Research Methods: Key Concepts & Design (Quick Reference)

Sample Size and Study Designs

n = number of participants; often very small in some designs (e.g., 1, 2, 4).
Small-n studies rely on descriptive/qualitative findings rather than broad generalization.
Cross-sectional designs: compare different ages at different times; not same individuals followed over time.

Strong correlation examples: r \approx 0.9 (closer to perfect) vs weaker correlation r \approx 0.6.
As one variable increases, the other may increase (positive correlation) or decrease (negative correlation).
Correlation does not imply causation: a relationship does not prove one variable causes the other.
Example: visual acuity unrelated to hair color; age may cause vision decline (confounding variable).
When describing relationships, distinguish correlation from causal mechanisms.

True experiments: independent variable (IV) manipulated, dependent variable (DV) measured; random assignment to groups.
Placebo control: a sugar pill used to control for placebo effects in drug studies (e.g., antidepressants).
Example: methylphenidate (Ritalin) study: IV = drug, DV = activity level.
Observational designs often cannot establish causation due to lack of randomization and control of confounds.

Quasi-experiments: experiments without random assignment.
Limitations: less control over confounding variables; weaker causal inferences.
Validity concerns: sampling methods (e.g., every seventh person from a phone book) can threaten representativeness and conclusions.

Hard sciences (biology) and social sciences (sociology/psychology) rely on manipulating/measuring variables to infer effects.
Common denominator: you work with what exists in the system or you manipulate a variable to observe outcomes; you cannot arbitrarily alter everything.
Analogy: you can manipulate rocks in geology but cannot alter the fundamental biology of a person for a study; this shapes study design (true vs quasi-experiments).

The U.S. has among the lowest breastfeeding initiation and duration in the industrialized world.
U.S. also has high infant mortality in certain comparisons; correlations with breastfeeding rates are not proof of causation.
Observational notes: some groups (e.g., Asian women in the U.S.) have high breastfeeding rates but different health outcome profiles; avoid simplistic conclusions.
Key takeaway: observational trends require careful analysis of confounders (immune status, environment, access to care, etc.).

Understand what n means and its impact on conclusions.
Distinguish correlation from causation; identify potential confounders.
Differentiate between IV and DV; use proper control groups (placebo where appropriate).
Recognize when a study is quasi-experimental and interpret causality accordingly.
Consider sampling validity and cross-disciplinary contexts when evaluating research evidence.