A Cross-Cultural Comparison of GSS and Non-GSS ... - CiteSeerX

Proceedings of the 29th Annual Hawaii lntemationul Conference on System Sciences -

I996

A Cross-Cultural Comparison of GSS and Non-GSS Consensus and Satisfaction Levels W ithin and Between the U.S. and Mexico Roberto Mejias College of Business Administration, University of Oklahoma: Norman, Oklahoma 73019 Litva Lazeneo , Alvaro Rico, Ana Torres Instituto Tecnologico y de Estudious Superioresde Monterrey (ITESM);Monterrey, Nuevo Leon, Mexico Douglas Vogel Karl Eller Graduate School of Management,University of Arizona:Tucson, Arizona, 85721 Morgan Shepherd College of Business,University of Colorado, Colorado Springs, Colorado, 80933

Abstract Relatively few studies have considered cultural dimensions in their analysis of Group Support Systems (GSS) and even fewer have employed empirical data to test their hypotheses. A cross-culturalfleld experiment was used to measure the effects of “‘national culture” (U.S. andMexican) upon group consensus levels and individual satisfaction levels in GSS environments and non-G%’ environments. Experimental results indicate that within the U.S. culture, there were no signiJicant dtfferences in consensus levels between GSS and Manual groups in the ranking of ideas, but that U.S. GSS groups reported greater changes in consensus levels. Results within the Mexican culture, reported higher ranking consensus levels for manual groups, but greater changes in consensus leveb for GSSsupported groups. With regard to satisfaction levels, while U.S. groups reported no dfferences between treatments, Mexican GSS groups reported higher satisfaction levels than Mexican manual groups. While U.S. groups reported no differences in perceived participation equity, Mexican GSS groups reported higher participation equity than Mexican manual groups. There were no dfferences in perceived participation equity between GSS and non-GSS groups reported within either the U.S. or Mexico. However, in comparing U.S. with Mexican groups there were signtficant differences in satisfaction and perceivedparticipation equity between both cultures across all experimental treatments.

Introduction Modem organizations are increasingly confronted with a competitive and often hostile businessenvironment. Corporate executives,particularlyfrom multinational corporations (MNCs), are often faced with the complex task of integrating their operations, frequently across nationally and culturally diverse environments. Managers and employees are subsequently spending an increasing portion of their work day (30%~80%) attendinga broad spectrumof meetings and working with a wide rangeof cross-functional,multi-national and cross-cultural work

1060-3425196$5.00 0 1996 IEEE

408

Proceedings of the 1996 Hawaii International Conference on System Sciences (HICSS-29) 1060-3425/96 $10.00 © 1996 IEEE

groups (Mintzberg, 1983). Making the most effective and eflicient use of these group meetings has clearly, become a high priority for executives. Organizations have increasingly looked to the information technologies (IT), particularly electronic support systems such as Group Support Systems (GSS), to maximize the effectivenessand quality of group meetings. Globally, GSS have increasingly been used by industrialized nations and to a lesser degree,by developing nations to support numerous decision making activities. However, the conceptual and theoreticalfoundationssupportingthe applicability of current GSS designs have based upon North American concepts of “desirable”group goalsand “desirable”group behavior. The vast majority of empirical research regarding GSS meeting technology and its effects on group decision-making has been confined to U.S. subjects within U.S. organizations and envinmments.Few studieshave consideredcultural dimensions in their analysis of electronic support systemsand even fewer have employed empirical data to test their hypotheses. In view of the sign&ant developmentsin information technology and the proliferationof telecommunicationsystemson a global basis, this has been a disappointing shortcoming of current IT research. This studyis a cross-cultural field experiment that measured the effects of “national culture” (U.S. and Mexican) upon consensuslevels and satisfaction levels in environments using computer mediated support (Group Support Systems)and nonGSS normal environments. The field study was conducted at two research sites: the University of Arizona, in Tucson, Arizona representing the U.S. national culture and ITESM (Institute Tecnologico y de Estudios Superiores de Monterrey) in Monterrey, Nuevo Leon, Mexico representing the Mexican nationalculture. The study included an analysis of consensus and satisfaction levelsboth within and between cultures as respective groups were exposed to three experimental treatments (GSSAnonymous, GSS-Identified and Manual-Identified).

