• No se han encontrado resultados

C ONDICIONES DE ELECCIÓN , UTILIZACIÓN , ALMACENAJE Y MANTENIMIENTO DE LOS E QUIPOS , M ÁQUINAS Y / O

PLIEGO DE CONDICIONES TÉCNICAS ESPECÍFICAS DE SEGURIDAD DE LOS EQUIPOS, MÁQUINAS Y/O MÁQUINAS-HERRAMIENTAS

7.2. C ONDICIONES DE ELECCIÓN , UTILIZACIÓN , ALMACENAJE Y MANTENIMIENTO DE LOS E QUIPOS , M ÁQUINAS Y / O

The consequences of the previously presented hypotheses for the research model will be revealed in the following sections. I shall describe the required statistical analyses, the measures for team results and the sample.

3.4.1 Methods for the Statistical Analysis

For the analyses of the longitudinal relationship between team responsiveness and team results, I use a regression model that predicts the results in one year by the responsiveness in an earlier year. This shows, to speak in terms of Gladstein (1984), whether team responsiveness affects results after a “significant time lag”. Figure 8 illustrates schematically the measurements of both responsiveness and results for each of the years.

Figure 8 Timeline of measurements for responsiveness and results

2001 t1 2002 t2 2003 t3 Timeline

For each year I perform a cross-sectional analysis to test the direct effect of team responsiveness on team results. Cross-sectional analyses are conventional in the literature and, besides, most of the Volvo data is useful for cross-sectional analyses only, as I will show later on. The consistency of results of these analyses for the different years of measurements will show how durable the relationships between team responsiveness and team results are. After that, I will test for a

longitudinal effect for t2 and t3, by adding the long-term input variables of t1 and t2

into the same model, where possible, and finally add the short-term input variables of either t2 or t3.

This longitudinal analysis is approached by a hierarchical multiple regression model (Field 2000; Miles & Shevlin 2001)). I will use regression analysis, because this “is a technique for modeling the relationships between two (or more) variables” (Miles & Shevlin 2001). “In regression analysis we fit a predictive model to our data and use that model to predict values of the dependent variable from one or more independent variables” (Field 2000). The method is called multiple because more than one predictor is used (in fact, three dimensions of team responsiveness and each through time) and it is called hierarchical because certain predictors are expected to have higher effects than other predictors (see my hypotheses).

The model starts with the earlier responsiveness (xt-1) as input, implying that the results (output) of today are first and foremost dependent on earlier responsiveness and that the later responsiveness (xt) contributes extra to these earlier responsiveness. Hierarchically, first the longitudinal sections of the model are entered, with team responsiveness in 2001 as input to the results in 2002 and team responsiveness in 2002 and/or 2003 as input to the results of 2003. Eventually, in the final step of the model, the cross-sectional section is added. Furthermore, the three dimensions of responsiveness are entered in a hierarchical order. In doing so, I follow the order suggested by the hypotheses. As a “basis for team results” (Campion, Medsker, & Higgs 1993; Gladstein 1984; Marks, Mathieu, & Zaccaro 2001) joint management will be entered first into the models for BP and QWL. Boundary management is expected to be the main effect for BP, while job management is expected to be the main effect for QWL. Additionally, job management will be added in the model for BP and boundary management for the model of QWL, since it is expected that both will have a positive (minor) effect as well.

In total there are three possible cross-sectional models and a maximum of nine longitudinal models, depending on the availability of data (table 8 and 9). Model 1 predicts team results by joint management, model 2 predicts BP by boundary management and QWL by job management, and model 3 predicts BP by job management and QWL by boundary management. Model 4, 5 and 6 add team responsiveness in the same order to model 1, 2 and 3, however, predicting team results by team responsiveness from the previous year. Model 7, 8 and 9 add team responsiveness in the same order to model 1 to 6, however, predicting team results in 2003 by team responsiveness in 2001.

Table 8 Overview of Statistical Models of Team Responsiveness Model Dimensions of team responsiveness

Model 1 Joint management

Model 2 Job management*

Model 3 Boundary management*

Model 4 Joint management xt-1

Model 5 Job management xt-1

Model 6 Boundary management xt-1

Model 7 Joint management xt-2

Model 8 Job management xt-2

Model 9 Boundary management xt-2

* The here presented order of entering job and boundary management is used for testing QWL models, whereas job and boundary management are entered in opposite order for all models that are used for testing the relationship with BP

Table 9 Overview of Cross-Sectional and Longitudinal Models of Team

Results Responsiveness 2001 Responsiveness 2002 Responsiveness 2003 Results 2001 Cross-sectional models 1, 2 and 3 Results 2002 Longitudinal models 4, 5 and 6 + cross-sectional models 1, 2 and 3 Results 2003 Longitudinal models 7, 8 and 9 + longitudinal models 4, 5 and 6 + cross-sectional models 1, 2 and 3

3.4.2 Measures of Business Performance

Suzaki’s Quality, Costs and Delivery (1993) and Cohen and Bailey’s “performance effectiveness assessed in terms of quantity and quality of outputs” (Cohen & Baily 1997) cover the measures for business performance (BP). These BP measures are only available for teams in the production departments (see table 10). However, due to the process structure of the paint-shop, no BP data are available on team level for this department either. Teams from supporting departments and the paint- shop are therefore excluded from the analysis of the relationship between responsiveness and BP.

For business performance I only use objective measures. Volvo’s measure of so- called “Direct OK” is used for product quality. “Direct OK” means the percentage of products or parts produced right directly, the first time around. In case of the paint- shop, these figures are only available on department level. Therefore, they are not

of interest for this study; I am interested in how the team’s responsiveness contributes to the team’s performance. In the press and detail shop and for most teams in the final assembly these data are available and therefore useful for this study. For some teams, data are only available on the level of sub-department (a number of teams belonging to the same team manager). In such cases I have randomly picked only one of the teams to use for the analysis. For all other departments, like supporting departments, no measures are available for quality. Quality in these departments is often difficult to measure. One could measure the quality of the financial department of course, for example by measuring satisfaction of internal customers (if possible to determine exactly), but such measures are hard to compare to those used in the production.

For the overall concept of costs both cost-index and capacity utilization are used as measures. Cost-index (shortly referred to as costs) relates to the percentage by which a certain unit, like a team or department, exceeds the budget it has been granted with for a certain period of time. Each month the financial department calculates for each cost center to what extent this budget is exceeded. The cost centers can be compared, because an index figure is used. The cost indices are not available on team level for the majority of teams, therefore one team was picked randomly from each cost center for the analyses. Utilization stands for the used capacity of man-hours per team per week. If a pre-flow team consists of six people, who each work eight hours a day, the capacity of that team for that day is 48 hours. If the pre-assembly on a cab takes four hours in total, the pre-flow team should produce twelve cabs during a day to have 100 percent utilization. If the team produces less, the utilization rate is lower. For each cab it is pre-calculated what the production time should be, based on its specific requirements. In other words, there is a variation in production time. The system to calculate the utilization considers these differences, but it also considers the differences per team. That is, if a team member is absent for a certain time, the available capacity in hours for that team also is lower. For each team, every day the available number of man- hours are calculated and compared to the number of cabs that are produced. For innovation (Dunphy & Bryant 1996) no objective data were available, while for delivery precision (Suzaki 1993) unfortunately too few data were available on team level to be of use for this study.

In table 10 I summarize the data of the three specific BP measures on plant level.

Table 10 Data on Business Performance (average per team)

2001 2002 2003

Product quality n 54 73 53

M 95.72 91.96 96.36

2001 2002 2003