Unit 2: Research Methods

Merton’s Norms of Science

1. Organized Skepticism

2. Universalism

3. Communitarianism

4. Disinterestedness

1. Organized Skepticism

Question findings until the evidence is overwhelming

• Somebody makes any claim/finding, we all should be very suspicious for a long time until almost none of us are

• The opposite: when individuals fail to be skeptical, if everyone agrees to it, it must be true

• Skeptical but not Cynical

2.  Universalism

Scientific truth is independent of who produces it

• The opposite is particularism, whatever you think must be the truth especially if you’re an important person

3. Communitarianism

Data, methods, theories, and results must be shared

• Response to replication crisis

4. Disinterestedness

Researchers should minimize personal and political bias

• Ex. Payday loans:

  • High interest rates (50-70%), researchers looked at how predatory these companies were, said there should be laws against them.

• Risk: researchers may become advocates, allowing moral commitments to influence the interpretation of data

• Being socially responsible without compromising objectivity

Hypothesis

Testable predictions derived from a theory

• Ex. Bystander effect

  • Kitty Genovese—supposedly 38 witnesses who did not intervene,

  • Hypothesis: “the more people present the less likely you are to get help”

Meehl

Real sciences have hard theories

• Not whether an effect exists, but how much

• If the number of bystanders increases by x, helping decreases by y

True Experiment

• Independent Variable

• Dependent Variable

• Reliability, Validity

  • Internal Validity

• Random Assignment

Independent variable

variable we manipulate, the cause

Dependent variable

the measured outcome

Reliability

Consistency

• Going to the gym, there’s a scale, that says 230, again it says 180, then it says 225, etc. A different number every time, the scale is not reliable, is it consistent

Validity

Accuracy (measuring what it should measure)

• Every time you step on a scale it says 210, but that’s not the weight, it’s not valid, is it measuring what it’s supposed to be measuring

Internal Validity

Confidence that the IV caused the DV

• Threatened by confounds and endogeneity

Confounds

Mixing of effects where another variable changes alongside the IV, making it unclear what actually caused the outcome

• Ex. The shotgun

  • Woman hears a bang and an object explodes. Concludes the bang caused the explosion. Misses the shotgun. Her presumed IV is actually a result, not the cause

• In the example on leadership, one school could be failing while the other is doing good, not the leader of the groups

Endogeneity

When a variable believed to be a cause is itself influenced by the outcome.

Random Assignment

eliminates a lot of threats to internal validity. Ensures groups are equivalent on average before the manipulation

• Fischer

Fischer

Fertilizer was applied unsystematically across fields, he argued plots must be randomly assigned to fertilizer vs. no fertilizer

• Said without randomization causal conclusions are impossible

  • Smoking and Cancer: Cannot randomly assign people to smoke, therefore alternative causes are always possible which is why evidence comes from quasi-experiments.

Quasi-Experiment

Looks like a true experiment but lacks random assignment

• Ex. Rich and poor, the groups exist naturally, cannot randomly assign

• Fischer’s argument about smoking

  • Claimed smoking cannot be proven to cause cancer, argued the relationship is genetic

Hill and Doll

Disagreed with Fischer. Wrote a paper showing smoking causes cancer, as smoking comes before cancer, smoking is correlated to cancer, the more you smoke the more likely you are to have cancer, if you quit, rates go down.