Background In an empirical study of 53 countries, which included over 116,000respondents,Hofstede’s (1980,199 1) model of cultural

Proceedings of the 29th Annual Hawaii International Conference on System Sciences -

differentiation described four dimensions of national culture alougwhich valuesystemsmay vary. Thesespecific dimensions were termed Power-Distance, Uncertainty-Avoidance, Individualism-Cvism, and Masculinity-Femininity. Power Distance describestherelativedistanceand relationshipbetween a supervisorand a subordinate. Power distanceis the extent to which a particular national culture acceptsand recognizesthe unequal distribution of power and influence in institutions and organizations. Countries that score high on power distance appearto emphasizeautocraticor paternalistic,while countries that scorelow on power distancefavor participative management relationsandthe use of “equal rights” and legitimatepower over coercivepower. Uncertainty Avoidance is the extent to which membersof a particular national culture feel uncomfortableor threatenedby uncertain or unknown outcomes(Hofstede, 1991). Countries that scoredhigh on the UncertaintyAvoidance dimensiontended to havea low tolerancefor uncertaintya greaterneed for formal rules. Additionally, countries with strong (high) Uncertainty Avoidance scoresalso had less tolerancefor people or groups with deviant ideas or behavior and were more likely to resist innovative ideas (Hofstede, 1991). Countrieswith weak or low Uncertainty Avoidance scoreswere inclined to take more risks andweremore likely to tolerate deviant behavior and innovative ideaswhen making group decisions(Hofstede,1980). Individualistic-Collectivistic refers to the relationship between the individual and his or her larger organization. Individualistic countries are cultures which are self-reliant and are expectedto look a&r only themselvesand their immediate families; relationshipsor links betweenoutsideindividuals are usuallynot very strong. Collectivistic countriesare cultureswith a preference for cohesive and tightly knit social frameworks. Collectivistic cultures strive to maintain harmony and avoid co&ontation or disagreementamonggroup members(Watson, Ho, Raman; 1994). The MasculinityFemininity dimensionindicatesthe relative trade-off betweenan assertiveenvironmentversusa supportive or nurturing environment. High Masculinity-Femininityscores indicate an emphasis on power, assertivenessand individual achievement. Cultural systems with low scores on the Masculinity-Femininitydimensionindicate a greateremphasison people,qualityof life, providing nurturing support and collective cooperation.

Prior GSS research Nearly all the empirical research in GSS literature haa adopted an input-output perspective and compared group outcomesof GSS groupswith traditional face-to-faceor manual groups (Ho, Raman, Watson; 1994). Traditional manual interactive techniques such as the Delphi method and the Nominal Group techniquehave been shown to improve group performance with idea generation and decision making tasks

1996

(Delbecq, Van de Ven, and Gustafson,1975) while the addition of a discussionleaderor “facilitator” has been shown to improve theperformanceof some groupswith “intellective”tasks (Maier and Solem, 1952). However, while these manual, interactive grouptechniqueshavebeenshownto be us&d, past researchhas repeatedlyshownthat GSS technologyhas tremendouspotential for improvinggroup performanceand productivity (Valacich, et al., in press). Since the exchange of information among participantsis the principle activity in group meetings,the main iuteresttoresearchers hasbeenwhether or not electronic support tech.nology (e.g., GSS) cau improve group meeting communication and enhanceface-to-face groups to be more productive and efficient. Prior GSSliterature,however has traditionally been directed at understandinghow new information technologiesare ditfused within and acrossU.S. firms (Straub, 1994). Few studieshave beendirectedat learninghow firms select,adopt and utilize new IT within cross-cultural environmentsand even fewer studies have advanceda theoretical perspective on possible cultural effects(Straub, 1994). The moderating effect of culture and nationality upon the behavior and attitudes of employeesin foreign operativeshas bcenwell establishedby many researchersin early crosscultural studieson psychologicalneed perceptionsand need satisfactions ((Porter, 1963; Zurcher, 1963; Richman, 1965; Zurcher, Meadow and Zurchcr, 1965; Simonetti and Weitz, 1967; IvancevichandBaker, 1969; Slocum, 1971; Mejiaa, 1980). For example,in a comparativestudy of American managerslocated in the United Statesand American managerslocatedin Europe, IvancevichandBaker(1969) found that culture had a significant inIlucnce even ifthe respondentwas not of that culture. Prior researchhas found that cultural di&rcnces between Asiatic and Western-English speaking cultures can be pronounced(Ho, Raman,Watson; 1989). For example,while U.S companiesexploit the advantagesof cmail, Japanesefirms do not use email as extensively. The Japanese,however, do utilize FAX more often than U.S. firms (Straub, 1994). The differences between these contrasting uses of IT may be attributed to cultural differences (Straub, 1994). Some researchers believethatpreferencefor even the “type”of analysis may differ amongdi&rent cultures(Sauter, 1992). For example, Europeans and Japaneseseem to prefer more “analytical” solutions and normative models while Americans prefer “heuristics” and descriptive models (Watson et al., 1994). Clearly, if differencesin design preferencesfor certain models exist among different cultures, one would expect observable diiliin the designof auy electronic support systemor GSS within crosscultural environments.

