Taking a closer look at the impact of culture onMulti Rater Feedback PhD Student Copromotor Jouko van Aggelen, MsC Josje Dikkers, PhD Managing Consultant Cubiks VU University Amsterdam Jouko.firstname.lastname@example.org email@example.com Promotor Copromotor Prof. Paul G.W. Jansen, PhD, Rob Feltham, PhD, VU University Amsterdam Director IPT Cubiks firstname.lastname@example.org Rob.Feltham@cubiks.com
Content Introduction, theoretical background Hypothesis Research method Results & Discussion Next steps, questions & suggestions
Introduction• For many organizations it has become common practice to use some kind of Multi Rater Feedback.• “As these instruments gain popularity around the world the resulting need to conduct multisource research across and within different countries intensifies.” (Atwater, Brett, & Charles, 2007; Atwater, Waldman, Ostroff, Robie, & Johnson, 2005; Brutus, Leslie, & McDonald-Mann, 2001).• Although research has shed some light on the phenomenon, its use in a cross-cultural environment is less explored. This is even more relevant when one realizes the impact they can have on individuals
Introduction• This current research combines my daily life as a consultant (working for multinational organisations) with a more scientifically grounded approach.• I would like to share some first results and collect your feedback and suggestions.
MRF• Three-hundred-and-sixty degrees feedback was first used (and detonated as a trademark!) by the American Consultancy Firm ‘Teams inc.’ in 1985 (e.g. Jansen and Vloerbergs, 1998).• It was Jack Welch who started the glory days of this method in 1994. Definition: “A process whereby raters from multiple perspectives rate a subjects performance“ (Zimmerman, Mount & Goff III, 2008, p. 123)
Cross-cultural usage of MRF• People from different groups, or cultures have a different way of acting and performing (Kruger & Roodt (2003).• Contextual differences have a greater influence on employees’ receptiveness than personality. One of the most impactful contextual differences can be national culture (Funderburg & Levy,1997)• MRF ratings are based upon interpersonal interactions and shared feedback (Atwater et all 2009).• These processes, interaction and the sharing of feedback, are highly influenced by culture (Ashford, 1989, Varela & Premeaux, 2008)
Some examples of previous researchfindings• Gillespie (2005): Employees from different countries and cultures, working for the same multinational, interpreted and responded differently to the same MRF questionnaire.• Shipper, Hoffman and Totondo (2007): MRF has the most effect (gaining actionable knowledge out of the process) in individualistic and low power distance cultures.• Varela and Premeaux (2008): The discrepancy between peer- and self- ratings was the least and direct reports gave the highest ratings to their bosses in high collectivistic and high power distance cultures.• Atwater, Wang, Smither and Fleenor (2009): High assertiveness and Power Distance seems to stimulate the relationship (= low discrepancy) between self and reports and between self and peer ratings.• Gentry, Yip, & Hannum, (2010): The discrepancy seems to be wider in high power and individualistic cultures, mainly due to the subjects’ self-ratings, not the ratings of others.
Hofstede’s culture typology• One of the most popular and most used classifications of cultural differences (Atwater, Waldman, Ostroff, Robie & Johnson, 2005, Shipper, F. Hoffman, R.C. & Totondo 2007)• An accepted way to study cultural differences.• Hofstede’s dimensions (1980, 2002, 2010) – Power Distance: The distribution of power by nature. – Identity: Individual freedom vs. focus on the collective, group harmony. – Gender: A caring ‘Feminine’ attitude vs. an assertive Masculine one. – Truth. Tolerance for ambiguity and uncertainty vs. uncertainty avoidant – Long term vs. short term orientation. – Indulgence vs. Restrain
Focus of this study• Moderating role of culture on with respect to three of Hofstede’s cultural Dimensions: Power Distance, Identity & Gender. The discrepancy between the subjects’ self-ratings and the ratings of bosses & the possible interaction effects,
Some examples PDI COLL-IDV FEM-MAS Russia 93 39 36 New Zealand 2 79 58 US 40 91 62 Venezuela 81 12 73 Japan 54 46 95 Sweden 31 71 5 Netherlands 38 80 14
