Cross-Cultural Lecture Flashcards
An additional level of bias not in lecture but in the reading:
sampling bias:
• Refer to the characteristics of a sample
that is used for comparison.
- In cross-cultural samples there is a weird bias where most of the samples are university students. This produces biases especially in third world countries where being apart of tertiary education makes you a social elite. Variations in the competitiveness of getting into university, the level of prestige it holds, and socioeconomic variations in students for each society which influence samples motivations, attitudes and belief to cultural variables and influence their responses on a survey.
- Interpretation-paradox: it is relatively easy to find differences between samples in social, cultural and economic dimensions but it’s more difficult to explain why these differences exist and what factors can explain them.
what are uniform and non-uniform bias?
An issue with Metric Bias: Factor Loadings (slope)
- A differential relationship between the observed indicator and the latent variable is called non-uniform bias (i.e., changes in latent variable are not uniformly associated with the observed behavior =lines are crossed).
- Uniform bias (one line uniformly above the other) indicates that there is some difference in the relationship between latent variable and the outcome across groups, but it is not due to the latent variable
- methods used in CC psychology?
- Ethnography
- Interviews/Focus Groups
- Cultural Product & Content
- Psychometric Measures (bias and equivalence are applicable).
- Experiments
- Priming
Postmodernism and Post-Positivism
Postmodernism: developed in critique to cross-culture’s positivist view. Views culture as being constructed and reconstructed by language use and the focuses on a way of being rather than searching for similarities and differences. Argues meanings are too subjective and fluid to be quantitively measured or generalized. They favor unstructured and qualitative methods.
Post-postmodernism (post-positivism): developed in response to the critiques against cross-cultural psychology. accept that knowledge is contingent on the process of social construction and is inevitably incomplete: but they still view there to be an external reality, and the scientific method is still the best way to describe both objective and constructed realities. Knowledge is viewed as something not absolute but open to modification based on further investigation believe it is impossible to understand someone’s subjective experience, so they treat psychological variables as a latent variable.
*people should use a mixed-methods approach and researchers should engage in reflexivity
of their ontological and epistemological views that can bias research.
Psychometric measures:
- Most objective method to obtain data using standardized paper-pen questionnaires or online surveys to collect data on psychological constructs.
- Standardized in the sense that they have fixed formats (i.e., Likert-scale asks people to rate how much they agree or disagree to each statement on a fixed scale).
- Involve a level of introspection by the participant to answer questions.
- Provides us valid and reliable information about latent constructs/variables which cannot be directly measured or observed.
- Reliability is inferred by looking at pattern in response by individuals across items.
- Validity based on how well the items are at measuring the construct or latent variable: construct validity is higher order more abstract form of validity and can be supported by identifying convergent and discriminatory validity. It correlates with validated scales on the same construct or same outcomes (convergent) and does not correlate with items unrelated to the construct (discriminant).
- The setting in which the scale is administered will impact on the objectivity of the measurements.
Culture content and cultural products
- Advantage of these approaches is that they tap into data that individuals produce either for public consumption or in order to communicate to the outside world and represent the making of culture.
- Downside is that it requires trained coders, careful interpretation and carefully prepared coding methods.
- People tend to subconsciously alter their behavior to the stereotypical norms of the language items are presented to them in. Therefore, the items should use the language native to the individual to ensure we are not under-representing cross-cultural differences. Is an effect that is exploited in priming methods.
e.g. twitter, websites, newspaper etc.
Experiments
- Culture is a subject variable therefore people cannot be randomly assigned to a condition. Thus, a true experiment cannot be conducted, and quasi-experimental designs are used.
- The independent variable their nation or ethnicity is hypothesized to indirectly measure the effects of culture (IV). Unless a strong mediator variable is included, we cannot make claims about cultural effects.
- Cultural background of participants moderates the effect of the experimental manipulation, but it often remains packaged (i.e., it’s unclear what specific effects it has on the IV-DV relationship).
- The success of an experiment hinges on the researcher’s ability to show the manipulation was effective in both contexts and rule out alternative explanations.
*experimental studies select samples based on underlying cultural dimensions that are
thought to influence the dependent variable of interest. Experiments in cross-cultural
psychology are quasi-experimental because participants cannot be randomly assigned
to experimental conditions. They are always open to alternative explanations.
- Priming Studies:
- Priming to temporarily activate procedural and tactic knowledge (cultural mindsets) and mental representations (beliefs, norms, goals and values) in people’s minds.
- This evoked cultural mindset serves as an interpretive framework for processing any subsequent information and switches to heuristic processing strategies that are relevant to a given context.
- Participants engage in a task that is designed to prime them (i.e., activate cultural relevant goals, motivations ad knowledge) and is designed to have a spillover effect onto their performance in the next task (i.e., the cues made ore salient from the first task effect their performance on the second task). DV embedded in second task. IV=prime.
