makes a clear comparison
✅ Describes direction of relationship
✅ Is testable

Weak vs. Strong Hypotheses

Weak:

Some people are more likely to donate to campaigns.

❌ No causal variable
❌ Not testable

Strong:

In a comparison of individuals, those who attend church weekly will be more likely to donate to campaigns than those who do not.

✅ Clear causal story
✅ Ready to test

Avoid Common Pitfalls

❌ Tautologies: Circular reasoning
❌ Vagueness: Undefined terms
❌ Overload: Too many variables at once

Stick to one cause and one effect at a time.

Testing Hypotheses: Compare Groups

Hypothesis:
Democrats are more likely than Republicans to support increased Social Security spending.

We compare group means or percentages.

Two Main Testing Methods

Cross-tabulation
- When both IV and DV are categorical (nominal/ordinal)
Mean comparison
- When IV is categorical, DV is interval

Cross-tab Example (Simulated Data)

Mean Comparison (Simulated Data)

Describing Relationships

Linear: Consistent effect across values

Nonlinear: Effect weakens, reverses, or curves

Example: Diminishing Returns

Thinking in Causal Chains

Education → Political efficacy → Voting

Each link can be tested as a separate hypothesis.

This is where causal mechanisms come in.

Probabilistic Thinking

Political science is not deterministic.

We say:

“More education tends to increase turnout.”

We don’t say:

“Everyone with a PhD always votes.”

Framework for Causal Research

Pick a dependent variable
Identify a causal variable (IV)
Write a hypothesis
Think through the mechanism
Choose a testing method
Look for linear or nonlinear patterns