Journal of Occupational and Organi:(ational Psychology (1999), 72, 25-40

Printed in Great Britain

25

© 1999 The British Psychological Society

The validity of collective climates Vicente Gonzalez-Roma*, Jose M. Peiro, Susana Lloret and Ana Zornoza Departamento de Metodologia, Psicobiologia y Psicologia Social, Universidad de Valencia, Spain

The objective of this study is to test the validity of the collective climate concept. It was expected that membership in collective climates was related to membership in the collectivities defined by departmental membership, hierarchical level, shift, job location and organizational tenure. The study sample was composed of 195 employees from a central administration agency. Using a combination of hierarchical and non-hierarchical clustering methods, three different collective climates were obtained. The results showed that only hierarchical level was related to collective climate membership. Based on all the results obtained, the debate on the validity of collective climates is reconsidered, and research on climate formation and relevant factors in that process is emphasized.

At the individual level, it can be stated that 'essentially, climate is individual descriptions of the social setting or context of which the person is a part' (Rousseau, 1988, p. 140). These individual descriptions are C2i\^&d psychological climate (James, 1982; Joyce & Slocum, 1984; Rousseau, 1988). Aggregation of psychological climates has heen used to represent the climate of larger units of analysis (teamwork, department, organization), hut prior to aggregation to a higher level of analysis agreement among individuals' climate perceptions (that is, psychological climates) must he demonstrated (James, 1982; Joyce & Slocum, 1984; Roherts, Hulin & Rousseau, 1978). Previous research has shown that this criterion has not often heen achieved for different suborganizational aggregates such as department, hierarchical level and work-team (Joyce & Slocum, 1984; Patterson, Payne & West, 1996). To overcome this problem, Joyce & Slocum (1984) proposed looking for clusters of organizational members with similar climate perceptions and then aggregating their individual climate scores. This procedure would warrant agreement among climate perceptions of individuals in the same cluster. These clusters of people are called collective climates, and are determined by means of statistical clustering methods Qackofsky & Slocum, 1988; James, 1982; Joyce & Slocum, 1984). James and his colleagues have advocated the usefulness of the collective climates concept, since 'aggregate climate perceptions may provide a powerful explanatory and predictive tool' (James, 1982, p. 221), and 'it allows for the •Requests for reprints should be addressed to Vicente Gonzalez-Roma, Departamento de Metodologia, Psicobiologia y Psicologia Social, Facultad de Psicologia, Universidad de Valencia, Av. Blasco Ibanez, 21, 46010-Valencia, Spain.

26

Vicente Gonn^dle^Roma et al.

description of organizational settings in psychological terms' (Joyce & Slocum, 1984, p. 722). Researching the validity of collective climates was the purpose of Joyce & Slocum's (1984) study. In doing so, these authors studied whether collective climates met the following criteria: (1) discrimination, or demonstrable differences between mean perceptions between climates; (2) predictable relationships to organizational or individual criteria (such as performance and job satisfaction); and (3) internal consistency, or agreement in perceptions within collective climates. Joyce & Slocum (1984) studied the relationships between collective climates (formed from perceptual agreement on six dimensions: rewards, autonomy, motivation to achieve, management insensitivity, closeness of supervision and peer relations) and five facets of job satisfaction and job performance in a sample of first-line foremen of three plants of a heavy-duty truck manufacturer. After determining different collecdve climates in each plant, they found that membership in a collective climate was significandy related to job satisfaction in plants 2 (except for satisfaction with pay) and 3, but not in plant 1 (except for satisfaction with promotions), and to job performance (in plants 1 and 2). They also studied the relationship between individual antecedents (time in position, work experience, management experience, age, education and salary) and organizational antecedents (function supervised, location, shift and leader initiating structure and consideration) of collective climates and membership in the collective climates obtained. They found that physical location was related to climate membership in plant 2 (^<.O6). In plant 3, function supervised was associated to climate membership (/)<,07). In both of these plants, the initiating structure and consideration behaviours of the foreman's immediate supervisor were strongly associated with membership in a collective climate. In contrast, in plant 1 climate membership was only associated with individual variables: job tenure, work experience and age. Joyce & Slocum (1984, p. 736) concluded that their research 'provided evidence that supports the validity of collective climates as one type of aggregate climate. Multiple collective climates were obtained within three different work settings, and membership in these climates was related to performance and job satisfaction.' In another study, Jackofsky & Slocum (1988) investigated the relationships between collective climates (formed from perceptual agreement on seven dimensions: supervisory style, task characteristics, coworkers, work motivation, employee competence, decision making, performance rewards) and intrinsic and extrinsic job satisfaction, job performance, leader reward behaviour (positive and negative), intention to quit, turnover, departmental assignment and the following individual variables: sex, age, marital status, hotel experience, education level and past employment in other hotels. The study sample was composed of 63 non-managerial employees at a newly established location of a major hotel chain. Measurements were obtained at two times. They found a significant reladonship between departmental assignment and collective climate membership at time 1, but not at time 2. There were no significant differences between the collective climates, at either time, regarding the individual variables studied. The results obtained by Jackofsky & Slocum (1988) also showed that membership in a collective climate was related to intrinsic job satisfaction and negative leader reward behaviours only