Research hypotheses Basedupontheforegoingdiscussionand Hofstede’smodel of cultural differentiation, we postulated a set of hypothesesthat

409

Proceedings of the 1996 Hawaii International Conference on System Sciences (HICSS-29) 1060-3425/96 $10.00 © 1996 IEEE

Proceedings of the 29th Annual Hawaii International Conference on System Sciences- 1996 soughtto predictthe behaviors of U.S. and Mexican groups both within their respective cultures and compared across both cultures. Since the experimental study was divided into two major areas of analysis, (i.e., the analysis of GSS vs non-GSS performanceand perceptionswithin each national culture and a comparisonof GSS performance and perceptions betweenU.S. and Mexican groups), two groups of hypothesesare proposed.

in stating that the presence of consensusis fine, insofar as it represents an open and genuine reconciliation of differences. Clearly, the role and relationship of consensus in socially interactive environments, like group meetings, are significantly d&rent for collectivistic (e.g., Mexico) and individualistic (e.g., U.S.) cultures. In general, there seems to be mixed research findings regarding group consensus levels and relative changes in Hypothesis Group A: (Analysis WHIZ Each Culture) comensuslevels.Several studies state that GSS technologymay actually reduce consensuslevels (Gallupe, DeSanctis, Dickson, Hypothesis A-l: The levelsof consensusgeneratedwith regard 1988). Still other studies report no change in consensuslevels to the ranking of idea categories will be higher in non-GSS (Watson, DeSanctis, Poole, 1988). Only a few studies (Turloff (Manual) groups than in GSS groups within both U.S. and and Hiltz, 1982; Stevik, 1987) report that GSS technology increasedgroupconsensuslevels. It has been suggestedthat the Mexican national cultures. anonymityfeatureof GSS removesthe identity of group members Hypothesis A-2 - The change in the levels of consensuswith regard to the ranking of idea categorieswill be higher in non- from their respective input and therefore, reduces the level of GSS groups than in GSS groups within both U.S. and Mexican commitment (Javenpaaet al. 1988). On the whole, empirical studies seem to be consistent with the findings of lower national cultures. Hypothesis A-3 - Participants from GSS groups will express consensus levels for GSS-assisted groups than for non-GSS more satisfaction with their groun’s decisions than participants groups (Ho et. al, 1989). It is therefore, predicted that GSS from non-GSS groups within both U.S. and Mexican national groups across both U.S. and Mexican cultures would generate lower comznms levels thannon-GSS(i.e., manual) groups as the cultures. Hypothesis A-4 - Participants corn GSS groups will express level of identity proportionately decreased. Specifically, as the more particinationeauity with the GSS processthan participants identity level of the group participants decreasesTom Manual from non-GSS groups within both U.S. and Mexican national (Identified) to GSS (Identified) to GSS (Anonymous), the level of group consensuswould be correspondingly lower. cultures, With regard to changes in consensuslevels, since the role and relationship of consensus is different between high Hypothesis Group B: (Analysis Comparing Cultures) “individualistic” countries like the U.S. and high “collectivistic” Hypothesis B-l- Mexican GSS groups, acrossall experimental (or low “individualistic”) countries like Mexico (Hofstede, 1980, 1991),instructinga groupto reach consensusmay be interpreted treatments,(GSS and non-GSS) will experiencehigher levels of diierently within each society (Watson et al., 1994). Clearly, consensusthan U.S. GSS groups. GSS and its supportingdecision making technology,reducesthe Hypothesis B-2 - The change in the levels of consensuswith regard to the ranking of idea categories will be higher for traditional social cues found in conventional meetings and thereby increasessocial distance among members which may Mexican GSS groups than for U.S. GSS groups. lead to a lower levels of group consensus(Jarvenpaa,Rao and Hypothesis B-3 - Participants from Mexican GSS groups will expresshigher levels of satisfactionwith their groun’s decisions Huber, 1988). Consequently, both U.S. and Mexican GSS groupsmay have fewer social cues and interactive opportunities than participants from U.S GSS groups. that arefnzquentlyutihzedby face-to-facemanual groups to sway Hypothesis B-4 - Participants from Mexican GSS groups will expressmore particination eauitv with the overall GSS process or pemuademembersto “reconsider”their position (e.g. ranking) and possibly change their vote (e.g. second ranking) on some than participants from U.S. GSS groups. Hypothesis A-l and A-2 are based upon researchfindings topic (e.g.ranking of idea categories). As a result, it is predicted from previous GSS experimental research. The concept of that electronically unaided groups (e.g., Manual groups), will consensus may be interpreted differently within dissimilar havethe ability to generategreaterchanges in ranking consensus cultural settings. For example, in high “collectivistic” countries, than electronically aided groups (e.g., GSS groups). Although therehasbeen sparseresearchon GSS satisfaction like Singapore consensus is given a high priority to further national goals (Watson, Ho, Raman, 1994). Mexicans, like andperceptionlevelsin a cross-culturalsetting, the GSS features Singaporeans may view dissent and non-consensusas something of equalandparallel processing,anonymouscommunication, and to be minimized so that the host country may be viewed as structured group processes are predicated to have similar politically and socially stable and thereby, attractive to foreign positive perception results in the Mexican national culture as work groups have previously experienced in the U.S. investment (Watson et al., 1994). On the other hand, in high “individualistic” societieslike the Specifically, groups using GSS technology within each national United States,Watson et al.,( 1994) cites U.S. SenatorFulbright culturewould be predictedto be more satisfied with their group’s

