1 emmanuel chéron, ph.d. 上智大学 sophia university graduate program in global studies...
TRANSCRIPT
1
Emmanuel Chéron, Ph.D.上智大学
SOPHIA UNIVERSITYGRADUATE PROGRAM IN GLOBAL STUDIES
International Business/EconomicsTokyo
Data Equivalence in Cross-cultural Research: Methods and Tools
2
Agenda Emic/etic controversy Emic/etic approach implementation steps Overall framework of data equivalence in cross-cultural research Equivalence of research topics Equivalence of data collection Equivalence of data preparation Statistical tests of measurement equivalence Alternative approaches Conclusion
3
Emic/etic controversy
• Emic school (phonemic):
Attitudes, interests and behavior are unique to each culture
An emic perspective implies an ethnographic research approach with limitations in terms of comparison and generalization
4
Emic/etic controversy• Etic school (phonetics): Attitudes and behavior are common across cultures allowing researchers to conduct inter-cultural measurements and comparaisons Any comparison conducted in international marketing research postulates a valid etic model exposing to a “pseudo-etic” bias risk (forced etic school) Validity and reliability of measures between countries need to be checked (a simple internal reliability coefficient such as Cronbach alpha is not enough)
5
Emic/etic implementation steps
Research steps Culture A Culture B (native) (foreign)
1. Start in A
2. Transfer in B
3. Discovery of B
4. Comparison of cultures
5. Comparison not possible
6. Comparison possible
EmicA
EmicA
EmicA
EmicA
ForcedEtic
EmicB
EmicB
EmicB
Shared Etic
EmicB
`
6
Key point of emic/etic controversy
Examples of sources of difference: Geographical, climate, demographics, political system, economics, regulations, ethics, cultural, social, religious, distribution Networks and channels, etc.)
Be aware that cultural contexts between culturesoften do not completely overlap
7
Equivalence in cross-cultural research
Problemdefinition
Datacollection
Datapreparation
Dataanalysis
Statistical tests of data equivalence
Equivalence of research topics• Functional, conceptual, category
Equivalence ofresearch methods• collection, stimuli
Equivalence ofresearch units
• definition, selection
Equivalence of administration
• timing, interaction
Equivalence of datahandling
• response translation, categories
Configural invariance• basic factor patterns correspond
Metric Invariance• factor loadings correspond
Scalar invariance• relationships of contructs-observed
Equivalence of data in cross-cultural research
• comparability of data
Multi-groupSEM (CFA)
orLatent trait
theory
8
Equivalence of research topics?
• Fonctional equivalence: meaning of physical training, jogging, shopping, of owning certain objects?
• Conceptual equivalence: meaning of stimuli (couleurs, nombres, symboles, objets) of behavior, gestures, social rituals (graduation, marriage, funeral ceremonies, gift giving)
• Category equivalence: category of objects (beer as an alcoholic beverage, milk with meals, hot vs cold continuum of parfumes in France, meaning of marital status, of a biological mother in Mali, ranking of professional status)
9
• Danger of self-reference to ones native culture• Importance of cultural understanding
Key points of problem definition
10
Equivalence in cross-cultural research
Problemdefinition
Datacollection
Datapreparation
Dataanalysis
Statistical tests of data equivalence
Equivalence of research topics• Functional, conceptual, category
Equivalence ofresearch methods• collection, stimuli
Equivalence ofresearch units
• definition, selection
Equivalence of administration
• timing, interaction
Equivalence of datahandling
• response translation, categories
Configural invariance• basic factor patterns correspond
Metric Invariance• factor loadings correspond
Scalar invariance• relationships of contructs-observed
Equivalence of data in cross-cultural research
• comparability of data
Multi-groupSEM (CFA)
orLatent trait
theory
11
• Equivalence of research methods Collection techniques Stimuli (verbal, visual)
• Equivalence of research units Administrative units (urban, rural) Consumption unit Buying decision roles
• Equivalence of administration Comparable timing Interaction with respondents
Equivalence of data collection?
12
Key point of equivalence of datacollection
Be aware that cross-cultural data equivalencemust be balanced with limitations involved in local data collection administration
13
Equivalence in cross-cultural research
Problemdefinition
Datacollection
Datapreparation
Dataanalysis
Statistical tests of data equivalence
Equivalence of research topics• Functional, conceptual, category
Equivalence ofresearch methods• collection, stimuli
Equivalence ofresearch units
• definition, selection
Equivalence of administration
• timing, interaction
Equivalence of datahandling
• response translation, categories
Configural invariance• basic factor patterns correspond
Metric Invariance• factor loadings correspond
Scalar invariance• relationships of contructs-observed
Equivalence of data in cross-cultural research
• comparability of data
Multi-groupSEM (CFA)
orLatent trait
theory
14
• Translation equivalence Limitation of back-translation and decentering
(Schadenfreude, なつかしい nattsukashii)
• Measurement systems equivalence Currency exchange, purchasing power parity Physical measurement systems (comparable quality standards)
• Equivalence of measurement scale
Equivalence of scoring scale No-saying and yeah-saying effects Equivalence of response style
(extreme response style, response range)
Equivalence of data handling?