The validity of collective climates at time 1, and to extrinsic job satisfaction, intention to quit and positive leader reward behaviours at both time points. No associations were found between collective climates and job performance and turnover. Jackofsky & Slocum (1988) concluded that their research 'provided additional evidence that supports the validity of collecdve climates as an important phenomenon. Agreement on climate decisions was related to job sadsfacdon, leader reward behaviours, and intendons to quit' (pp. 332-333). Payne (1990), commendng on Jackofsky & Slocum's (1988) study, has cridcized the concept and meaning of^ collecdve climates, stadng that 'the important point, however, is not just that people see the organizadon in a similar way, but that similarity is rooted in some formal or informal structured collecdvity. Clusters of people who share common percepdons of their company might arise from such things as having similar personalides, values, interests, etc. To demonstrate that you can find clusters of people with similar views of their organizadon is no more than to demonstrate that clustering techniques work' (p, 78). Certainly, accomplishment of criteria 1 and 3 established by Joyce & Slocum (1984) to study collecdve climates validity are pracdcal consequences of the stadstical methods employed, so they should not be used with Joyce & Slocum's purpose. From Payne's (1990) view, 'the clusters have to have some sensible socio-psychological idendty if they are to represent a meaningful scientific concept' (p. 78). Payne (1990) suggested that clusters might occur around departments, work-teams, cliques, cabals, trade-union membership, etc. Taking into account this argument, Payne (1990) stated that the conclusions reached by Joyce & Slocum (1984) and Jackofsky & Slocum (1988) regarding the validity of collecdve climates are quite unjusdfied, since 'the individuals could be clustered on age, sex and body weight but the clusters would not have a great deal of conceptual udlity in helping to understand the funcdoning of organizadons' (p. 78). Payne (1990) suggested that Jackofsky and Slocum study the overlap exisdng between collecdve climate membership and departmental membership. Answering Payne's cridcisms, Jackofsky & Slocum (1990) advocated social interacdon as the basis of collecdve climates; that is, their argument was that perceptual agreement among collecdve climate members stems from social interacdon between individuals. Taking into account that the measurement for their study took place approximately 45 and 150 days after the hotel had opened (dme 1 and dme 2, respecdvely), Jackofsky & Slocum (1990) pointed out that 'the finding of a reladonship between perceptual agreement and departmental membership at dme 1 but not at dme 2 provides support for our argument that social interacdon between subjects was a basis for the formadon of collecdve climates in dme 2' (p. 82). From their point of view, only when pardcipants had a greater opportunity to interact across organizadonal boundaries, could perceptual agreement be predicated on social interacdon and not on structure. Recendy, Patterson, Payne & West (1996) invesdgated the socio-psychological significance of collecdve climates by tesdng whether they mapped onto different social groups. In a sample composed of 10 work-teams of a construcdon company which operated independendy from each other and were spadally distant, they invesdgated the relationships between collecdve climate and work-group

27

28

Vicente Gon^dlei^-Romd et al.

membership, regional membership, job type and hierarchical level. No reladonship between the aforemendoned variables was found out. The collecdve climates idendfied were composed of individuals from different social groups. Patterson et al (1996) concluded that 'the collecdve climates appear to have no obvious sociopsychological meaning as many individuals in the same collecdve did not interact with each other. Nor the clusters reflect similar organizadonal experiences stemming from the same job content or hierarchical posidon' (p. 1686). The methodological procedure used to idendfy collecdve climates has also been used to identify organizadonal subcultures. Jermier, Slocum, Fry & Gaines (1991) employed clustering techniques to idendfy the organizadonal subcultures exisdng in a police organizadon. They found that membership in the five clusters they idendfied was associated with structural variables (rank, departmental assignment and shift), individual variables (tenure) and outcome variables (organizadonal commitment and work performance). Although Jermier and colleagues' study was not focused on collecdve climates, it is perdnent for the issue addressed here. Unlike Slocum and colleagues Qackofsky & Slocum, 1988; Joyce & Slocum, 1984), Jermier et al found that cluster membership was associated with membership in the social coUecdvides defined by the structural variables of rank, departmental assignment and shift, and with organizadonal tenure. These results, together with convergence with subcultural forms described by police ethnographers, led Jermier et al. to assert that the clusters they obtained were not stadsdcal ardfacts and that the cluster solution had been successfully validated. Moreover, the aforemendoned associadon also contributed to Jermier et al viewing the idendfied organizadonal subcultures as realides which are socially constructed by interacdng individuals pertaining to the same collecdvity such as the same department or hierarchical level. Turning back to collecdve climates, after reviewing the literature it is clear that the empirical evidence regarding the meaning and udlity of the concept of collecdve climates is sdll scarce. Payne (1990) argued that, if collecdve climates are to represent a meaningful sciendfic concept, perceptual agreement within clusters should be rooted in some formal or informal structured coUecdvides; that is, collecdve climates will 'have to have some sensible socio-psychological idendty' (p. 78), Collecdve climates have been incorporated into the variety of climate types (Rousseau, 1988), but as Patterson et al (1996) have stated 'there is a need to demonstrate that collective climates has a meaningful idendty in terms of informal or formal organizadonal groupings before claiming it represents a valid descripdon of a meaningful social construct' (p. 1678). The main objecdve of the present ardcle is to test the validity of the collecdve climate concept. It will be tested whether collecdve climates membership is related to membership in the collecdvities defined by departmental assignment, hierarchical level, shift and job locadon, and to organizadonal tenure. Several arguments can be provided to justify the expected reladonship between collecdve climate membership and departmental assignment. First, people in the same department share tasks and work processes, Consequendy, practical demands of their environments may become very similar. Moreover, they have common general goals (those of the department). Common perceptions and beliefs may

The validity of collective climates emerge as the result of common demands and interests (Jermier et al., 1991). Second, functional relationships imposed by the work system operating within each department promotes social interaction among department members. Through social interaction, individuals develop shared perceptions of their setting (Ashforth, 1985; James, Joyce & Slocum, 1988). Third, belonging to the same department implies having the same formal leader. Leaders are important agents of socialization who may exert relevant influences on department members' perceptions (Kozlowski & Doherty, 1989). Sharing the same formal leader may contribute to developing common environmental perceptions, since 'the leader serves as an interpretative filter of relevant organizational events, features and processes' (Kozlowski & Doherty, 1989, p. 548). And fourth, different departments may implement distinct socialization tactics. The expected reladonship between collective climates membership and hierarchical level is based on three arguments. First, job requirements and demands vary with hierarchical level. The higher the hierarchical level the more demanding and complex is the job. Individuals within the same level are exposed to similar job demands, and this may be the source of shared perceptions of the environment (Jermier et al, 1991). Second, meetings of managers may become frequent in some organizations (as was the case in the present study). IJnder these circumstances, shared perceptions among managers may develop through the social interaction processes which take place in those meetings. And third, organizations often implement management development training programmes. These programmes are a relevant way of influencing the views of managers. In the study organizadon, the aforemendoned training programmes are implemented. Having the same shift and the same job locadon provides job incumbents an opportunity to interact. In the case of having the same shift, interacdon is facilitated by temporal coincidence in the organizadonal context. When job incumbents have the same job locadons, interacdon is facilitated by spadal proximity. In both cases, shared percepdons of the environment may develop through interacdon processes. Regarding the hypothesized associadon between organizadonal tenure and collecdve climate membership, it is based on the following argument: employees with the same tenure were socialized similarly and have been exposed to the organizadon stimuli and events during the same dme. These two factors may lead to those employees perceiving the organizadon in a similar way. Finally, considering that one of the methodological criteria which has been employed to assess the validity of different types of aggregate climates is that these climates are expected to show stadsdcally significant relationships to organizadonal or individual criteria (Pritchard & Karasick, 1973), it will be tested if membership in collecdve climates is significandy related to employees' job sadsfacdon, job involvement and organizadonal commitment. Method Sample The study sample was obtained from the population of employees who worked in the main building of a central administration agency. This building has six floors and the agency is structured in six

