PSY3217 – Cultural Issues in Psychology: Cross-Cultural Research Methods

Identify universals vs culture-specific aspects of human behaviour
Broaden samples beyond WEIRD (Western, Educated, Industrialised, Rich, Democratic) populations
Clarify how culture shapes behaviour and how behaviour maintains, reproduces, or changes culture
- Bidirectional influence → dynamic systems view

Participants from several African sub-cultural groups shown a two-dimensional drawing (elephant, antelope, hunter, horizon line).
- Questions: “Which animal is nearer?”, “What is the man doing?”
- Some groups interpreted picture in a flat, non-perspectival way (no depth cues) → thought elephant was smaller or nearer, hunter aiming at elephant, etc.
- Demonstrates that perception of pictorial depth is learned and culturally variable, not a strict universal.

Method-validation studies
- Purpose: confirm that an existing scale/measure assesses the same psychological construct in another culture.
- Typical question: “Does item X load on the same factor across cultures?”
Indigenous cultural studies
- Explore phenomenon within a single cultural context, using locally relevant concepts.
- Avoids imposing outsider categories.
Cross-cultural comparisons
- Compare two or more cultural groups on the same phenomenon.
- Central question: “Are mean levels / structure of construct Y different across cultures?”

Exploratory vs Hypothesis-testing
- Exploratory: open search for any differences.
- Hypothesis-testing: theory-driven predictions (e.g., collectivism → higher interdependent self-construal).
Contextual factors
- Must ask why a difference occurs. Is it genuinely cultural or due to a confound (e.g., $SES$ , education, urbanicity)?
Structure- vs Level-oriented
- Structure: “Is the factorial configuration the same?” (qualitative form)
- Level: “Is the mean score higher/lower?” (quantitative magnitude)
Unit of analysis
- Individual-level (within each culture)
- Ecological-level (aggregated culture/nation means, norms, GDP, etc.)

Research question (RQ) must link a cultural variable to a specific psychological variable.
- Move beyond mere nationality labels → specify mediators/moderators (e.g., values, norms, institutions).
“Unpackaging” = identifying concrete mechanisms behind cultural differences instead of reifying “culture”.
Toolbox for unpackaging:
- Experiments (e.g., manipulate independence vs interdependence primes)
- Priming studies (cultural frame switching in biculturals)
- Behavioural tasks, physiological measures, naturalistic observation
- Context variables (parenting style, educational system, relational mobility, pathogen prevalence, etc.)

“White Americans have higher average IQ compared with African-Americans.”
- Raises questions of measurement equivalence, socio-economic confounds, historical oppression, stereotype threat.
“Children in Colombia develop theory of mind later than children in Australia.”
- Could reflect linguistic structure, schooling, parental mental-state talk.
“When recalling happiness, Japanese students cite socially interdependent situations; North Americans focus on personal achievement.”
- Example of culturally shaped emotion concept & memory.

Conceptual bias: Is the theory/construct meaningful in all cultures tested?
Method bias
- Sampling, linguistic, procedural, administration differences.
Measurement bias
- Psychometric inequivalence, lack of structural equivalence.
Interpretational bias
- Over-generalisation, cultural attribution errors.

Culture often treated as independent variable in designs $Culture \rightarrow Outcome$ , yet samples may differ on age, education, $SES$ , rural/urban status, internet literacy.
Remedies
- Match samples on potential confounds.
- Use multi-level modelling to partition individual vs cultural variance.

Lexical gaps: e.g., many Indigenous Arctic languages have numerous words for “snow”; English does not.
If a term has no direct counterpart, conceptual equivalence is threatened.
Back-translation procedure
1. Translate original English instrument → target language.
2. Independent bilingual translates it back to English.
3. Compare versions, resolve discrepancies.
Humorous real-world mistranslations (illustrate stakes):
- “Ladies, please leave your clothes here and spend the afternoon having a good time.” (Laundry, Rome)
- “Drop your trousers here for best results.” (Bangkok dry cleaners)
- “Specialist in women and other diseases.” (Italian doctor)
- “We take your bags and send them in all directions.” (Dutch airline)
- “The manager has personally passed all the water served here.” (Acapulco restaurant)

Psychometric operationalisation: same indicators must relate to latent construct similarly across groups.
- Techniques: Confirmatory Factor Analysis (CFA) for configural, metric, scalar invariance.
Structural equivalence: identical underlying factor structure.
- Example: González-González et al. (2015) intercultural empathy scale validated for cross-cultural use.
Without invariance, comparing raw means is meaningless.

Extreme responding (tendency to choose endpoints)
- Cultural scripts: collectivist norms (“The nail that sticks up gets pounded down”) may suppress extremes.
Acquiescence bias (“yea-saying”)
Socially desirable responding
- a) Self-deceptive enhancement (“I’m not racist!”) – unconscious.
- b) Impression management (“You mustn’t think I’m racist!”) – strategic.

Remedies: balanced keying, forced-choice formats, anonymity assurances, behavioural or implicit measures.

Statistical significance p < .05 ≠ practical or theoretical significance.
- A tiny mean gap may be statistically significant with large $N$ but irrelevant in real life.
Visualisations: overlapping distributions of males vs females (example slide) remind us effect sizes matter.
Researchers must check own cultural “blinkers” – tendency to explain all differences as cultural even when economic or historical.
Cultural attribution fallacy: Ascribing causal power to culture per se without measuring mediators.

Pre-register hypotheses & specify the cultural mechanism.
Use multi-method convergence (survey + behavioural + qualitative).
Check measurement invariance before mean comparisons.
Ensure rigorous translation/back-translation and pilot testing.
Report effect sizes ( $d$ , $\eta^2$ ), confidence intervals, and contextual variables.
Interpret cautiously; avoid stereotyping or essentialising cultures.