410

Proceedings of the 1996 Hawaii International Conference on System Sciences (HICSS-29) 1060-3425/96 $10.00 © 1996 IEEE

Proceedings of the 29th Annual Hawaii International

decision@&pothesisA-3) and perceive more equal participation (Hypothesis A-4) within GSS supported environments than traditional manual technology. The predictions for Hypotheses B-l, B-2, B-3, and B-4, which compareconsensuslevels and perceptions levels between different national cultures are based upon the sparse crosscultural GSS researchpresentedto date and Hofstede’smodel of cultural differentiation which depicts Latin American countries, (e.g.,Mexico) as ideologicalopposites to the U.S. culture. Since the U.S. is characterizedby high individualism and low power distance, U.S. individuals tend to see themselves as “I” and strive for self-actualization (Watson, Ho, Raman, 1994). The Mexican culture is characterizedby low individualism (or “high collectivism”) and high power distance and see themselvesas part of “we”and will strive for belongingnessand the good of the w interest (Hofstede, 1980, 1991). This almost bipolar difference in cultural dimensions result in less cultural pressure on U.S. groups to reach agreementbecauseU.S. society places a highervalue on individualism. Subsequently,Mexican groups as a whole, were predicted to generatehigher group consensus levels than U.S. groups as a whole. With regard to the comparison of changes in consensus levels between the U.S. and Mexico, as the Mexican culture possessesa higher “collectivistic” cultural index than the U.S. culture, there would be more pressure on Mexican groups to reach agreementor at least achieve a higher level of consensus than U.S. groups. Additionally, since GSS technology removes some of the social cues and therefore reduces or neutralizes some of the “high power distance”effects existent within the Mexican culture, we predicted that Mexican groups would, generategreater changesin consensuslevels than U.S. groups. With regard to satisfaction and perception, since Mexicans generatea possessa higher ‘collectivism” score than the U.S. and are more concernedwith making a decision for the good of the collective group, it was predicted that not only would Mexicans generate a higher consensus level than an individualistic culture (e.g., U.S.), but that Mexican participants may express more satisfaction with their group’s decision that their U.S. GSS counterparts. Additionally, the “collectivistic” group-supporting features of GSS technology will reduce many of the high power distance barriers that have been traditionally present in face-to-face meeting such that Mexican will experience and perceive more “participation equity” than U.S. groups, who normally take “equal participation” in most group setting for granted.

