Within minimal group paradigm (MGP) studies, group members make decisions about the distribution of valued resources such as money or points to anonymous ingroup and outgroup individuals. The distribution of such resources is usually made using allocation options known as the “Tajfel Matrices” (Tajfel et al., 1971). These matrices are used to assess how much group members are “tempted” by contrasting orientations such as parity (P), ingroup favoritism (FAV), maximum differentiation (MD), and maximum joint profit (MJP).
The first matrices used by Tajfel et al. (1971) opposed the strategy of ingroup favoritism (FAV = MIP + MD) against the more economically rational strategy of maximum joint profit (MJP; pull of FAV on MJP; pull of MJP on FAV). Note that ingroup favoritism (FAV) is a combination of two strategies: maximum ingroup profit (MIP) and maximum differentiation (MD). Another matrix is designed to pit maximum differentiation (MD) against a combination of maximum ingroup profit (MIP) and maximum joint profit (MJP; pull of MD on MIP +MJP; pull of MIP +MJP on MD).
In this case, respondents choosing the MD option do so at the cost of sacrificing absolute ingroup gain (MIP and MJP) for the sake of achieving a maximum differentiation between the ingroup and outgroup outcome, this difference being in favor of the ingroup.
In another matrix, Billig and Tajfel (1973) pitted the parity (P) option (equal number of points to each group) against ingroup favoritism (pull of P on FAV; pull of FAV on P).
By comparing each allocation response on these classic matrices, distinct “pull scores”
were derived which represented the relative strengths of the different distribution strate- gies adopted by group members. Note that within each matrix, the option to choose the parity (P) response was ALWAYS available. The advantage of using the Tajfel matrices lies in the fact that one can measure the strength of different types of discrimination strate- gies (MD, FAV on P, FAV on MJP) independently of more socially desirable strategies such as parity (P) and maximum joint profit (MJP). Methodological, statistical, and scaling issues related to the use of the Tajfel matrices were discussed by Brown, Tajfel, and Turner (1980) while Bourhis, Sachdev, and Gagnon (1994) provided a step-by-step guide to construct and calculate “pull scores” from the Tajfel matrices.
Both theory and data suggest that social orientations measured using “pull scores” based on the Tajfel matrices do provide a “convenient and representative description of the actual
distribution strategies” employed by group members in laboratory and field intergroup sit- uations (Brown et al., 1980, p. 409). Post-experimental questionnaires have been used in various MGP studies to verify the congruence between participants’ self-reported and actual use of the distribution strategies measured by the Tajfel matrices. For instance, we conducted correlations between actual use of distribution strategies on the Tajfel matrices and self-reports obtained in post-session questionnaire items for the following MGP studies: Gagnon and Bourhis (1996) n =116; Gagnon and Bourhis (1997) n = 470;
Rabbie, Schot, and Visser (1989) n=131; and Sachdev and Bourhis (1985) n=200. For computation purposes correlations were weighted as a function of the number of respon- dents in each study. Combining the above studies, results showed positive and significant correlations between actual use of distribution strategies and the allocation motives expressed by subjects in the questionnaires for parity: P on FAV: r=.44; and for the three discrimination strategies: FAV on MJP: r=.51; FAV on P: r=.46; MD on MIP +MJP:
r=.46. However, correlations between actual allocations and reported use were not sig- nificant for maximum joint profit: MJP on FAV: r=.08 and the MIP +MJP on MD strat- egy:r=.07. Thus group members correctly report their use of key strategies such as parity and discrimination while they tend to systematically overestimate their use of the maximum joint profit (MJP) relative to the virtual absence of this strategy in actual use.
The Tajfel matrices have been adapted successfully for use in various laboratory and field settings. The Tajfel matrices have been modified to measure: performance evalua- tions (Sachdev & Bourhis, 1987); real-life salary increases and salary cuts (Bourhis &
Hill, 1982); teachers’ allocation of financial resources to rival labor federations (Bourhis, Gagnon, & Cole, 1997); the distribution of punishments such as obnoxious noise and unpleasant tasks (Otten, Mummendey, & Blanz, 1996); the allocation of additional course credits for participation in experiments (Bourhis, 1994a) and the allocation of sweets by children using three-column matrices presented as dominoes (Vaughan, Tajfel,
& Williams, 1981). The diversity of these measures shows that the Tajfel matrices can be adapted to suit the valued resources of contrasting group members in different types of intergroup laboratory and field settings.
