Back to Journal Cover Page


Volume 6, Number 7
Submitted: November 21, 2000
Resubmitted: March 7, 2001
Accepted: March 19, 2001
Publication date: March 20, 2001


Hein F. M. Lodewijkx
Utrecht University, The Netherlands

J. E. M. M. Syroit
Utrecht University, The Netherlands


Dutch students who want to join a student organization have to go through lengthy and demanding initiations, lasting three to four weeks, in order to become fully accepted members. The present study compares the longitudinal data collected among the initiated newcomers of two such organizations; one organization imposing a severe initiation, the other a more mild one. These different initiation practices constitute naturalistic severe and mild initiation conditions, enabling us to examine the validity of two hypotheses proposed to account for the expected increase in group attractiveness after newcomers have endured a severe initiation. (1) Aronson and Mills’ (1959) severity-attraction hypothesis (SAH), that builds upon cognitive dissonance theory. (2) A severity-affiliation-attraction hypothesis (SAAH), which assumes that the proposed increase in group attractiveness is due to affiliation-related processes under the threatening circumstances of an initiation (Schachter, 1959). Results revealed no support for the SAH or the SAAH. Findings suggest the existence of a mere initiation-affiliation-attraction relationship: Irrespective of their severity, initiations seem to induce a common fate, increasing newcomers’ attraction to the group through the process of short-term, affiliate exchanges.



Scarce research has been conducted to investigate the socialization and initiation practices used by real groups of various kind to permit newcomers entrance into their groups (Dornbusch, 1955; Vaught & Smith, 1980; see Moreland & Levine, 1989, for a review). Of particular interest in this field is the phenomenon that group entrance is often made very hard for the newcomers. Newcomers have to endure lengthy, demanding and demeaning practices in order to become full and accepted members (Brown, 2000). Counter-intuitively, in such circumstances the relationship seems to be: The more extreme and degrading newcomers' experiences, the more attracted to the group they seem to become. In the social psychological literature this relationship is better known as the severity-attraction hypothesis (SAH, Aronson & Mills, 1959), that builds upon Festinger’s (1957) cognitive dissonance theory. Succinctly, it can be summarized as follows: "If I am willing to endure all this, I must really like the group."

To our knowledge the SAH has hitherto only been investigated in the laboratory, but not in naturalistic settings. The present study aims to fill this gap and directly compares the effects of naturalistic severe and mild group initiations on group attractiveness and related variables. The study further examines a so-called severity-affiliation-attraction hypothesis (SAAH), derived from the early work of Schachter (1959) on affiliation under threat, exchange theory (Homans, 1968), and the health research by Rook (1987) on companionship. A more detailed presentation of both hypotheses will follow in later sections.

It should be noted that the present research uses the data of two longitudinal field studies (Lodewijkx & Akkersdijk, 1995; Lodewijkx & Syroit, 1997), separately investigating the merits of the SAH and the SAAH. The studies were conducted among the newcomers of two different Dutch student organizations, one organization imposing a severe initiation, the other utilizing a more mild initiation. The studies did not support the SAH, but found evidence corroborating the SAAH, raising doubts about the validity of the SAH. However, in both studies only correlational data were presented, and the studies were not compared with each other. Correlational data do not only compromise causality, they also do not give any information concerning the degree to which naturalistic severe and mild initiations affect newcomers’ level of group attractiveness at the end of the initiation. Only a direct comparison between the two naturalistic settings would provide an answer to this question. Therefore, in the present study severe and mild initiations are directly compared with each other in an effort to assess their effects upon group attractiveness and related variables. In this sense the present study clearly extends the aforementioned field studies. Before moving on to an outline of the SAH and the SAAH, however, a short description will be given of the proceedings of the initiation practices of the two student organizations in which the studies were conducted. These descriptions will perhaps help the reader to value the data presented and discussed here.



Initiation rituals can be considered as a ‘rite of passage’ (Van Gennep, 1977), defined as rites which accompany every change of place, state, social position and age. The change from high school to college or university is an example of such a passage. In the initiations imposed by the two Dutch student organization several stages can be discerned, resembling the stages distinguished by Van Gennep (1977) in his seminal cultural anthropological work: separation, transition, and incorporation.

The ‘Severe Initiation’ Student Organization
During the first, separation stage, some two-hundred female newcomers are literally separated from their familiar environment and are obliged to live together at a campsite in a rural area for about a week, surrounded by fellow group members and a few senior members who are in charge. No males are allowed. Each newcomer is given a different name, which they carry throughout the initiation. All newcomers have to wear the same bag-like clothes. The newcomers live in tents in groups, and every day the group composition is changed deliberately by those in charge. No clocks are present at the site and watches are forbidden. Newcomers hardly get enough sleep and they are subjected to hard work every day (digging ditches, pulling trees); to roll-calls that can last from 1 hour up to 4 hours in day- or nighttime; and they are given hardly enough food or drink. In physical terms, newcomers suffer hardship.