Research methodology The research study was composed of two separate, but closely interrelated areas of analysis. The first area of analysis investigated the effects of GSS supported meetings upon group performance and individual perceptions within the Mexican national culture and within the U.S. national culture. Each

Conference on System Sciences -

1996

national experiment therefore, was a 2 x 2 “within subjects” factorial design study where supporting technology (GSS Vs. non-GSS)and identification features (anonymousVs. identified) constituted the two independentvariables. The second area of analysis represented a cross-cultural comparisonof GSS technologyupon group consensuslevels and perceptions between the U.S. and Mexican national cultures. This areaof analysistherefore,constituted a 2 x 2 x 2 factorial design where supporting technology (GSS Vs. non-GSS), identification features (anonymous Vs. identified), and national culture (U.S. & Mexican) represented the three independent variablesin this cross-cultural field experiment. The particular emphasis however, throughout this research study was on national culture and its relative effect on group performance and the perceptions of their respective group members. Ideally, this 2 x 2 x 2 design would create eight cells, but becauseof the perceived artificiality of an “anonymous communication” treatmentwithin the Manual group treatment, two (2) of the cells were not consideredappropriate or realistic for experimentation. Specifically, most organizations would not employ an “anonymoustlgenerationof ideas and an “anonymous”discussion of their viewpoints within a traditional face-to-face meeting environment. Therefore, this left a total of 6 experimental treatment cells in the final researchdesign. Research sites

Several Mexican research sites were considered for the currentcross-culturalfield experiment. After severalfield visits and considerations of several potential sites, the Institute Technologic~y de Estudious Superioresde Monterrey (ITESM) in Monterrey, Nuevo Leon, Mexico proved to be the best match for the U.S. site (University of Arizona, Tucson, Arizona) in terms of similar academic programs, relative emphasis on research and teaching and similar equipped electronic meeting rooms. Additionally, the ITESM campus at Monterrey used the identical DOS version of Ventana GroupSystemssoftware that was used at the University of Arizona researchsite. The use of identical groupware software was especially critical so that diflkences recorded within eachculture and acrossboth cultures would could not be attributed to differences in software interfaces. Subjects

The U.S. national culture sample consisted of 22 work groups, or 230 upper division production management and business administration undergraduate college students. The Mexican nationalculturesampleconsisted of 20 work groups, or 239 upper division business administration and industrial engineering college students. Both national samples were divided into three experimental treatments: GSS-Anonymous, GSS-Identified, and Manual-Identified. There were 6-7 groups

411

Proceedings of the 1996 Hawaii International Conference on System Sciences (HICSS-29) 1060-3425/96 $10.00 © 1996 IEEE

Proceedings of the 29th Annual Hawaii International

for eachof the threeexperimentaltreatments,for each culture, for a total of 42 groups. Task and experimental