HypothesesPower Distance • Johnson, Kulesa, Cho & Shavitt (2005): PD is positive related with extreme response styles and negative with acquiescent response behavior (“yah-saying”) • Balzer et al., 2004, Hofstede 1983, Varela 2008: Reports ratings will be inflated since they want to maintain the relationship and that believe that their manager will know best. • Carl, Gupta, & Javidan: “Under the higher power distance style of supervision, there is virtually no rapport between the leader and subordinate” ( 2004, p. 535). 1. The discrepancy between self and boss becomes wider when the subject comes from a (more) high PD culture. 2. The discrepancy between self and boss becomes wider when the boss comes from a (more) high PD culture 3. The discrepancy is the least wide when both are from a low PD culture and the discrepancy is the widest when both are from a high PD culture
HypothesesIdentity • In collectivistic cultures giving feedback is not so easy. People will be less ‘tough’ on each other. In individualistic cultures people are focused on saving own face vs. saving face of others. This could result in a less critical self-assessment. • Reports gave the highest ratings to bosses in high collectivistic cultures (Varela and Premeaux, 2008) • The discrepancy seems to be wider in individualistic cultures (Gentry, Yip, & Hannum, 2010) 4. The discrepancy becomes more wide when the subject is from a more individualistic culture 5. The discrepancy becomes less wide when the boss is from a more collectivistic culture 6. The discrepancy is widest when both come from an individualistic culture and smallest when subjects come from a collectivistic and bosses from an individualistic culture
HypothesesGender • Goffin and Anderson (2006): High achievement orientation and a high self-esteem (masculine characteristics) are related to overestimation. • Another characteristic that is associated with Masculine cultures is assertiveness, which seems to be related to more comfortable with giving critical feedback. 7. The discrepancy is wider when subjects are from masculine cultures. 8. The discrepancy is lower when bosses are from masculine cultures 9. The discrepancy is widest when both come from an masculine culture and smallest when subjects come from a femnine and the boss from an masculine culture
The data • Three data sets – Company specific MRF’s and competences – Conducted with the same objective for more or less the same target groups – All with a developmental focus – All available in multiple languages • Three different global operating companies 1. Western European Food and beverage 2. Northern European Mechanical Engineering 3. Asian Steel Company • Three different regional spreads 1. Eastern and western Europe + Africa 2. US, Western and Northern Europe + some Asian countries 3. Europe, North and South America and Asia
Cultural spread Sample 1 Sample 2 Sample 3 Sample 1 Sample 2 Sample 3 295 1170 320 2719 10228 2931Subjects Reviewers Power Distance Power Distancemin 11 11 35 min 11 11 11max 104 104 104 max 104 104 104mean 70.50 49.03 61.72 mean 60.97 49.57 60.95SD 27.68 16.66 13.93 SD 27.48 19.51 14.44 Identity (individualism vs. Collectivism) Identity (individualism vs. Collectivism)min 11 14 12 min 8 13 12max 90 91 91 max 91 91 91mean 46.54 72.95 62.52 mean 47.89 71.70 63.00SD 19.05 15.84 17.80 SD 17.10 16.37 17,67 Gender (Masculinity vs. Femininity) Gender (Masculinity vs. Femininity)min 14 5 14 min 14 5 5max 110 79 88 max 110 110 110mean 54.94 49.25 51.64 mean 53.33 47.75 51.21SD 27.67 19.19 10.45 SD 27.12 19.02 10.36
Measures & analytical procedure• Correlations company specific competences all high, we chose for 1 single overall rating per reviewer• Calculated discrepancies between subject’s and boss’ rating• The cultural interaction is calculated by multiplying the ‘Cultural Z- scores’ of subject and boss• Multiple regression – Independent variables: The cultural dimensions of the subject, the boss and their interaction – Dependent variables: The subjects’ self rating, the reviewer/boss rating and the discrepancy.
First: Overall analysesSelf, boss & discrepancy ratings Data set 1 Data set 2 Data set 3 Mean self 3.26 3.61 3.12 SD Self .44 .46 .41 Mean Boss 3.25 3.46 3.09 SD Boss .52 .57 .45 Mean discrepancy .024 .141 .055 SD discrepancy .656 .666 .594 T –test discrepancy T = .664 T = 7.410 T = 2.115 p = .507 p = .000 p < .035 • Rating scale 1-5 • The mean discrepancy in data set 2 & 3 varies significantly
ResultsPower Distance as moderator on the self-boss discrepancy Data set 1 Data set 2 Data set 3 1 Subject (wider if subject is high PD) β = .021 β = .123 β = .029 p = .774 p. = .090 p. = .604 2 Boss (wider if boss is high PD) β = -.074 β = -.002 β = -.008 p =.308 p = .979 p = .979 3 Interaction (least if both low PD + vice versa) β = -.394 β = .105 β = -.066 p = .000 p < .002 p = .186 • (1) and (2) are not supported • (3) is supported in two of the three samples Only the interaction between the power distance of self and boss seems to be significant.