After this, the transition stage follows, which lasts about one and a half week. The newcomers leave the campsite and return to the city. They get acquainted with the norms and values of the student organization, with the senior members and with student life and all kinds of activities are programmed. During this stage the so-called 'evening gatherings' take place, that are regarded as a menace by the newcomers, because during these meetings they all run the risk of being ‘hazed’, bullied and embarrassed in front of everyone.

Next follows the incorporation stage during which the newcomers are officially inaugurated as members of the ingroup. The inauguration takes place at the office building of the organization. All newcomers are celebrated, have a meal together and party with the senior members during the so-called 'integration party'.


The ‘Mild Initiation’ Student Organization
This organization consists of male as well as female members and they are initiated together. Politically, this organization has a more progressive / liberal signature than the former one, and, historically, a religious background. The organization emphasizes the personal growth of its members. The initiation practices used are much less official, and are less controlled, compared to the more ‘severe’ organization. For example, in this ‘mild’ organization the use of alcohol and cannabis is allowed, whereas this is strictly forbidden in the ‘severe’ organization. Moreover, anonymity is less prominent in this ‘mild’ organization: The newcomers keep their own name, their own clothes and watches. The size of the group is also much smaller (n = 47). In this organization Van Gennep' s stages could not be distinguished as easily compared to the more ‘severe’ organization. Again, the newcomers were separated from the outside world for about week and, again, they stayed at a campsite. During this period, however, newcomers also made acquaintance with the senior members. So, the separation and transition periods are integrated here. Importantly, no threatening ‘evening gatherings’ took place during this stage. After the separation / transition stage the newcomers returned to the city, and after a week they were inaugurated in the presence of the senior members at the office buildings. The whole initiation took two and a half weeks.

Preliminary unstructured interviews held among several students of the two organizations suggested the differences described above; yet, these descriptions remain anecdotal, lacking independent verification. We also emphasize that there are many differences between the two student organizations we cannot control for: i.e. the size and the composition of the groups involved, the duration of the initiation, and the rather conservative versus progressive / liberal norms and values newcomers should incorporate in order to become fully accepted members. We are aware that these are all potentially confounding variables, compromising the internal validity of the present study. However, as we will try to show, the consistency in results between the present study, related field studies (Lodewijkx & Akkersdijk, 1995; Lodewijkx & Syroit, 1997) and some of the laboratory studies (Schopler & Bateson, 1962) -- all conducted to critically examine the SAH -- seem to suggest that these shortcomings are less damaging to the validity of the study than might be assumed.


The Severity-Attraction Hypothesis (SAH)
Imagine that, during an initiation, group members are ordered by those in charge to clean a small bridge over a canal with a toothbrush (as happened a few years ago in the more ‘severe’ organization). While doing this difficult job, several of the group members might find some of the others a nuisance, perhaps because of having the feeling that they are trying to take a free ride on them. According to cognitive dissonance theory (Festinger, 1957) in such circumstances cognitive dissonance is aroused in the newcomers once they realize that the group of which they are a part is not as positive and pleasant as they anticipated it to be. In this case two elements are dissonant to each other: The voluntary initiation and the negative experiences with the group. Because dissonance is hypothesized to be an aversive motivational state, the new members will try to reduce it. According to Aronson and Mills’ (1959) SAH newcomers will try to reduce this aversive state by residing to cognitive mechanisms, that is, by "overestimating the attractiveness of the group" (ibid. p.180).

The scarce experimental research aimed at examining the SAH was conducted about forty years ago (Aronson & Mills, 1959; Schopler & Bateson, 1962; Gerard & Mathewson, 1966). With the exception of the Schopler and Bateson (1962) study, these studies did not check for the degree of dissonance allegedly aroused in the motivated participants by imposing an embarrassing treatment upon them. As already noted, cognitive dissonance is an inferred motivational state, which has aversive consequences for those who suffer from it. Importantly, cognitive dissonance cannot be measured directly.

In the two recent field studies, alluded to above, Lodewijkx and Syroit indirectly tried to assess the distressing state newcomers are supposedly in, when experiencing dissonance. They argued that self-reported measures of negative affect --more specifically, depressive mood-- might be regarded as an indirect measure of the dissonance aroused in the newcomers during naturalistic initiations. Both these studies, however, showed reliable, negative correlations between newcomers’ depressive mood and group attractiveness: The more depressed newcomers’ mood, the less likely they were to rate the group favorably. This unsuspected finding raised doubts concerning the validity of the SAH. However, because both studies utilized correlational data, the question whether severe or mild initiations do differentially affect newcomers’ depressive mood and level of group attractiveness, could not be answered. Only a direct comparison of the two naturalistic settings would provide an answer to this question. Therefore this comparison is made in the current study. In this comparison, the SAH straightforwardly predicts that newcomers in a severe initiation condition will increase the attractiveness of the group compared to newcomers in a mild initiation condition (H. 1), providing of course that dissonance is aroused in the newcomers by the negative experiences with the group. If depressive mood constitutes an indirect measure of experienced dissonance, as suggested by Lodewijkx and Syroit (1997), this means that this variable can be considered a mediating variable (Baron & Kenny, 1986) in the severity-attraction relationship. This implies that the predicted severe / mild main effect on group attractiveness (as stated in H. 1) should disappear when statistically controlling for newcomers’ depressive mood (H. 2).

The Severity-Affiliation-Attraction Hypothesis (SAAH)
In his classic affiliation studies Schachter (1959) has shown that individuals, when facing threatening or stressful situations, strongly tend to seek the safe company of other individuals. This affiliate need is most prominent when the individuals have gone through a similar predicament and share the same emotional experience. Although there may be many reasons for this affiliate need, Schachter contended that self-evaluative social comparisons were the main reason. When facing threat or danger, people often do not know how to react, or how to label their emotions. According to Schachter (1959), self-evaluative social comparisons with similar others will help people to reduce and cognitively label their emotional uncertainty. Building upon this theory, and assuming that initiations can be considered threatening, Lodewijkx and Syroit


(1997) proposed that the severity-attraction relationship could therefore also be mediated by affiliation-related processes. Noteworthy, this affiliation approach to severe initiations was based on Aronson and Mills’ own analyses. As these researchers acknowledged themselves, affiliate exchanges among the initiated newcomers were considered to be an important variable to increase group attractiveness. Specifically, participants in their study were not allowed to affiliate in order "to counteract any tendency to identify more strongly with the group as a result of feelings of having shared a common unpleasant experience" (ibid. p. 178). So, in their view as well as in ours, affiliation-related processes might be an important factor in naturalistic group initiations to foster feelings of group belongingness and to increase the attractiveness of the group.

On these grounds, Lodewijkx and Syroit (1997) argued that the need for affiliation during threatening naturalistic initiations would be reflected in short-term, rewarding, spontaneous interactions or companionate exchanges amongst the newcomers (Rook, 1987; Buunk & Verhoeven, 1991; Buunk & Peeters, 1994; Peeters, 1994). During these exchanges contact with others is sought voluntarily with the only intention and goal to have some pleasure. These companionate exchanges show themselves in various ways, for example, in having a few laughs with another person, in making jokes and chatting, or in talking about personal things like hobbies or sports. In line with this approach, findings of the Lodewijkx and Syroit field studies indeed showed small but reliable positive correlations between newcomers' self-reported affiliate exchanges and group attractiveness. A severity-affiliation-attraction hypothesis (SAAH) can be derived from this perspective, predicting that, first, a severe initiation will lead to more companionate exchanges amongst the newcomers than a mild initiation (H. 3). Second, it predicts that companionship will mediate the severity-attraction relationship, such that the expected severe / mild main effect on group attractiveness (as stated in H. 1) will disappear when statistically controlling for companionship (H. 4).

As already outlined above, the Lodewijkx and Syroit field studies did not include a direct severe versus mild comparison of newcomers’ experienced level of companionship, thereby precluding an assessment of the validity of H. 3 and H. 4. The design of the present study allows for such an assessment. If support is found for H. 3 and H. 4, there is some indication that a severe initiation constitutes a sufficient condition to activate affiliate exchanges among the newcomers, whereby the attractiveness of the group is increased.



Hypotheses were tested using a 3 (Type of initiation: Severe 1 versus Severe 2 versus Mild) X 2 (Initiation stages: t1 versus t2) repeated measurement design. The three-level 'Type of initiation' between-subjects factor was constructed in the following way. Severe initiation conditions were created by employing the data from the Lodewijkx and Syroit (1997, Study 1) field study conducted among the newcomers of the ‘severe initiation’ student organization (see also Lodewijkx & Akkersdijk, 1995). Two random subsamples with n = 46 were drawn from the total of 202 novices of that organization (the n is comparable to the total n obtained in the 'mild' student organization, see below). We drew two subsamples to examine the reliability of the findings between the two 'severe' subsamples. The type of initiation factor was completed by adding to it the data of the newcomers of the student organization whose newcomers went through a more mild initiation (n = 46; Lodewijkx & Syroit, 1997, Study 2).