procedures

The field experiment across both cultures was composed of seven (7) identical phases for group participants from each national culture (e.g., U.S. and Mexican) 1. Explanation and Purpose of Experiment to Participants 2. Idea Brainstorming 3. Categorization of Brainstormed Ideas into “Idea Categories’ 4. 1st Ranking of Idea Categoriesin Order of Importance 5. Discussion of 1st Ranking 6.2nd Ranking of Categories 7. Completion of Perceptions Questionnaire Before the experimental session actually was underway, facilitators at both research sites explained the nature and purposeof the experiment The experimental task questionswas displayedvia a preparedoverheadtransparencyor by writing the “task question”on a white board. Additionally, three examples of appropriate ideas or answers to the task question were also given to the group along with the explanation that there were really no “right” or “wrong” ideas that could be submitted. Depending on the what experimental treatment a group was assigned,participantswere askedto either submit their real name (for the GSS-Identified and Manual treatments) or “anonymous pseudonym” (for the GSS-Anonymous) for tracking and matching purposes. Participants will then asked to submit (either manually or electronically)as many ideas or comments as possible in an idea “brainstorming”type fashion for exactly 15 minutes. During the idea brainstorming session,facilitators simultaneously (i.e., “on the fly”) createda list of “idea categories”of similar ideas. After the 15 minute brainstorming session, the group participants reviewed the list of “idea categories”generatedby the group facilitator andwere askedif the idea list accuratelyreflected their ideasand comments.Participantswere allowed to add categories to the list (in the event that the facilitator may have omitted some ideas),combinesimilar categories,and/or correct the wording of the idea categorieslist. Both GSS-supportedand Manual group members were then senta “ballot”with the list of the categories generatedduring the session. Participants were then asked to rank, from “most important” to “least important”, each of the idea categories. All ranking for all experimental treatment groups was done electronically, even for the manual groups. After all group members had submitted their individual rankings, a group “concordance level” statistic (e.g. 35%, 41.5%, 62.8%) was generated by the supporting software along with a graphic distributionof individual rankings for each of the idea categories and was displayed on a large screen in front of the group. The facilitator then earmarkedfor discussion those particular idea categories whose standard deviation was high with the purposeof encouragingthe group to come to a higher consensus

1996

or level of agreement in their rankings of various categories. Group participants were allowed to raise their hands and argue why someidea category should or should not have been ranked as highly/lowly as was displayed on the screen. After ten minutesof verbal discussion,group members were sent a second ballot with the same list of idea categories that they had previouslyranked before, (but now, in random order) and asked to again rank, in order of importance, the list of idea categories, based on the past discussion. After ranking, the group then viewed the results along with a new higher or lower group consensus level. After l-2 minutes of discussion, each group members was asked to fill out a post-test questionnaire (GSS Group Perceptions Questionnaire) that asked them to rate their individual satisfaction with the group process and various other group process gains and process losses. Dependent Variables

and Measures

There were four (4) dependent variables for this crosscultural field experiment. Level of Consensus, was the degree of group agreement or “concordance”among group members regarding the relative ranking of importance of the various ideas or idea categories generatedby the group. Change in Level of Consensus, is the degree of change in the level of concordancefi-om the group’s first ranking to the second ranking of the idea categories. The relative change in rankings was recorded by the supporting software in each of the experimental rooms. Satisfaction with the Group’s Decision, is the satisfaction of group members with the final rankings of the idea categories. Satisfaction levels were recorded from a 63 item perceptions questionnairethat used a 1 to 7 Likert Scale. Participation Gquity is the opportunity for equal participation among group members in a meeting. Perceived participation equity was recorded from a 63 item perceptions questionnaire that used a 1 to 7 Likert Scale.

412

Proceedings of the 1996 Hawaii International Conference on System Sciences (HICSS-29) 1060-3425/96 $10.00 © 1996 IEEE

Conference on System Sciences -

Experimental results The results from this study were divided into two sections: the analysis within each national culture (Hypothesis Group A) and the analysis between or comparing cultures (Hypothesis Group B). For both analysis,a significant level of 0.05 was used for hypothesis testing. Detailed tables of results are available from the lead author. For the analysis comparing results within each national culture, t tests and one-factor ANOVAS for the three experimental levels in this field study (GSS-Anonymous, GSSIdentitled,Manual-Identified)were used to test the hypothesesin

Group A. The one-factorANOVA helpedto deternineif observed differences between the means of the dependent variables were due to the particular combination effects (i.e.,

Proceedings of the 29th Annual Hawaii International

Conference on System Sciences -

1996

interesting main effect for the three experimental treatmentsin the study. The experimental data indicates that as the level of identity of the individuals within each experimental treatment increased(from electronic anonymousto manual identified) the level of rankingconsensusincreased.This trend held true across both national cultures for both rankings. Perception - With regard to perception levels, the experimental data indicates highly signitkant differences betweenU.S. and Mexican groups due to the main effects of the culturevariable. Mexican groups acrossnearly all experimental treatments,mcordedsignificantlyhigher levels of satisfaction and participation equity than U.S. groups (p < 0.001). Mexican groups reported signiticantly higher levels of satisfaction with their group’s decision than U.S. groups across all three experimental treatments (p