Interaction effects Power distance Data set 2• Most impact: PD Boss • Lowest discrepancy: LPD Boss, LPD self• Discrepancy is most negative if HPD Boss • Widest discrepancy: HPD Boss• Discrepancy is most positive if LPD Boss • Discrepancy is most negative: HPD Boss, LPD Self • Discrepancy is most positive: HPD Boss, HPD Self
ResultsIdentity as moderator on the self-boss discrepancy Data set 1 Data set 2 Data set 34 Subject (more wide if subject is β = -.172 β = -.164 β = -.080 individualistic) p <.005 p < .004 p = .2075 Boss (less wide if boss is collectivistic) β = .024 β = .074 β = .082 p = .701 p = .193 p = .2006 Interaction (widest if both from β = .160 β = .209 β = -.186 individualistic, least of subject is p < .032 p = .000 p < .001 collectivistic, boss ind.) • (4) is supported in two of the three samples • (5) is not supported • (6) is supported in all three samples. Individuality / collectivism of the subject seems to have an effect The interaction between the subject and boss also seems to have an effect.
Interaction effects Identity Data set 2 Data set 3 • Widest and most positive discrepancy: COL subject, COL boss • Lowest discrepancy: IND subject• Most negative discrepancy: IND Boss, IND subject • Widest & most positive discrepancy: COL subject,• Most positive discrepancy : IND boss COL subject, Col Boss
ResultsGender as moderator on the self-boss discrepancy Data set 1 Data set 2 Data set 3 7 Subject (wider if subject is β = -.074 β = .012 β = .049 Masculine) p = .276 p = .889 p = .378 8 Boss (wider of boss is Masculine) β = -.143 β = .049 β = -.083 p < .034 p = .566 p = .132 9 Interaction (widest when both from β = -.021 β = .126 β = -.031 masculine, smallest when subjects p = .751 p = .000 p = .512 is feminine, boss is masculine) • (7) is not supported in all three samples • (8) is not supported in two of the three samples • (9) is not supported in two of the three samples Masculinity / Femininity does not seem to have much impact, most hypotheses with respect to this dimension are not supported.
Interaction effectsGender • Widest discrepancy: MAS boss Data set 2 • Lowest discrepancy: FEM boss and subject
Conclusions• Power Distance • The cultural background of the subject and his/her boss itself does not seem to have moderate the wideness of the discrepancy • However the interaction between the two does seem to moderate the wideness of the discrepancy (2 out of 3). Based upon the current analyses the interaction effect itself is difficult to interpret, it’s really blurry• Identity (collectivism vs. individuality) • The cultural background of the subject does seem to moderate the wideness of the discrepancy (2 out of 3 significant) • The cultural background of the boss does not seem to moderate the wideness of the discrepancy • The interaction between the two (subject, boss) does seem to moderate the wideness of the discrepancy, however again the effect is difficult to interpret.
ConclusionsContinued • Gender (feminine vs. masculine • The cultural background of the subject does not seem to moderate the wideness of the discrepancy . • The cultural background of the boss does not seem to moderate the wideness of the discrepancy (1 out of 3). • The interaction between the two does seem to moderate the wideness of the discrepancy (1 out of 3)
Implications• Power Distance and Identity seem to moderate the discrepancy between the self and the boss rating in MRF, especially the interaction between the cultural backgrounds of the two.• This is in line with past research, where these two are seen as having the most impact (also these two are by far the most studied).• How this interaction works needs to be clarified, further studied, however it does confirm our concerns the usage and interpretation of MRF results worldwide in more or less the same way.
Next steps1. Expand the data set + with the different rater groups (peers, reports).2. Expand with other cultural dimensions3. Analyse the moderating role of the organisational culture (vs national culture).4. Analyse the moderating role of culture with respect to the used competences (what is actually assessed).