Regarding the 'initiation stage' within-subjects factor, points of measurement were chosen on the basis of the different stages that could be distinguished during the initiations. The first point of measurement (t1) was chosen when newcomers were staying at the campsite. The second point of measurement (t2) was chosen one to two weeks after the initiation was terminated, but shortly before newcomers were officially accepted as full members of their organization (that is, shortly before the integration or incorporation parties took place).

The sample of the ‘severe’ student organization consisted of 220 female newcomers (age 18-23 years). Eight of the newcomers cancelled their registration and ten decided to leave the campsite in the first week. A total number of 202 novices, 91% of the total sample, participated in the research. From these 202 students two random subsamples were drawn (each n = 46). In the more mild student organization, 47 male and female students enrolled as members, who together went through the initiation. Forty-six newcomers filled out the questionnaires. Twenty were male, 26 were female. Mean age was 21.5 years. In this study, no reliable gender differences were observed on all dependent variables, allowing us to collapse the data across gender, and consequently, to compare the student organizations with each other.


For each of the measurements in both studies newcomers were provided with labels on which a private code number was printed, thereby preserving anonymity. When asked to check the questionnaire, newcomers were reminded of their labels and, on request, stuck the labels to the questionnaire. At t1 all newcomers gathered in tents at the campsites. The questionnaires were administered and collected by female researchers when newcomers were staying at the campsites for three to five days. The posttest at t2 was administered one to two weeks after the end of the initiation. At t2 all the newcomers gathered in the office buildings of the student organizations. Questionnaires were administered as the newcomers entered the office building, where female researchers waited for them. After filling out the questionnaires, the researchers immediately collected them. At t1 and t2, newcomers were requested to check the questionnaire individually and care was taken that they complied with this request.

Assessments and Data Reduction
The DIRO, a Dutch translation (Buunk & Verhoeven, 1991; Buunk & Peeters, 1994; Peeters, 1994) of the Rochester Interaction Record (RIR; Wheeler, Reiz, & Nezlek, 1983; Cutrona, 1986) was used as the main instrument. It was chosen because the DIRO / RIR allows the analysis of individuals' affective reactions to stressful experiences and degree of experienced companionship in various group settings. The operationalizations of companionship and the mood checklist (see below) were both derived from the DIRO. Attractiveness and severity measures are of our own making. All questionnaire items were checked by five-point scales (1 = not at all, 5 = very much), except for some open-ended items with which the severity measures were operationalized (see below). Separate principal component analyses (PCAs), using varimax rotation, were performed at t1 and t2 to reduce the data. All extracted components had eigenvalues greater than 1. By aggregating the items, composite measures were constructed for all scales consisting of more than one variable. Descriptive statistics are summarized in Table 1.

Asking newcomers to rate their liking for the group at t1 and t2 assessed group attractiveness (e.g. "I belong to the group"; "I feel at home in the group"; "In enjoy the group"; 10 items). PCA extracted one factor 'Group attractiveness' at both points of measurement with high internal consistency. Companionship was operationalized by three items at t1 and t2 (e.g. "I often stop by and have a chat with someone after which I feel better"). The three items showed high loadings on the factor companionship at t1 and t2 separately. Cronbach's alphas for companionship were 0.61 and 0.63 respectively. Newcomers' moods were assessed by a mood checklist at t1 and t2 (14 items, e.g., "angry, lonely, relaxed, relieved, frustrated, irritated, elated, tired, inhibited"). PCA yielded similar one-factor solutions for the two time periods, labeled 'Depressive Mood', with high internal consistency.


Table 1. Descriptive statistics, reliability coefficients and intercorrelations for dependent variables.










(1) Attractiveness (t1)


(2) Attractiveness (t2)

61*** 88            

(3) Companionship (t1)

36*** 22* 61          

(4) Companionship (t2)

22* 24* 52*** 63        

(5) Depressive Mood (t1)

-53*** -34*** 09 15 90      

(6) Depressive Mood (t2)

-43*** -39*** 09 10 70*** 89    

(7) Severity (t1)

-09 -06 06 18 37*** 28 -  

(8) Severity (t2)

-13 -04 06 07 19 24   -

* p < .05; ** p < .01; *** ; p < .001.
Note. Cronbach’s alphas are on the diagonal. Decimal points omitted for alpha- and correlation coefficients. Dfs = 76 / 94 due to missing values.