29

30

Vicente Gonsale^Romd et al.

departments. However, there is not an exact overlapping between floors and departments. Members of three departments (A, B and C) are located across two floors (second and fourth, ground floor and third, and first and third, respectively), whereas the members of the three remaining departments (D, E and F) are located within one floor each (third, first and fifth, respectively). A battery of quesdonnaires was distributed among the employees by the agency technical personnel staff. Along with the battery, a letter from the research team guaranteeing anonymity and confidentiality and a letter from the top manager encouraging participation were distributed. The answered questionnaires were returned in a closed envelope to the agency technical personnel staff. One hundred and ninety-five participants responded and returned the questionnaires. The response rate was 43%. Twenty-eight participants did not provide department assignment data, and four presented some missing data on the climate items. These 32 participants were dropped from the data file. Therefore, the study sample size was 163. Of the sample, 57.5% were female. The average age was 35.2 (SD = 7.9). Top managers accounted for 3.7%; this group included the department directors and the provincial director. Eight per cent were section managers, 12.9% subsection managers, 12.3% were team leaders and 63.2% were lower level employees. The average organizational tenure was 10.7 years (SD = 8.5).

Measurement Organizational climate was measured using the questionnaire developed by the FOCUS (First Organizational Climate/Culture Unified Search) research group. This questionnaire measures four dimensions of climate: support orientation: the extent to which there are kindly and supportive relationships among organizational members (8 items); rules orientation: the extent to which organizational members' behaviour is regulated by formal norms and rules (6 items); goals orientation: the extent to which activities and behaviours are oriented towards the attainment of previously established objectives (14 items); and innovation orientation: the extent to which there is openness to new ideas and projects (12 items). Participants were asked to describe, not evaluate, the climate within their organization. This process was introduced 'to maximize the respondent's use of actual experiences as a basis for describing a climate' 0oyce & Slocum, 1984, p. 727). Response to each item is by means of a 6-point Likert scale. The associated consistency reliabilities (Cronbach's alpha) for each climate scale in the study sample were: .73 (support), .61 (rules), .81 (goals) and .79 (innovation). A number of single questions were used to gather information about employees' departmental membership, hierarchical level, shift, job location and organizational tenure. Job location was operationalized by means of the floor location of respondents' jobs. Hierarchical level had five categories (1 = top managers, 2 = section managers, 3 = subsection managers, 4 = team leaders and 5 = lower level employees); while shift presented only two: continuous (only in the morning) and split-shift (in the morning and in the afternoon). Job satisfaction was measured using a 12-item scale which covers different job aspects such as work conditions, supervision, goals and relationships with the organization (Melia & Peiro, 1989). Items were responded to by using a 7-point Likert scale ranging from '1. very dissatisfied' to '7. very satisfied'. The internal consistency computed for this scale (Cronbach's alpha) in the study sample was .88 (Af=4, SD = 1.1). Job involvement was measured by means of Lodhal & Kejner's (1965) 6-item scale. Items were responded to using a 7-point Likert scale ranging from '1. strongly disagree' to '7. strongly agree'. The Cronbach's alpha coefficient for this scale computed in the study sample was .77 {M— 3.5, SD — 1.1). Organizational commitment was measured using the eight items of O'Reilly & Chatman's (1986) scale which measure organizational commitment based on identification and internalization of organizational aims and values. Items were responded to using a 7-point Likert scale ranging from '1. strongly disagree' to '7. strongly agree'. The Cronbach's alpha coefficient was .76 {M— 3.9, SD = 1.2).

Analysis Collective climates were obtained following the general procedure outlined in Joyce & Slocum (1984). Participants were clustered on the basis of profile similarity on the four climate dimensions measured.

The validity of collective climates

31

Both hierarchical and non-hierarchical techniques were used. When using hierarchical techniques, 'the results at succeeding levels of clustering generally are dependent. Consequently, allocation decisions made early in the clustering affect subsequent clusters, and nonoptimum clusters are generated' 0oyce & Slocum, 1984, p. 728). After a hierarchical solution is determined, non-hierarchical methods may be used to improve hierarchical initial clusters to obtain a better solution (Joyce & Slocum, 1984). Initial clusters were determined by graphing the number of hierarchical clusters against the fusion or amalgamation coefficient. This test is analogous to the scree test of factor analysis (Aldenderfer & Blashfield, 1984). 'A marked "flattening" in the resulting graph suggests that no new information is portrayed by the following mergers of clusters' (Aldenderfer & Blashfield, 1984, p. 55). A complementary procedure examines the values of the fusion coefficients to discover a significant 'jump' in the value of the coefficient. 'A jump implies that two relatively dissimilar clusters have been merged; thus the number of clusters prior to the merger is the most probable solution' (Aldenderfer & Blashfield, 1984, p. 57). After a set of initial clusters had been selected in this way, a non-hierarchical k means clustering procedure was used to improve the results. Once an initial partition had been specified, the means (centroids) of the clusters were computed, and Euclidean distances to all cluster means were calculated. Then cases were assigned to the nearest cluster thus reducing the pooled within-group variance. This procedure was repeated until cluster assignments were stable and subsequent iterations of the procedure failed to produce a decrease in the pooled within-cluster variance (Aldenderfer & Blashfield, 1984; Joyce & Slocum, 1984). These cluster analyses were carried out by means of the SPSS programme. Once cluster (collective climate) composition was established, additional analyses were carried out in order to determine the adequacy of clusters, and whether aggregation of individual scores in each cluster was justified. First, univariate ANOVAs were developed to find out if there were statistically significant differences between clusters across the climate dimensions. Second, James (1982) pointed out that, prior to aggregation of individual climate scores to obtain collective climate scores, perceptual agreement should be demonstrated. To evaluate perceptual agreement on each climate dimension in each cluster or collective climate, the intra-class correlation coefficient 1 ([CC(1)) was used (James, 1982). A high ICC(l) indicates that participants in each cluster agree on their description of their work environment, and hence aggregation is justified. To investigate the association of collective climate membership with the other variables considered in this study, a number of contingency table analyses and univariate ANOVAs were carried out.