- Can randomly assign people to a condition (i.e., collectivism or individualism prime or experimental and control condition).
- Between subject’s design is designed to replicate cross-cultural differences.
- Prime culture with national symbols or languages.
Interviews and focus groups
- Require interaction between the researcher and members of the cultural group.
- Unstructured or structured
- The negative of using an unstructured interview is that it makes findings hard to compare cross-culturally raising concerns about validity and reliability.
- The negative of a highly structured interview is that it can fail to explore the experiences and viewpoints of the interviewee on a specific topic, but the findings are more easily comparable.
- These are primary methods used in indigenous research.
- Cross-cultural psychologists are more likely to use structured interviews.
unstructured interviews are more likely to be in ___ and structured in ___
indigenous (qualitative) cross-cultural (quantitative).
Interviews and focus groups
- Require interaction between the researcher and members of the cultural group.
- Unstructured or structured
- The negative of using an unstructured interview is that it makes findings hard to compare cross-culturally raising concerns about validity and reliability.
- The negative of a highly structured interview is that it can fail to explore the experiences and viewpoints of the interviewee on a specific topic, but the findings are more easily comparable.
- These are primary methods used in indigenous research.
- Cross-cultural psychologists are more likely to use structured interviews.
- The advantage of focus groups is that the discussions reveal areas of contention and multiple perspectives from members of the same or different cultural groups.
- Negative of focus groups is that the personality characteristics (assertive or shy) may change the dynamic of the interview and differences in social status in some cultures may change the discussion outcome as well making some opinions more salient and suppressed.
unstructured interviews are more likely to be in ___ and structured in ___
indigenous (qualitative) cross-cultural (quantitative).
- The choice in whether a structured or unstructured interview or focus group is used depends on the researcher’s ontology and epistemology views.
- Structured interviews are used for people who believe in monism, objectivity, positivism and that themes of discussion are quantifiable and subject to statistical analysis.
- Unstructured approaches are adopted by researchers who view reality as being more subjective, are more likely to use qualitative methods to identify novel themes by engaging with the participant without seeking to identify universals or broad generalizations about the study.
ethnography
2 problems in cc
- the researcher must have the scientific goal in mind to understand the complete society rather than just reporting the salient factors to a foreign observer.
- Common in anthropology, cultural psychology and some indigenous approaches.
Problems with ethnographic methods:
o Researcher may impose their beliefs and positionality onto their interpretations of their findings (i.e., impose feminist beliefs).
o The status of the researcher in the group being studied (i.e., the presence of the observer changes their behavior).
A national level dimension of culture raises two important questions: 1) will the instruments and the constructs at the nation level still have the same structure at the individual level (___) 2) does the relationship between psychological constructs and a third variable in the same nation level as it is in the individual level (__).
isomorphism, homology
Can cultural groups such as nations be compared using individual-level scale means?
No
Can individuals be compared using nation-level value scales?
No
*patterns of value structure are different at the individual level relative to the national level.
why are nation level analysis of individual’s aggregated data valid?
nation- level variance is valid in its own right because we are dealing with social aggregated data of individuals. Nation level comparisons are made with mean scores, so they are statistically independent of individual level correlations between individuals.
At a nation level we are looking at eco-logic rather than psycho-logic.
What is found at the national level cannot be meaningfully linked to individuals.
Multi-Level Modelling:
- Data that has been aggregated to the national level, averaged within each nation and factor analysis has been conducted to determine dimensions of culture at a national level.
- We can explore the correlations with other indicators to establish validity.
- Mathematically the within nation and between nation variance is statistically independent. However, when the means of each group are different this will influence the correlation of means (national level) will be different to the correlation of means within groups.
- More advanced statistical analysis can allow for us to test whether the aggregated dimensions (national level) have an influence on the psychological mechanism on an individual’s level whilst controlling for third variables at an individual level.
- When can separate the effect of individual level value priorities from the effects of living in a particular context in which certain values are prioritized.
- We can use moderation analysis to examine the strength of the relationship (IV-DV). To determine if the effects is stronger for certain groups of people (i.e., who).
MODERATOR
The Unpackaging of Culture:
- We can unpackage the culture by doing an extension of basic cross-cultural research at test if a specific variable (i.e., mediator) is the cause of the effect we see.
- Unpackaging culture can also be referred to as linkage studies, covariate studies or mediation studies.
In summary, the unpackaging of culture involves two inter-related features 1) the identification of theoretical factors or processes which may produce cultural differences in psychological outcomes of interest 2) the explicit empirical test of the proposed process leading to these outcomes.
when is unpackaging of culture used?
when their is insufficient number of samples to use multi-level modelling.
Meta-Analysis: A useful tool for integrating research:
• A quantitative technique which aggregates results across many published studies.