Severity was measured at t1 and t2 in two ways. (i) Through four open-ended questions newcomers checked which experiences they believed to be unpleasant, two of which referred to the past, the other two to anticipated experiences. (ii) Per open-ended question newcomers rated the level of unpleasantness on five-point scales. At t1, during the campsite stage, severity items were filled out that referred to past and anticipated experiences. At t2, the posttest, severity ratings only referred to past experiences. In the current study analyses were performed using the item on which the newcomers checked their first mentioned negative experience in the past (also because the number of missing values on the remaining severity items were too many). Mean severity ratings were high at both time periods (see Table 1). Pearson correlations computed between the single severity item and the depressive mood scale were positive and reliable for t1, r(94) = .37, p < .001, and for t2, r(94) = .28, p < .006. Similarly, the correlation between the severity rating at t2 and depressive mood at t2 was also positive and reliable, r(76) = .24, p < .04. These correlations indicate that the more severe newcomers' experiences became, the more depressed their moods tended to become. These findings indicate that newcomers' severity of experiences was validly measured.



Table 2 shows mean ratings of the dependent variables, broken down by type of initiation and initiation stage. All hypotheses were tested by performing 3 (Type of initiation: Severe 1 versus Severe 2 versus Mild) X 2 (Initiation stage: t1 versus t2) repeated measures, mixed ANOVAs, with the dependent variables measured at the different time periods as the within-subjects factor.

As a ‘check of the manipulation’ in a first analysis we examined whether the severe / mild conditions affected newcomers' severity ratings in the expected way. ANOVA revealed a main effect for type of initiation, F(2, 77) = 9.17, p < .001, and an interaction effect, F(2, 77) = 3.89, p < .025. The within-subjects factor did not reach significance, F < 1. Consistent with the severe / mild conditions Table 2 shows that the severe conditions indeed gave rise to more extreme severity ratings (M sample 1 = 4.03; M sample 2 = 4.28) relative to the mild initiation condition (M = 3.56). Comparisons further revealed that the two severe conditions did not differ significantly from each other (p < .20), while both severe conditions differed significantly from the mild initiation condition (ps < .03 by Tukey HSD using harmonic means). However, these results are qualified by the significant interaction effect. This interaction effect was further explored by testing the within-condition mean squares of the within-participant effect against the overall error term of the analysis, using overall degrees of freedom (df = 1, 77). The analyses showed that, within both severe initiation conditions, there is a decrease in the level of experienced severity from t1 to t2, Fs(1, 77) > 3.98, ps < .05, whereas there is a small but unreliable effect in the other direction in the mild initiation condition, F(1, 77) = 2.41, p > .10. These observations indicate that, first, we were successful in finding naturalistic severe and mild initiation conditions, and second, that newcomers' experiences were most unpleasant in the severe conditions, in particular when newcomers endured the trials and tribulations during the campsite period at t1. These findings allow us to continue hypothesis testing.

Table 2. Mean Ratings of Dependent Variables broken down by Severe and Mild Initiation Conditions and Initiation Stage (t1 / t2).


Severe Initiation
Sample 1

Severe Initiation
Sample 2

Mild Initiation

  t1 t2 t1 t2 t1 t2

Severity of Experiences

4.20 3.85 4.42 4.14 3.35 3.77

Group Attractiveness

3.78 3.96 3.89 4.16 3.99 4.12

Depressive Mood

3.11 3.18 3.09 2.98 2.35 2.34


4.19 3.93 4.26 4.19 3.92 3.74

Note. The higher the ratings, the more severe newcomers’ experiences, the stronger the perceived attractiveness of the group etc. (lowest = 1; highest = 5). Dfs of the ANOVAs vary due to missing values.


To test the SAH (H.1), attractiveness measures were included as the within-subjects variable in a 3 X 2 repeated measures ANOVA. The analysis only yielded a main effect for initiation stage, F(1, 111) = 18.30, p < .001. Neither the main effect for type of initiation F(2, 111) = 1.41, p < .25, nor the interaction effect, F < 1, was statistically reliable. These results do not support the SAH, since no main effect was obtained for the severe / mild initiation conditions on group attractiveness. As for the significant initiation stage effect, the means in Table 2 show that group attractiveness is highest at the end of the initiation (M t2 = 4.09) compared to the campsite stage (M t1 = 3.90). The increase in group attractiveness from t1 to t2 is observed within the severe and the mild initiation conditions separately and, because the interaction effect was not reliable, appears to occur independently of the severe or mild initiation practices newcomers were subjected to.