Results Three collecdve climates were idendfied in the study sample. Collective climate 1 was composed of 111 pardcipants, of whom 61.3% were female. The average age was 34.7 (SD = 7.9). In this group, 3.6% were managers, 11.7% middle-level managers, 12.6% team leaders and 72.1% lower level employees. No top manager was classified into this cluster. Collecdve climate 2 was made up of 49 pardcipants, of whom 51% were female. The average age was 36.5 (SD = 8.3). In this group, 12.2% were top managers, 18.4% were secdon managers, 16.3% were subsecdon managers, 12.2 team leaders and the remaining 40.8% lower level employees. Finally, collecdve climate 3 was composed of only three pardcipants, of whom one was female. The average age was 31.3 (SD = 2.9). All were lower level employees. Table 1 shows the collecdve climates' means in each climate dimension. The univariate ANOVAs carried out to test if the differences among the collecdve climates' means in each climate dimension were stadsdcally significant and provided significant F rados in all cases. To assess within-cluster agreement, tbe intra-class correladon coefficient 1 (ICC(l)) and tbe inter-rater reliability (IRR) coefficient {r^^Q-^ proposed by James, Demaree & Wolf (1984) were calculated.

32

Vicente Gon^dlet^^Romd et al.

Table 1. Collecdve climates means across climate dimensions and intra-class correladon coefficients 1 (ICC(l)) Clusters 1 2 3 F

ICC(l)

Support

Innovation

Rules

Goals

2.5 3.6 1.8

2.4 3.2 2.0

4.0 4.9 1.5

2.7 3.8 1.4

72.5** 0.57

37.4** 0.40

46.1** 0.46

81.2** 0.60

Tbe intra-class correladon coefficient 1 computed for eacb climate dimension varied from .40 (innovadon) to .60 (goals) (see Table 1). Tbe IRR coefficient was computed for tbe tbree collecdve climates across tbe four climate scales. Tbe coefficients obtained ranged between .80 and .89. All tbese results indicated tbat tbe tbree collecdve climates were discriminable and tbat tbere was an acceptable degree of witbin-cluster agreement; consequendy aggregadon was jusdfied. Two similarides were observed between tbe patterns of clusters 1 and 2 across tbe four climate dimensions. First, botb presented a peak on tbe rules climate dimension; and second, tbeir sbapes were quite similar. Mainly, tbe differences between tbese two cluster patterns bad to do witb level ratber tban witb pattern sbape. Finally, cluster 3 sbowed a weak climate witb low scores on tbe four climate dimensions. To test wbetber collective climates' membersbip was related to membersbip in tbe coUecdvides defined by departmental assignment, bierarcbical level, sbift and job locadon, a number of condngency analyses were carried out (see Table 2). In order to prevent metbodological problems, cluster 3 was eliminated from tbe analyses because of its small size. Tbe results obtained sbowed tbat bierarcbical level^ and sbift were significandy related to collecdve climate membersbip. Job locadon was only marginally related to collecdve climate membersbip since tbe cbi-square stadsdc computed to test tbe null bypotbeses of independence between botb variables yielded a^ value of .058. Department assignment was not related to collecdve climate membersbip. Given tbat tbe autbors knew tbat top managers were used to work on a split-sbift basis, tbe reladonsbips among bierarcbical level, sbift and collecdve climate membersbip were furtber analysed. First, a condngency analysis between bierarcbical level and sbift revealed tbat tbere was a stadsdcally significant associadon 'The contingency table between hierarchical level and collective climate membership yielded 3 out of 10 cells with expected frequencies of less than five. It has been recommended that all expected frequencies be at least five. However, Everit (1977) indicates that this condition is probably too stringent and can be relaxed. Welkowitz, Ewen & Cohen (1976) have also asserted that the greater the degrees of freedom associated, the less severe the requirement. To complement the chi-square statistic, the Goodman & Kruskal tau statistic (a measure of association not based on the chi-square statistic) was computed. The value obtained when collective climate membership was considered as the dependent variable was .18 (/i<.0001).

The validity of collective climates

33

Table 2. Results of condngency analysis: reladonsbips between collecdve climates membersbip and departmental membersbip, bierarcbical level, sbift and job locadon Variables Hierarchical level Shift Job location Department

y^

d.f.

p"

28.58 10.05 10.67 4.59

4 1 5 5

.001 .002 .058 .769

"One-tail probability.

between botb variables (%^(4) = 48.19, ^ < .0001)^. Second, in order to determine wbetber tbe individual contribudon of eacb variable to tbe predicdon of collecdve climate membersbip was stadsdcally significant, a number of logisdc regression analyses were carried out. Tbis metbod was preferred because it requires far fewer assumpdons tban discriminant analysis. Tbe SPSS programme makes it possible to include categorical variables as predictors by creadng tbe necessary dummy variables. To test tbe null bypotbesis tbat tbe coefficient (Bj) of one variable (X) is zero, tbe likelibood-rado (LR) test bas been recommended (Norusis, 1993). Tbe likelibood associated witb eacb model is tbe probability of tbe observed data given tbe parameter esdmates for eacb model. Tbe LR is obtained by dividing tbe likelibood for tbe model not containing X (tbe reduced model) by tbe likelibood for tbe full model. Tbe quandty minus two dmes tbe log of tbe LR stadsdc ( ~ 2 log LR) bas a cbi-square distribudon witb degrees of freedom equal to tbe difference between tbe number of predictor variables in tbe full and tbe reduced model. Tbis can be used to test tbe nuU bypotbesis tbat tbe coefficient of X is zero. Table 3 sbows tbe — 2 log LR values obtained wben bierarcbical level and sbift were removed from tbe model. Tbe results sbow tbat tbe null bypotbesis tbat tbe coefficient of sbift is zero could not be rejected ( — 2 log LR = .10,^ — .75). Tbe null bypotbesis tbat tbe coefficient of bierarcbical level was zero was rejected ( ~ 2 log LR = 19.12, jf)< .001). Tberefore, wben tbe significant influence of bierarcbical level on collecdve climate membersbip was controlled, sbift was not a significant predictor of tbe former variable. To test wbetber collecdve climates membersbip was related to organizadonal tenure, an univariate ANOVA was carried out. Tbe results obtained sbowed a stadsdcally significant reladonsbip between botb variables {F- 6.61,p< .OW). Tbe average tenure for tbe two collecdve climates (CC) idendfied were 9.6 (CCl) and 13.3 (CC2) (standard deviadons were 7.3 and 10.4, respecdvely). Given tbat in bureaucracies organizadonal tenure is an important factor in promodons, tbe reladonsbip between tenure and bierarcbical level was examined. Botb variables were significandy correlated (r= — .57, ^<.OO1). Next, tbe ^In this case, the contingency analysis yielded 4 out of 10 cells with expected frequencies of less than 5. The Goodman & Kruskal tau statistic obtained when shift was considered as the dependent variable was .33

34

Vicente Gonq^dlef^-Romd et al. Table 3. Results of logistic regression analysis: tests of the null h)rpotbesis tbat tbe coefficients for hierarcbical level and shift wben predicting collective climate membership are zero Variable removed from the full model Hierarchical level Shift

- 2 Log LR

d,f,

p

19,12 0,10

4 1

,0007 ,75

Table 4. Results of logistic regression analysis: tests of the null hypothesis that tbe coefficients for hierarchical level and tenure when predicting collecdve climate membership are zero Variable removed from the full model Hierarchical level Tenure

- 2 Log LR

d,f,

p

20,65 0.01

4 1

,004 ,91

relationships among hierarchical level, organizational tenure and collective climate membership were further analysed by means of logistic regression analysis. To determine wbether tbe individual contributions of hierarcbical level and organizational tenure to tbe prediction of collective climate membership were statistically significant, the likelihood-ratio (LR) test was implemented (see Table 4). The results sbow that the null hypothesis that the coefficient of organizational tenure is zero could not be rejected ( — 2 log LR = .1, p= .91). Tbe null hypothesis tbat the coefficient of hierarchical level was zero was rejected ( ~ 2 log LR = 20.65,^< .01). Therefore, wben tbe significant influence of hierarchical level on collective climate membership was controlled, organizadonal tenure showed no significant effect on tbe former variable. In summary, tbe multivariate techniques employed showed tbat collective climate membership was only significandy related to hierarcbical level. Under the hypothesis of independence between collective climate membership and hierarcbical level, it was expected tbat only 30% of tbe top managers would be classified into collective climate 2. However, the observed data showed that all of them were included into that cluster (see Table 5). Under the same hypothesis of independence, it was expected tbat 44.4% of tbe section managers would be classified into collective climate 2, but the observed data showed tbat 69.2% were classified into that cluster. For tbe subsection managers and tbe team leader levels, differences between observed and expected frequencies were very small (less tban 1,7); 61.9% of the subsection managers and 70% of tbe team leaders were classified into collective climate I. Finally, under the independence hypothesis it was

The validity of collective climates

35

Table 5. Contingency table between collective climate membership and hierarchical level Hierarchical level Top managers Section managers Sub-section managers Team leaders Lower-level employees

Collective climate 1 0 0% 4 30,8% 13 61,9% 14 70% 80 80%

CoOective climate 2 6 100% 9 69,2% 8 38,1% 6 30% 20 20%

Note: In each cell, the upper numhers are counts, the lower numbers are row percentages.

expected that 69,4% of lower level employees would be classified into collective climate 1, but tbe observed data sbowed that 80% were included into that cluster. The analysis of the contingency table between collective climate membership and hierarchical level suggests the following interpretadons. First, collective climate 2 basically represented tbe organization view of top managers just as it went through the other hierarcbical levels. The average scores of top managers on the four climate dimensions (support = 3,5, innovation = 3.7, rules = 5,1, goals = 4.1) seem to support tbis assertion since they were similar or higher than the average scores for collective climate 2. Second, tbe top managers' view differentially went tbrougb the other hierarcbical levels, so that the lower tbe hierarcbical level, tbe lesser the penetration of the top managers' view. Among section managers tbat view still was the prevailing one, but in the subsequent hierarcbical levels it was the minority view. And third, collecdve climate 1 essentially represented tbe view of middle- to low-level employees. To test whether membership in collective climates was significantly related to employees' job satisfaction, job involvement and organizational commitment, three univariate ANOVAs were performed. The results obtained pointed out tbat membership in collective climates was significantly related to employees' job satisfaction (/^= 3.41, ^<.001), job involvement iF~ A.A2,p<.00\) and organizational commitment (/^= 5,87,/>< ,001), Collective climate 2 showed the highest means on the tbree criteria. Once the association between collective climate membership and hierarchical level was demonstrated, and following Payne's (1990) recommendation, we further investigated the relationship between hierarcbical level and organizational climate perceptions. Differences among the five hierarchical levels on tbe four studied climate dimensions were analysed by means of univariate ANOVAs. The results obtained sbowed tbat tbere were significant differences on all tbe climate dimensions (see Table 6). Post hoc Bonferroni tests revealed that significant

36

Vicente Gon^dlev^Romd et al.