15
Key points of equivalence of data handling
Make sure that translations, measurement systems,scoring systems and response styles are equivalent
• Baumgartner and Steenkamp (JM, 2001) using GfK survey data on 11 European countries found an average response style effect of 8% on the variance of 60 5-point Likert (degree of agreement) scales When measuring consumer ethnocentrism and health consciouness, they found a respective effect of 11 to 23% and 12 to 29% depending on the country The relative effect size between countries was found smaller than between scales
16
Equivalence in cross-cultural research
Problemdefinition
Datacollection
Datapreparation
Dataanalysis
Statistical tests of data equivalence
Equivalence of research topics• Functional, conceptual, category
Equivalence ofresearch methods• collection, stimuli
Equivalence ofresearch units
• definition, selection
Equivalence of administration
• timing, interaction
Equivalence of datahandling
• response translation, categories
Configural invariance• basic factor patterns correspond
Metric Invariance• factor loadings correspond
Scalar invariance• relationships of contructs-observed
Equivalence of data in cross-cultural research
• comparability of data
Multi-groupSEM (CFA)
orLatent trait
theory
17
• Configural invariance Test of the measurement model within culture Test of cross-cultural configural invariance
• Metric Invariance Test of score equivalence given cross- cultural configural invariance
• Scalar Invariance Test of a common cross-cultural scale origin (partial equivalence?)
• Invariance of latent response Test of cross-cultural equality of parameters of the response probability model of each survey question
Statistical tests of data equivalence?
18
Key points of statistical testsof data equivalence
The choice of equality constraints may change the test results of scalar invariance Qualitative empirical judgement is still neededto identify invariant items between culture
19
Compare actual cross-cultural buying behavior rather than non-observable survey data• Data mining of sales transaction• Latent class analysis to identify market segments
Cross-cultural comparison of observed response data in experimental setting• Neuromarketing (Functional Magnetic Resonance Imaging)• 3D Simulation of commercial setting• Eye-tracking
Alternative approaches
20
Brand Recognition and Cultural Differences -- Heatmap Data real-time eye-tracking system
(Source: JCMR, "Brand recognition and cultural impact, 2005.10")
http://www.jmrlsi.co.jp/english/case/jmarket/2006/02_study_examples.html
21
Conclusion
There are many complex requirements for cross-cultural data equivalence when measuringnon-observable variables (attitudes, opinions,perceptions)
In spite of many refinements available to improvecomparison of non-observable variables, observed actual buying behavior and responsesto experimental settings offer attractive alternatives
22
References
Baumgartner, Hans and J-B Steenkamp (2001) «Response Styles in Marketing Research:A Cross-National Investigation», Journal of Marketing Research, Vol. 38, May, 143-156.Chéron, Emmanuel and Hideo Hayashi (2001), «The Effect of Respondents'Nationality andFamiliarity with a Product Category on the Importance of Product Attributes in Consumer Choice:Globalization and the Evaluation of Domestic and Foreign Products», Japanese PsychologicalResearch, Volume 43, No. 4, 183-194.Chéron, Emmanuel J.; Tetsuo Sugimoto and Hideo Hayashi, (1994), «Usage Frequency andPurchase Motives of Consumer Products: A Comparison between Canada and Japan»,Asian Journal of Marketing, Vol. 3, December, 7-20.Chéron, Emmanuel J.; Thomas C. Padgett and Walter A. Woods (1987), «A Method forCross-Cultural Comparisons for Global Strategies», Journal of Global Marketing, Vol. 1,Nos. 1 & 2, Fall/Winter, 31-51.Laroche, Michel; Linda C. Ueltschy; Shuzo Abe; Mark Cleveland and Peter P. Yannopoulos(2004) «Service Quality Perception and Consumer Satisfaction: Evaluating the role ofCulture», Journal of International Marketing, Vol. 12, No. 3, 58-85.Salzberger, Thomas; Rudolf R. Sinkovics and Bodo B. Schlegelmich (1999) «Data Equivalencein Cross-cultural Research: A Comparison of Classic Test Theory and Latent Trait Theory BasedApproaches», Australasian Marketing Journal, Vol. 7, No. 2, 23-38.Steenkamp J-B and Hans Baumgartner (1998) «Assessing Measurement Invariance inCross-National Consumer Research», Journal of Consumer Research, Vol. 25, June, 78-90.Usunier, Jean-Claude and Julie Anne Lee (2005), Marketing Across Cultures, 4/E, Pearson,Prentice Hall Europe. ISBN: 0-273-68529-5.