ANOVA performed on the depressive mood scale yielded a main effect for type of initiation, F(2, 106) = 47.37, p < .0001. The main effect for initiation stage, F < 1, and the interaction effect, F(2, 106) = 1.71, p < .19, were not statistically reliable. Mean depressive mood ratings in Table 2 show that newcomers in the severe conditions score higher on this variable (M sample 1 = 3.14; M sample 2 = 3.03) compared to newcomers in the mild initiation condition (M = 2.34). Comparisons further showed that the two severe conditions did not differ significantly from each other (p < .44), while both severe conditions deviated significantly from the mild condition (ps < .001). These results thus corroborate the main effect for type of initiation obtained on the severity ratings. Results concerning the mediating role of depressive mood in the severity-attraction relationship (H. 2) will be presented below.

H. 3 expected that a severe initiation would strengthen the affiliate exchanges among the newcomers compared to a mild initiation. The 3 X 2 ANOVA, including the companionship ratings as the repeated measures, revealed main effects for type of initiation, F(2, 111) = 5.97, p < .003, and initiation stage, F(1, 111) = 7.39, p < .008. The interaction effect was not significant, F < 1. Table 2 shows that newcomers in the more severe initiation conditions (M sample 1 = 4.06, M sample 2 = 4.23) have higher ratings on the companionship measure compared to the mild initiation condition (M = 3.83). Again, the means of the severe conditions did not differ significantly from each other (p < .39), and again, both deviated significantly from the mild initiation condition (ps < .05). The initiation stage main effect indicates that newcomer score lower on companionship at t2 (M = 3.94) compared to t1 (M = 4.11). This latter result is plausible. It indicates that newcomers have less companionate exchanges after the initiation has ended when the group was more fragmented than during the initiation, when, as a whole group, all newcomers together stayed at the campsite.


Mediation Analyses
Hypotheses 2 and 4 aimed at gaining more insight into the interrelationships between the variables under study and mediating relationships between them were proposed. According to Baron and Kenny (1986) three conditions must be met in order to be able to conclude that such relationships exist. (1) A predictor variable (i.e. the severe / mild independent variable) should show a significant relationship with a dependent variable (i.e. group attractiveness). (2) It should also have a significant relationship with a mediating variable (i.e. depressive mood and companionship). (3) The mediating variable should show a significant relationship with the dependent variable. With respect to the first condition, as noted above, our data do not meet this requirement, because no severe / mild main effect was obtained on group attractiveness. The rationale to perform mediating analyses is therefore not warranted in the present case. As for the second condition, results showed significant effects of the severe / mild independent variable on the proposed mediating variables depressive mood and companionship. Regarding the third condition, correlations were computed between the relevant variables, revealing that for some of the variables (i.e. the severity ratings), the data did not meet Baron and Kenny's third condition, while for other variables (i.e. companionship) they did.

To illustrate this, contrary to the severity-attraction hypothesis, severity ratings did not correlate significantly with group attractiveness at t1, r(94) = -.09, p = .37, or at t2, r(94) = -.06, p = .58. A similar pattern was observed for the severity rating at t2 with group attractiveness at that point of measurement, r(76) = -.04, p = .71. The companionship measure a t1, however, did show positive correlations with group attractiveness at t1 r(94) = .36, p < .001, and at t2, r(94) = .22, p < .03, and the same holds for companionship measured at t2 with attractiveness at t2, r(94) = .24, p < .02. Thus, as expected, some of the variation in group attractiveness at t2 is accounted for by affiliation- related processes, R2 = .081, p < .01. Correlations further revealed no significant relationships between newcomers' self-reported severity of experiences and companionship at t1, r(94) = .06, p = .54, or at t2, r(76) = .007, p = .96. The same pattern can be seen for the correlations between depressive mood and companionship at t1, r(94) = .09, p = .40, and t2, r(94) = .10, p = .34. Depressive mood at t1, however, shows negative correlations with group attractiveness at t1, r(94) = -.53 and t2 r(94) = -.34 (ps < .001). The same holds for depressive mood at t2 with attractiveness at t2, r(94) = -.39, p < .001.



Results indicated that the severe initiation conditions differed from the mild initiation condition in the degree of self-reported severity of experiences, depressive mood and companionate exchanges. On all these dependent variables newcomers in the severe conditions had higher ratings than newcomers in the mild condition, while no reliable differences were obtained between the two severe conditions. The latter findings contribute to the reliability and validity of the results. Importantly however, the severe conditions did not differ from the mild condition on the most important variable in the severity-attraction relationship: Group attractiveness. This hypothesis would predict higher ratings on group attractiveness in the severe conditions, because in these conditions newcomers’ experiences were significantly more unpleasant and depressing. Thus, the present results show no evidence for Aronson and Mills’ severity-attraction hypothesis (SAH).