• It is an analysis of previous analyses. Meta-Analysis can be conducted if:
o You have an effect size: a numerical measure of the expression of psychological characteristics (i.e., a means frequency).
o The strength if an association between two psychological constructs (i.e., correlation).
o The mean difference of a psychological characteristic between two or more groups.
• Results from previous studies may need to be converted to allow for comparisons (i.e., weighted samples smaller sampled studies hold more weight or adjusted for criteria of quality with higher quality studies holding more weight).
• Once we have a combined mean effect, we can test whether the studies are homogeneous (i.e., do they show the same effect).
• If the studies show different effects and are not homogenous that the researcher can begin to look for moderator variables which are producing these differences.
• Sensitive measure, good at synthesizing collective studies, identifying gaps in previous literature and studying novel hypothesis.
(5) Guidelines of cross-cultural research:
- Researchers should recognize the cultural contingencies of their own beliefs and values when conducting research and should resist evaluating other cultures against our own cultural beliefs.
- Researchers should study psychological processes in their native language of participants, using English is likely to lead to cultural accommodation of responses to the perceived cultural stereotypes of the language being used (i.e., people responses are responsive to the language the items are phrased and can underestimating cross-cultural differences).
- Samples should be matched closely as possible in order to rule out alternative explanations for observed differences in outcomes being studied (i.e., confounds).
- Instruments need to be valid and reliable in all cultural groups in order to accurately detect cross-cultural similarities or differences. Researchers should try to ensure that instruments, measures, and manipulations are understood in comparable ways in each location.
- Experiments are more persuasive in their evidence if a) they include a manipulation check on the effectiveness if the crucial manipulation in all cultural samples, and b) the experimenter specifies the relevant psychological processes and relates then to the dependent variable of interest, to rule out alternative theoretical explanations for the observed cultural difference.
Are cross-cultural comparisons are open to interpretation? how do we determine 1) if an interpretation is valid? 2) If the instrument[s] producing the observed differences are valid?
Yes. Should check for (3) biases: • Constructs Examined (construct bias) • Administration procedures (method bias) • Operationalisation (item bias)
Construct Bias:
- Incomplete (rather than no overlap) overlap of definitions of the construct across culture.
- Differential appropriateness (sub) test content (i.e., the skills being studied are not in the behavioural repertoire of all cultural groups).
- Poor sampling of all relevant behaviours (short instrument i.e., survey length).
- Incomplete coverage of the construct (i.e., all of the relevant domains to the construct are not tested, domain under-representation, where stable relationships is considered to be a part of intelligence in collectivist cultures).
example of construct bias:
The classic marshmallow test
• Does the ability to delay gratification in
childhood lead to better adult outcomes?
• Follow up test found that people who did better at waiting did better on their SAT scores, more likely to have tertiary education and had a lower BMI score.
• People interpreted this to mean that willpower is stable trait over time that is determined in early childhood and predictive of life success.
• Does it measure willpower?
No: It has construct bias for two reasons:
o Brain scans show that the waiters had less reward activation than the non-waiters which indicates that the meaning of the test may be fundamentally different between the two groups.
o It may not be measuring willpower at all. Brain scans show that for children who waited it wasn’t the brain regions associated with self-control that were activated.
Method Bias
- Is the same construct measurable in the same way across cultures?
- Is the instrument equally familiar to each cultural group?
- Differential social desirability (i.e., language the items are written in prime socially reliable behaviours based on the activated culture).
- Differential response styles such as extremity scoring and acquiescence (i.e., favouring one side of the scale regardless of item content).
- Differential stimulus familiarity (i.e., money in economic game etc.).
- Lack of Comparability of samples (i.e., differences in educational background, age or gender composition).
- Differences in physical/social conditions of administration (i.e., temperature of the room, presence of others).
- Differential familiarity with response procedures (i.e., pen and paper).
- Tester/interviewer effects (I.e., status of the researcher, ethnicity, professional status, religion or whether they’re local or foreigners)
- Communication problems between respondent and tester/interviewer in either cultural group.
Example of method bias
• E.g. the audio illusion do you hear laurel or yanny?
o 47% hear Yanny and 53% hear Laurel.
o Why do we hear them? Priming, the video asks do you hear laurel or yanny. If you were not primed to hear these words you may never have heard them.
o The audio waves are very similar but what you hear will depend on the sound quality of the device you are using.
How does this demonstrate method bias?
o If the test was to identify which word was being said. The overlay of higher pitched sounds bias this test against hearers who are more likely to detect the higher pitches, making it unfair due to the method testing recognition.
o Showing the two written word for us to read primes us to hear one or the other. Otherwise we may have heard something completely different.
Item Bias:
*specific items or questions in a larger test.
- Poor translation or formulation of items (i.e., complex wording, accepted method is to use back translation).
- Nuisance factors (i.e., items which invoke additional traits or abilities).
- Incidental differences in appropriateness (i.e., school curriculum of one group).
- Cultural specifics (i.e., connotative meaning, general appropriateness).
- Example:
