Responsiveness: Processes of Work Teams
2.4 Methodology and Results
This section will present the operationalization of my responsiveness concept and the data collection. Next I will discuss the data analysis and its outcomes.
2.4.1 Data and Sample
The main source of data used in this study was a questionnaire filled out by all employees in the five production departments and five supporting departments of Volvo Umeå. The questionnaires were handed out by the team or department manager during one of the weekly team meetings and answered during work time. Three measurements have taken place between 2001 and 2003, each in the period March-April, with an intermediate period of one year.
For my data collection I used questionnaires containing items for all twelve key- aspects as defined in Table 2. The five-point Likert scale items were selected from questionnaires developed by Hut and Molleman during their research at Philips (1998), the Work Groups Effectiveness Model by Campion et al. (1993), other research at Philips by De Leede (1997) and items previously developed for research at Volvo (Kuipers & De Witte 2005a). These items (table 3) cover the previously defined twelve key-aspects. The table provides an overview of the number of items used for each aspect, example items, and the Cronbach’s Alpha’s calculated for each aspect (showing sufficient reliability) per year of measurement. I have specifically used items that reflect actions and behavior of teams, in other words, no items refer to emergent states (Marks, Mathieu, & Zaccaro 2001) or actions and behavior of individual team members. A total of 46 items were used to measure responsiveness.
Table 3 Items for the Twelve Key-aspects of Actions and Behavior in
Work Teams (alpha’s are subsequently for 2001, 2002 and 2003) Key-aspects No. of
items
Example item Cronbach’s
alpha’s*
Goal orientation 2 Team goals are formulated by the team and
based on the company’s goals
.76, .76, .74 Planning
activities
4 The team formulates its own weekly production
plan
Key-aspects No. of
items Example item Cronbach’s alpha’s*
Feedback 3 The team members address to each other in
case of mistakes in the task performance
.70, .63, .67 Conflict
management
4 The team members solve internal cooperation
problems without management interference
.78, .77, .81 Multi-functionality
& job rotation 5 The team members often interchange tasks .78, .77, .79
Delegated management & support tasks
4 The team carries out the routine maintenance .74, .75, .75
Work
communication
2 The team members share information about
the work
.64, .70, .66 Decision making
& control 6 The team divides the tasks .76, .84, .86
Performance
management 3 The team acts on mistakes .68, .67, .70
Improvement activities
4 The team members often take initiatives for
improvement
.76, .76, .77 Customer &
supplier relationships
4 The team solves problems with internal
customers
.87, .86, .88
Advanced management & support activities
5 The team arranges back-up and support when
necessary .69, .72, .73
* The samples range between n=1293 (in 2001) and n=1507 (in 2002)
During the three years of measurement more than 150 teams have been ‘followed’ with this questionnaire. Descriptions of these teams and the departments and processes in which they operate have been provided in 1.5. The total sample of 2001, 2002 and 2003 is summarized in table 4. The numbers of teams and individuals, as well as the response rates are mentioned for each year of measurement.
Table 4 Data Sample of Teams and Individuals for Team Responsiveness, Research Period 2001-2003
Department
2001 2002 2003
Press & Detail Shop Teams Individuals 23 (282) 25 (265) 26 (238) Body-in-white Dept. Teams Individuals 17 (263) 20 (272) 20 (245) Paint Shop Teams Individuals 22 (339) 22 (361) 21 (348) Pre-Assembly Dept. Teams Individuals 15* (183) 37 (369) 38 (405)
Final Assembly Dept. Teams Individuals 39* (630) 26 (442) 25 (365) Engineering & Technicians
Teams
Individuals 11 (284) 14 (278) 13 (274)
Material Handling Dept. Teams Individuals 14 (168) 12 (183) 12 (186) Supporting Dept.’s Teams Individuals 11 (66) 12 (63) 12 (64) Total teams Total individuals 152 1561 (73%) 168 1715 (76%) 167 1547 (68%)
* Departments reorganized by the end of 2001 (see section 1.5 )
2.4.2 Results of Factor Analyses
To explore the dimensions underlying the twelve key-aspects, a so-called factor analysis is carried out. Simply put, this analysis groups the items that highly correlate in factors that do not correlate with each other. These factors are also
called latent variables. With factor analysis one pursues factors that explain the maximum amount of variance with the smallest number of explanatory concepts (Field 2000).
The exploratory factor analysis was carried out on the 46 items of the questionnaire, for each year separately. Factor loadings ranged between .320 and .840. The outcomes of the KMO Measure of Sampling Adequacy can be considered as ‘superb’ (Field 2000), since all were above .9 (.94 in 2001 and 2002, .95 in 2003). Bartlett’s Test of Sphericity showed highly significant results for all three years (p< .001), which allows continuing the factor analysis.
First the scree-plots with eigenvalues were analyzed for 2001, 2002 and 2003, to understand the relative importance for each of the factors. In all three years the point of inflexion of the curve (cf. Field, 2000) lies with three factors, while for the remaining number of factors there are relatively low eigenvalues. Also the theoretical interpretation of these three factors (which will be discussed in the next section) appeared to be the easiest. As a result, a Varimax rotated factor analysis has been performed to provide three factors for each year of measurement (see Appendix A). It is not allowed to combine the data of different years in the same factor analysis, since data depends on the data of the year before. Therefore, first the analysis had to be carried out for each year separately and then a congruence test was used to determine the extent to which factors were comparable for each year of analysis. For this test the formula of Gorsuch (1974) has been applied6. To test this congruence the factor loadings for each year were compared; 2001 with 2002, 2002 with 2003 and 2001 with 2003. In order to claim replication of the factors for each pair of years the factor congruence should be above .9. This requirement is met for each of the factors, with .95 for 2001-2002, .99 for 2002- 2003 and .95 for 2001-2003. The explained variance for each of the factors lies between 10 and 20 percent per year, with a total explained variance between 41 and 45 percent for all three factors together (table 5). The Cronbach reliability per factor had alpha’s ranging between .85 and .93 per year, implying sufficient reliability (see also Appendix A).
Table 5 Percentage of Total Variance Explained (rotation sums of squared
loadings) Factor 2001 2002 2003 1 19.238 15.853 19.310 2 11.998 13.653 13.430 3 10.879 11.966 11.871 total 42.115 41.472 44.611 6
Jeremy Miles, Regression Theory at the Essex Summer school 2003, kindly provided an SPSS Syntax file and all possible help to perform this congruence test.
2.4.2.1 The Three Factors at Two Other Organizations
To have a first check of the “external” validity of the three factors, I obtained data of two other organizations7. Both studies used the same items for team responsiveness as I used at the Volvo study. The first study was at a Dutch production facility of a large automotive manufacturer, where data was collected of 23 production teams (Peppelman 2003). A total of 435 employees in this organization answered the questionnaire. The second study was at a waste management and transportation company in the Netherlands, where data was obtained on nine teams by 52 respondents (Bozon 2004). With the combined data- file of these two organizations, I carried out a Varimax rotated factor analysis for three factors. The factor loadings were tested for congruency with the factor loadings of Volvo in 2003. The congruency appeared to be .88, which is slightly below the suggested .9 required for claiming congruence (Gorsuch 1974). Deeper analysis of the single item congruencies, however, showed that two items for factor 3 (comparable to factor 3 in table 5) have a rather low score. After removing these items the total factor congruence improves to the required .9. I concluded that the same three factors are found as for Volvo, with the exclusion of two items. The Cronbach reliability alpha for factor 1 is .91, as it is for factor 2, while factor 3 has an alpha of .85. The explained variance for the three factors is respectively 22%, 14% and 10%, with a total of 46%.