Admittedly, the same is true for our severity-affiliation-attraction hypothesis (SAAH). Similar to the previous argument, the severe conditions differed significantly from the mild condition on companionship, indicating stronger affiliate exchanges in the severe conditions compared to the mild condition. Still, no reliable differences were observed on group attractiveness. Results concerning the SAAH are thus inconclusive.

There are many differences between the laboratory studies, examining the SAH, and the field studies. These differences were summarized by Lodewijkx and Syroit (1997) and relate to: (1) Newcomers’ expectations that they are facing a demanding experience or not; (2) the sizes and composition of the groups involved; (3) the duration of the initiation; (4) the degree and kind of severity of the experiences; (5) the voluntary nature of group entrance; (6) the commitment of the newcomers to their choice of student organization; (7) the motivation to become a member; and finally (8) the rather ‘closed’ nature of the student organizations, meaning that if the newcomers leave the group, they cannot become members of the organization in the same year. It is a difficult task to determine which of the many variables will contribute to the hypothesized greater group attractiveness in our naturalistic settings, apart from the severe or milder treatment of the newcomers.

Methodologically, we cannot control for the influence of these potentially confounding variables on group attractiveness, and this can be considered a weakness of the study. On the other hand, such drawbacks are inherently part of this kind of research. This, however, does not mean to say that therefore such research should not be undertaken, simply because there is always the chance that the findings offer opportunities to arrive at new insights on a theoretical level. We argue that this the case in the present study. First, because the pattern of results emerging from the former field studies and the present one is remarkably consistent and statistically reliable. Second, because these results are clearly in line with findings of the Schopler and Bateson (1962) experimental study, which was conducted to critically investigate the SAH. Contrary to this hypothesis, these researchers found that participants in the severe initiation condition who felt most embarrassed --and who therefore should have experienced the strongest cognitive dissonance-- rated the group less favorably compared to participants who felt less embarrassed. According to the SAH the more embarrassed participants should have rated the group more favorably.


When combining all these observations, two general conclusions can be drawn. (1) In initiation situations negative affect decreases group attractiveness. This decrease occurs irrespective of whether the negative affect is experienced by newcomers during naturalistic group initiations in terms of depressive mood, or by participants in a controlled laboratory situation in terms of feelings of embarrassment. (2) In initiation situations there might be a straightforward initiation-affiliation-attraction relationship. That is, independently of whether newcomers are treated severely or not, initiations seem to increase newcomers’ attraction to the group through the process of short-term, companionate exchanges. The reasons for drawing this second general conclusion will be outlined below.

As we have seen, companionship does significantly account for some 8 per cent of the variation in group attractiveness. Two observations further substantiate this conclusion. One, there are no significant relationships whatsoever between affiliate exchanges on one hand and severity of experiences and depressive mood on the other. Second, extent of companionship varied with initiation stage, with stronger companionship observed during the campsite stage, when all members had the opportunity to meet and exchange rewarding interactions amongst each other, and with lowered companionship at the last stage, where such opportunities were less abundant. These observations suggest that initiations, irrespective of their severity, might activate affiliate exchanges amongst the newcomers, whereby the liking for the group becomes stronger. The finding that group attractiveness increased from t1 to t2, whereas simultaneously companionship decreased from t1 to t2, does not need to undermine this conclusion. In particular because at both points of measurement, the zero-order correlations between companionship and group attractiveness were all positive and statistically reliable (mean r = .26, p < .05).

The significant severe / mild main effect obtained on companionship further supports this proposed initiation-affiliation-attraction relationship. The absence of any correlation between newcomers’ severity of experiences / depressive mood and level of companionship suggests that this severe / mild main effect can be attributed to the specific circumstances associated with the two different student organizations, but not to newcomers’ actual experiences. In his affiliation studies, Schachter (1959) convincingly showed that participants’ anticipation that they would receive a strong electric shock was already sufficient to induce a strong need for affiliation in these participants. This occurred even in conditions where they could not directly interact and communicate with other participants. We argue that the same anticipatory process also applies to the present field studies, in particular where the severe organization is concerned. At the campus students talk about their initiation experiences, often in negative terms. At times, the media extensively cover these initiations, especially when extreme incidents have occurred (like the death of a student through alcohol abuse, as happened a few years ago in Groningen, The Netherlands). Thus, newcomers, who want to join a student organization that imposes a


severe initiation, already anticipate that they have to go through some trying times in order to become accepted members. Similar to Schachter’ s affiliation studies, this anticipated common fate may have induced a strong need amongst the newcomers to seek the safe company of similar others even before the start of the initiation. In this sense, the inducement of this affiliate need may even occur irrespective of whether newcomers are treated harshly or not while being initiated; the anticipation will be sufficient. This ‘anticipatory’ affiliate need induction might explain the severe / mild main effect obtained on companionship in the present study. The same process might also be the reason why no reliable relationships were obtained between newcomers’ self-reported experiences of severity and the affiliate exchanges they encountered.