The classic minimal group discrimination effect is a robust phenomenon which can be monitored using evaluative and resource allocation measures other than the Tajfel matrices. As Brown et al. (1980) stated “. . . the minimal group paradigm is defined solely in terms of the independent variable, social categorization per se. It is not defined by dependent variables or response techniques and strategies” (p. 400). Resource allocation measures other than the Tajfel matrices have included the distribution of a fixed sum of money (or points) between ingroup and outgroup members using a “zero-sum” alloca- tion rule (Ng, 1981; Perreault & Bourhis, 1999) and the “free-choice” distribution of up to 100 points to ingroup members and of up to another 100 points to outgroup members (Locksley, Oritz, & Hepburn, 1980). Binary choice matrices have been used by Brewer and Silver (1978) while multiple alternative matrices (MAM) were proposed by Born- stein, Crum, Wittenbraker, Harring, Insko, & Thibaut (1983a,b). Researchers using the Tajfel matrices also proposed that discrimination could be measured more simply by using a “difference score” between points given to ingroup members over those assigned to out- group members (rather than the more complicated “pull scores”; Diehl, 1990; Platow, Harley, Hunter, Hanning, Shave, & O’Connel, 1997).
The above measurement alternatives tend to provide parity and discrimination options which are not as diversified and subtle as the strategies monitored using “pull scores” based on the Tajfel matrices. For instance, measures of ingroup favoritism using “difference scores” calculated from ingroup/outgroup allocations (free choice, zero sum, Tajfel matrices) cannot distinguish orientations such as maximum ingroup profit (MIP) and maximum differentiation (MD). Furthermore, such “difference scores” do not reveal whether ingroup favoritism was achieved at the cost of maximum ingroup profit (MIP) or maximum joint profit (MJP). Moreover, “difference scores” cannot distinguish when an equal distribution of resources between ingroup and outgroup members (ingroup - outgroup=zero) actually reveals a systematic strategy of parity (P) or the use of maximum joint profit (MJP) or minimum joint benefit (MJB).
Perhaps because of the initial lack of clear information on how to construct and score the Tajfel matrices, the measures were the focus of a methodological and conceptual debate in the 1980s. For instance, one debate centered on whether group members used mainly “pure” allocation strategies (Bornstein et al., 1983a,b) or were instead tempted by a combinations of both single strategies and compromises between alternative allocation strategies (Turner, 1983a,b). For Bornstein et al. (1983a), “the basic assumption of the multiple alternative matrices (MAM) is an outcome maximization assumption . . . that the preferred orientation will have a relatively high frequency of choice, and the remain- ing alternatives will have low or approximately equal frequencies of choice” (pp.
370–371). Using the MAM, group members were instructed to choose a single alloca- tion option out of those presented in each matrix. The allocation orientations offered as choices on each matrix always included the following options: Parity (P), ingroup favoritism (MIP, MD), maximum joint profit (MJPi, MJPo), and outgroup favoritism (OF). Matrices offering these same options were presented numerous times to the respon- dents, each matrix differing simply by the actual numbers portraying each orientation.
For instance, the same parity option could be presented across five different matrices as:
34/34, 35/35, 36/36, 37/37, and 38/38. Repeated presentation of the matrices allowed respondents to consistently choose the same allocation orientation on each matrix or to choose different allocation options from one matrix to the other. Results obtained with MAM measures showed that only 22% to 49% of respondents limited their allocations choice to a unique distribution orientation (Bornstein et al., 1983a, table 8, p. 338).
These results showed that the majority of respondents either had difficulty opting for only one strategy at a time, or preferred to compromise between pure orientations, a mode of resource distribution more in line with the premises of the Tajfel matrices. The latter assume that the measurement of resource allocations is similar to the measurement of attitudes representing continuous variables which on each Tajfel matrix can include com- binations of ingroup favoritism, parity, outgroup favoritism, and maximum joint profit (Turner, 1983a). Results obtained with the Tajfel matrices demonstrate that group members do prefer to compromise between such orientations while they rarely opt for unique orientations across sets of matrices.
Though not all issues of the debate have been settled, it remains that much empirical evidence suggests that the Tajfel matrices do monitor subjects’ social orientations in a valid, reliable, and sensitive manner (Brewer, 1979; Diehl, 1990; Messick & Mackie, 1989; Rabbie et al., 1989; Turner, 1980, 1983a,b). Consequently, and as noted by
Messick and Mackie (1989), it remains fruitful to devote attention to the conceptual and economic underpinnings of the Tajfel matrices, a theme which is the focus of the second part of this chapter.