Table 6. Hierarchical levels' means across climate dimensions, ANOVAS' F ratios and post-hoc comparisons CUmate dimension

1

Support Innovation Goals Rules

3,5 3,7 4,1 5,1

Hierachical levels 2 4 3 3,6 3,2 3,6 4,5

2,9 2,7 3,1 4,1

2,9 2,5 3,0 4,6

5

F

Post-hoc comparisons

2,6 2,5 2,9 4,2

8,0** 9,9** 8,7** 3,3*

1:5; 2:5,4 1:3,4,5; 2:4,5 1:3,4,5; 2:5 —

*;)<0.02; **;)<0.0001. Note: Hierarchical levels: 1, top managers; 2, section managers; 3, sub-secdon managers; 4, team leaders; and 5, lower-level employees. Post-hoc comparisons: 2:4,5 means that there are significant differences between group 2 and groups 4 and 5.

differences (/)<.O5) in the dimensions of support, innovation and goals only appeared between top managers and section managers on the one hand, and some of the remaining hierarchical levels on the other. Regarding the dimension of rules, no significant difference was obtained. These results suggested the existence of two hierarchical climates, with top managers and section managers having similar climate perceptions, but different from those of middle- to low-level employees. To test this proposition, first hierarchical level was dichotomized. Participants were classified into high level (top managers and section managers) and middle- to low-level (subsection managers, team leaders and low-level employees) categories. Then, univariate ANOVAs across the four climate dimensions were carried out. The results obtained showed significant differences between the two hierarchical climates across the four climate dimensions (support: /^= 26.2, ^ < .0001; innovation: /^= 32.6, ^ < .0001; goals: F- 28.4, ;><.0001; rules: F= 5.3,/)<.O3). Next, and following the suggestion of James (1982), the ANOVA results were converted to ICC(l)s to obtain measures of agreement within hierarchical levels across the four dimensions of climate. The ICC(l) values were .24 for support, .28 for innovation, .25 for goals and .05 for rules. These values seemed to suggest that there were medium-to-low levels of agreement within the five hierarchical levels for the four climate scales. However, as James has acknowledged (James, Demaree & Wolf, 1984), the ICC(l) cannot assume high values unless the between-group mean squares is large enough in relation to the within-group mean squares. Therefore, a result indicating sufficient perceptual agreement will be obtained only when within-group homogeneity accompanies large between-group discrimination. An alternative procedure to analyse within-group agreement is to use agreement measures which only focus on within-group homogeneity, such as the inter-rater reliability (IRR) coefficient ('wGO)) proposed by James, Demaree & Wolf (1984). This coefficient was computed for the two hierarchical climates across the four climate scales. The eight coefficients obtained ranged between .80 and .89. Considering that these values fell above the acceptable level of .70 (Nunally, 1978), it was concluded that there was

The validity of collective climates a good deal of agreement within hierarchical climates across the four climate scales. Therefore, the hierarchical climates identified in the present study meet the criteria of agreement and discrimination. The relationship between hierarchical climates and the individual criteria considered was also investigated. The results yielded by the univariate ANOVAs carried out showed statistically significant differences between the two hierarchical climates across the three criteria (job satisfaction: /^= 18.21, j&<.0001; job involvement: 7^=11.57, ^<.OOO8; and organizational commitment: 7^=31.29, p< .0001). The high-level employees group showed the highest means on the three criteria. So, the hierarchical climates identified here also met the criterion of showing significant relationships with individual criteria. Finally, to ascertain how much overlap there was between collective climates' membership and hierarchical climates' membership a contingency table analysis was carried out. The results obtained showed a high overlap between both variables (X^(l) = 23.70,/<.00001). The obtained contingency table showed that 78.9% of the high-level employees were classified into collective climate 2 by the clustering procedure employed, and 75.9% of middle-to-low employees were classified into collective climate 1. Discussion The aim of the present paper was to study the validity of the collective chmate construct. In the introduction to the present study, it was argued that if collective cUmates are to represent a meaningful scientific concept they will have to be related to membership in some formal or informal structured collectivities (Payne, 1990). From this perspective, it was expected that membership in collective climates was related to membership in the collectivities defined by departmental membership, hierarchical level, shift, job location and organizadonal tenure. The results obtained only yielded empirical support for the relationship between hierarchical level and collective climate membership. The analysis of the contingency table between the aforementioned variables suggested that collective climate 2 basically represented the organization view of top managers just as it differentially went through the other hierarchical levels. In turn, collective climate 1 essentially represented the view of middle- to low-level employees, levels in which top managers' view had not widely penetrated. According to these interpretations, the collective climates obtained in the present study do have a psychosocial meaning since they help to ascertain the penetration of top managers' organizational view through the other hierarchical levels. Louis (1985) has stated that, in order to analyse the penetration of a given organizational culture, sociological, psychological and historical aspects have to be considered. Sociological aspects refer to the dispersion or scope of the culture. Applying this rationale to the organizational climate concept and to the present study, the relationship between collective climate membership and hierarchical level showed the scope of the top managers' view in the studied organization. Taking into account all these arguments and results, it can be concluded that the validity of the collective climates concept has received some support from the present study.

37

38

Vicente Gonr^dle^Roma et al.