Aronson, E. & Mills, J. (1959) The effects of severity of initiation on liking for a group. Journal of Abnormal and Social Psychology, 59, 177-181.

Brown, R. (2000) Group processes (2nd Edition). Oxford, UK: Blackwell Publishers.

Buunk, A. P. & Verhoeven, K. (1991). Companionship and support in organizations: A microanalysis of the stress-reducing features of social interaction. Basic and Applied Social Psychology, 12, 243-258

Buunk, A. P. & Peeters, M. C. W. (1994). Stress at work, social support and companionship: toward an event-contingent recording approach. Work and Stress, 8, 177-190.

Cutrona, C. E. (1986). Behavioral manifestations of social support: A micro-analytic investigation. Journal of Personality and Social Psychology, 51, 201-208.

Dornbusch, S. M. (1955). The military academy as an assimilating institution. Social Forces, 33, 316-321.

Festinger, L. (1957). A theory of cognitive dissonance. Stanford, Ca.: Stanford University Press.

Gerard, H. B. & Mathewson, G. C. (1966). The effects of severity of initiation on liking for a group: a replication. Journal of Experimental Social Psychology, 2, 278-287.

Homans, G. C. (1968). Human behavior: Its elementary forms. New York: Harcourt, Brace, Jovanovitch.


Lodewijkx, H. F. M. & Akkersdijk, N. L. (1995). Ontgroening en groepsattractiviteit: Affiliatiebehoefte of dissonantiereductie (In Dutch). [Initiation and liking for the group: Need for affiliation or dissonance reduction]. In P. van Lange, F. Siero, B. Verplanken, & E. van Schie (eds.), Sociale Psychologie en haar toepassingen (Vol. VII, pp. 55-68). Delft: Eburon.

Lodewijkx, H. F. M. & Syroit, J. E. M. M. (1997). Severity of initiation revisited: Does severity of initiation increase attractiveness in real groups? European Journal of Social Psychology, 27, 275-300.

Moreland, R. L. & Levine J. M. (1989). Newcomers and oldtimers in small groups. In P. B. Paulus (ed.), The Psychology of Group Influence (pp. 143-186). Hillsdale, NJ: Lawrence Erlbaum Associates.

Peeters, M. C. W. (1994). Supportive interactions and stressful events at work: An event-recording approach. Doctoral dissertation, Katholieke Universiteit Nijmegen, The Netherlands.

Rook, K. S. (1987). Social support versus companionship: effects on life stress, loneliness and evaluations by others. Journal of Personality and Social Psychology, 52, 1132-1147.

Schachter, S. (1959). The psychology of affiliation. Stanford, Ca.: Stanford University Press.

Schopler, J. & Bateson, N. (1962). A dependence interpretation of the effects of a severe initiation. Journal of Personality, 30, 633-649.

Van Gennep, A. (1977) The rites of passage. London: Routledge & Kegan Paul Ltd.

Vaught, C. & Smith, D. L. (1980). Incorporation and mechanical solidarity in an underground coal mine. Sociology of Work and Occupations, 7, 159-187.

Wheeler, L., Reiz, H. T., & Nezlek, J. (1983). Loneliness, social interaction and sex roles. Journal of Personality and Social Psychology, 45, 943-953.



The authors are greatly indebted to the old-timers and the leadership of the two student organizations for conducting their research amongst their newcomers.


Hein Lodewijkx received his Ph.D. in social psychology from Utrecht University, The Netherlands. He is currently an assistant professor teaching group processes and research methodology. His research interests include intergroup conflict and aggression, group initiation, group creativity and productivity. Address correspondence to Hein Lodewijkx, Department of Social and Organizational Psychology, Utrecht University, PO Box 80140, 3508 TC, Utrecht, The Netherlands. Phone: 31.30.2534937. Fax: 31.30.2537584. E-mail: or

Jef Syroit received his Ph.D. in organizational psychology from Tilburg University, The Netherlands. He is currently an associate professor at Utrecht University, The Netherlands, teaching personnel psychology and organizational psychology. His research interests include intergroup relations and justice-related problems in organizational settings. Address correspondence to Jef Syroit, Department of Social and Organizational Psychology, Utrecht University, PO Box 80140, 3508 TC, Utrecht, The Netherlands. Phone: 31.30.2533460. Fax: 31.30.2537584. E-mail:


Back to Journal Cover Page