Chapter 3 Key Takeaways

The Research Methods Toolkit

Attraction research uses a range of designs: experimental, correlational, observational, survey-based, qualitative, and neuroimaging. Each has characteristic strengths and limitations; no single method is sufficient on its own.
Experimental designs allow causal inference through random assignment but typically sacrifice ecological validity — the conditions that make studies internally valid often make them less like the real world.
Qualitative methods are not merely preliminary to "real" research; they generate kinds of understanding that quantitative methods cannot capture, particularly regarding how people make meaning of their own attraction experiences.

The acronym WEIRD (Western, Educated, Industrialized, Rich, Democratic) captures the dramatic unrepresentativeness of the populations on which most psychological research is based.
Approximately 90% of published attraction studies have drawn samples exclusively from North America or Western Europe. This is not a minor sampling detail — it shapes what the field thinks it knows about human desire.
The solution is not merely to replicate studies in different countries, but to ask whether the concepts and instruments developed in Western research contexts are appropriate starting points for non-Western populations.

Statistical significance (p < .05) and practical significance are not the same thing. With large samples, trivially small effects can be statistically significant; with small samples, genuine effects may fail to reach significance.
Cohen's d provides a sample-size-independent estimate of effect magnitude: 0.2 = small, 0.5 = medium, 0.8 = large (rough guidelines, not absolute thresholds).
Most published effects in attraction research are small to medium in size, meaning they explain a modest proportion of the variance in attraction outcomes. This is important context for interpreting popular-science headlines.

The Open Science Collaboration (2015) found that only about 39% of a sample of 100 psychology studies replicated successfully — a result that prompted widespread re-evaluation of the field's confidence in its findings.
The mechanisms behind replication failures include underpowered samples, flexible analysis practices (p-hacking), and HARKing (Hypothesizing After Results are Known).
Publication bias — the tendency to publish significant results and file-drawer null results — systematically inflates the effect sizes in the published literature. Meta-analyses that do not correct for this will also overestimate effects.
Pre-registration, in which researchers publicly specify hypotheses and analysis plans before collecting data, is a partial but meaningful remedy for these problems.

Self-report, behavioral, and physiological measures do not always agree. Each offers a different window onto the psychology of attraction, and each has characteristic limitations.
The appeal to physiological measures as "objective truth" behind subjective experience is misleading: physiological responses are imperfect proxies for psychological states, and require careful interpretation.

IRB review exists to protect participants from harm, but ethical research involves more than compliance. Culturally sensitive, collaborative research design — like the approach Okafor and Reyes developed for the Global Attraction Project — is an ethical commitment, not a bureaucratic requirement.

When encountering any attraction research claim, ask: Who was the sample? What was the design? What is the effect size? Has it replicated? What does it not tell us? These questions are not expressions of cynicism — they are how good science works.