Tbis conclusion sbould not lead climate researcbers to take collective climate validity for granted. We bave to recall tbat in previous studies it was not supported. In Joyce & Slocum's (1984) study, sbift and job location sbowed no significant reladonsbip witb collective climate membersbip. Patterson and colleagues (1996) did not find any associadon between collecdve climate membersbip and work-team membersbip, bierarcbical level, job type and region wbere tbe work-team operated. Only Jackofsky & Slocum (1988) found an associadon between collecdve climate membersbip and departmental assignment, but only at dme 1, not at dme 2. Collecdve climates is an attracdve type of aggregate climate because it solves tbe agreement problem. However, an indiscriminate use of collective climates would be confusing. Before using collecdve climates to explain organizadonal criteria, it is necessary to demonstrate tbeir socio-psycbological meaning and sbow tbat tbey are useful for understanding some organizadonal processes (Payne, 1990). Only tben could bypotbeses regarding tbe reladonsbips between collecdve climates and organizadonal criteria be accurately formulated. Wben collecdve climates were proposed, previous researcb bad sbown tbat tbe perceptual agreement criterion bad not often been acbieved for different suborganizadonal collecdves and units sucb as department, bierarcbical level and work-team (Joyce & Slocum, 1984; Patterson et al, 1996). Collecdve climates were proposed as a way to meet botb tbe perceptual agreement and tbe discriminadon criteria for aggregate climates. But tbe difficuldes in acbieving tbe perceptual agreement criterion for exisdng suborganizadonal collecdves sbould not lead researcbers to presuppose perceptual beterogeneity witbin exisdng units and, tben, to immediately define collecdve climates as tbe only way one bas to obtaining aggregate climates wbicb meet tbe criteria of perceptual agreement and discriminadon. In tbe present study, following tbe alternadve approacb of invesdgadng discriminadon between and perceptual agreement witbin tbe exisdng collecdves defined by bierarcbical level, two bierarcbical climates were idendfied wbicb met tbe aforemendoned criteria and wbicb sbowed significant reladonsbips witb organizadonal or individual outcomes (Joyce & Slocum, 1984). Indeed, tbere was a good deal of overlap between tbe soludons yielded by tbe two approacbes. For botb soludons it can be concluded tbat tbere were two sbared views of tbe studied organizadon: one wbicb is beld and promoted by top and secdon managers, and tbe otber wbicb is tbe prevailing view among middle- to low-level employees. Botb soludons suggest tbat bierarcbical level may be a key factor in understanding tbe formadon of aggregate climates in bureaucracies like tbe studied organizadon. Tberefore, botb approacbes may be useful ways to ascertain tbe factors wbicb contribute to tbe formadon of aggregate climates. However, to acbieve tbis goal, researcbers using tbe collecdve climates approacb sbould invesdgate tbe psycbosocial meaning of tbe clusters obtained, tesdng if clusters are based on some formal or informally structured collecdvity (Payne, 1990). Researcbers wbo invesdgate aggregate climates as related to exisdng collecdves and units sbould consider using agreement measures based only on witbin group bomogeneity wben tbe perceptual agreement criterion is analysed.

The validity of collective climates

39

The results provided by the multivariate techniques employed in the present study have important implications for future research. The significant bivariate relationships found between collective climate relationship and shift and organizational tenure disappeared when the influence of hierarchical level was controlled and the aforementioned relationships were re-estimated using multivariate methods. Jermier et al. (1991) found significant bivariate relationships between subculture membership and department assignment, rank, shift, organizational tenure and organizational commitment. But when the unique contribution of the considered variables in explaining subculture differences was investigated using multivariate methods (discriminant analysis), only the contributions of organizational commitment and rank were statistically significant (^<.O5). These results and those reported in the present paper stress the necessity of using multivariate methods in the study of collective climate validity. Finally, investigating whether collective climates have some psychosocial meaning and are not only statistical artifacts implies looking for the factors which could influence climate formation. However, more longitudinal research is required in order to understand the processes of climate formation and to ascertain the variables that actually play a relevant role in it. This research should examine the role of attitudinal variables (such as job satisfaction) which have frequently been considered only as outcomes of climate perceptions, since some theoretical models (Payne & Pugh, 1976) and the results of cross-sectional studies (James & Jones, 1980; James & Tetrick, 1986) suggest that the relationship between them and climate perceptions is reciprocal. Acknowledgements The authors wish to thank two anonymous reviewers for their helpful comments on an earlier draft. This study was partially sponsored by a research grant from Direccion General Interministerial de Ciencia y Tecnologia, Research Project: PB94-0997.

References Aldenderfer, M. S. & Blashfield, R. K. (1984). Cluster analysis. Sage University Paper Series on Quantitative Applications in the Social Sciences, 07-044. Beverly Hills, CA: Sage Publications. Ashforth, B. E. (1985). Climate formation: Issues and extension. Academy of Management Review, 4, 837-847. Everitt, B. S. (1977). The analysis of contingency tables. London: Chapman & Hall. Jackofsky, E. F. & Slocum, J. W. Jr. (1988). A longitudinal study oi
40

Vicente Gon:(ale^Romd et al.

James, L. R. & Tetrick, L. E. (1986). Confirmatory analytic texts of three causal models relating job perceptions to job iinsi&cnoa. foumal of Applied Psychology, 71, 77-82. Jermier, J. N., Slocum, J. W., Jr., Fry, L. W. & Gaines, J. (1991). Organizational subcultures in a soft bureaucracy: Resistance behind the myth and facade of an official culture. Organisation Science, 2, 170-194. Joyce, V. F. & Slocum, J. W. (1984). Collective climate: Agreement as a basis for defining aggregate climate in organizations. Academy of Management foumal, 27, 721—742. Kozlowski, S. W. J. & Doherty, M. L. (1989). Integration of climate and leadership: Examination of a neglected issue, foumal of Applied Psychologf, 74, 546—553. Lodhal, T. & Kejner, M. (1965). The definition and measurement of job invohement. foumal ofApplied : Psychology, 49, 24-33. Louis, M. R. (1985). Sourcing workplace cultures: Why, when and how. In R. H. Kilmann, M. J. Saxton, R. Serpa and associates (Eds), Caining control of the corporate culture, pp. 126-136. San Francisco, CA: Jossey Bass. Melia, J. L. & Peiro, J. M. (1989). El cuestionario de satisfaccion SlO/12: Estructura factorial, fiabilidad y validez (The SlO/12 satisfaction questionnaire: Factor structure, reliability and validity). Revista de Psicologia del Trabajoy de las Organi^aciones, 4, 179-187. Norusis, M. J. (1993). SPSS for Windows: Advanced Statistics. Release 6.0. Chicago, IL: SPSS. Nunally, J. C. (1978). Psychometric theory. New York: McGraw-Hill. O'Reilly, C. & Chatman, J. (1986). Organizational commitment and psychological attachment: The effects of compliance, identification and internalization on prosocial behavior, foumal of Applied Psjchologf, 71, 492-499. Patterson, M., Payne, R. & West, M. (1996). Collective climates: A test of their socio-psychological significsince. Academy of Management foumal, 39, 1675—1691. Payne, R. (1990). Madness in our method. A comment on Jackofsky and Slocum's paper, 'A longitudinal study oi cyimates'. foumal of Organisational Behavior, 11, 77—80. Payne, R. & Pugh, D. S. (1976). Organizational structure and climate. In M. D. Dunnette (Ed.), Handbook of industrial and organisational psychology, pp. 1125-1173. Chicago: Rand McNally. iPritchard, R. D. & Karasick, B. W. (1973). The effect of organizational climate on managerial job performance and job satisfaction. Organisational Behavior and Human Performance, 9, 126—146. Robert, K. H., Hulin, C. L. & Rousseau, D. M. (1967). Developing an interdisciplinary science of organisations. San Francisco: Jossey Bass. Rousseau, D. M. (1988). The construction of climate in organizational research. In C. L. Cooper & I. Robertson (Eds), Intemational review of industrial and organisational psychology, pp. 137—158. London: Wiley. Welkowitz, J., Ewen, R. B. & Cohen, J. (1976). Introductory statistics for the behavioral sciences. New York: Academic Press. Received 7 February 1996; revised version received 17 November 1997

The validity of collective climates

merged; thus the number of clusters prior to the merger is the most probable solution' (Aldenderfer ..... Integration of climate and leadership: Examination of.

1MB Sizes 0 Downloads 366 Views

Recommend Documents

The Concept of Validity - Semantic Scholar
very basic concept and was correctly formulated, for instance, by. Kelley (1927, p. 14) when he stated that a test is ... likely to find this type of idea in a discussion of historical con- ceptions of validity (Kane, 2001, pp. .... 1952), the mornin

The Concept of Validity - Semantic Scholar
one is likely to build ever more complicated systems covering different aspects of .... metaphysics changes the rules of the game considerably. In some highly ...

The Effects of Ethical Climates on Bullying Behaviour in ... - CiteSeerX
study is to explore the effects of bullying behaviour upon ... Journal of Business Ethics (2009) 86:273–295. © Springer 2008 .... affect the definition of their workplaces (Altmann,. 2000). ...... Manchester Institute of Science and Technology,.

The Effects of Ethical Climates on Bullying Behaviour in ... - CiteSeerX
organizational commitment, supervisory support .... autonomy, trust, cohesiveness, support, recognition, ...... Manchester Institute of Science and Technology,.

Examination of the Predictive Validity of Preschool ...
.edu/techreports/dissemination.html#TechRep. Graduate ... get/procedures_and_materials/index.html. All research .... (Missail et ai., 2006). Kindergarten ...

Challenging the reliability and validity of cognitive measures-the cae ...
Challenging the reliability and validity of cognitive measures-the cae of the numerical distance effect.pdf. Challenging the reliability and validity of cognitive ...

Validity of the construct of Right-Wing Authoritarianism and its ...
Moreover, this scale appears to be the most reliable research tool since it is regarded as ..... with other scales (e.g. the Big Five) (Akrami & Ekehammar, 2006; Altemeyer, 2006). ... analysis that their data did not conform to a one-factor solution.

Validity of the construct of Right-Wing Authoritarianism and its ...
Moreover, this scale appears to be the most reliable research tool ... internationally comparable analytical tools and focus on their comparability across different.

The micro-empirics of collective action: The case of business ...
Jul 13, 2011 - increasingly important instance, the Business Improvement District. A ..... large enough number of agents choose to act as proponents. The ...... developer leadership dates as far back as the 1800 s for private streets in.

Validity: on the meaningful interpretation of ... - Semantic Scholar
achievement is a construct, usually inferred from per- formance on assessments such as written tests over some well-defined domain of knowledge, oral exami-.

Experiences of discrimination: Validity and ... - Semantic Scholar
Apr 21, 2005 - Social Science & Medicine 61 (2005) 1576–1596. Experiences of ... a,Г. , Kevin Smith b,1. , Deepa Naishadham b. , Cathy Hartman .... computer (in either English or Spanish), followed by a ... Eligible: 25 – 64 years old; employed

Experiences of discrimination: Validity and ... - Semantic Scholar
Apr 21, 2005 - (Appendix 1), based on the prior closed-format ques- tions developed by ..... times more likely than white Americans to file com- plaints about ...

On the predictive validity of implicit attitude measures - Semantic Scholar
Social psychologists have recently shown great interest in implicit attitudes, but questions remain as to the boundary conditions .... Hilton (2001)), nearly all were focused on the ability ..... information about his name, address, social security.

Duress, Deception, and the Validity of a Promise - Oxford Journals
different accounts of why duress and deception invalidate promises. According to ... Their main interest is in the remedies for breach of contract, in how costs ...

On the predictive validity of implicit attitude measures
information about his name, address, social security number, place of birth, current ...... These include individual difference variables (self-moni- toring, private ...

Factor Validity and Reliability of the Sport Friendship ...
degree of reciprocal influence over one another (Rubin et al., 2006). This ..... ogy researchers and a specialist in the field of friendship research. .... sidered to be their best friend in sports (on a team in their club sport or in physical educat

On the Validity of Econometric Techniques with Weak ...
However,. Luiz Cruz is a Ph.D. student of economics at the University of California at Berkeley. .... confidence intervals have coverage probability much smaller than the commonly ...... Journal of Business and Economic Statistics 13:225–35.

On the Validity of Simulating Stagewise Development ...
of Simulating Stagewise Development by Means of PDP Networks: Application of Catastrophe Analysis and an Experimental. Test of Rule-Like Network Performance. MAARTJE E. J. RAIJMAKERS. SYLVESTER VAN KOTEN. PETER C. M. MOLENAAR. University of Amsterdam

On the predictive validity of implicit attitude measures - Semantic Scholar
implications for the degree to which people use/apply their attitudes ... A total of 85 non-Black college undergraduates par- ...... San Diego, CA: Academic Press.

Validity of the phase approximation for coupled ...
original system. We use these results to study the existence of oscillating phase-locked solutions in the original oscillator model. I. INTRODUCTION. The use of the phase dynamics associated to nonlinear oscil- lators is a .... to the diffusive coupl

boundary layer climates oke pdf
Sign in. Loading… Whoops! There was a problem loading more pages. Whoops! There was a problem previewing this document. Retrying... Download. Connect ...