2.1. Participants
The survey was distributed to three workplaces (a university, a hospital, and a construction site) in the city of Gävle, Sweden. The questionnaires were distributed to 623 respondents in May 2009. In September 2009, 344 completed surveys were collected, giving a total response rate of 55.2%. The characteristics of the sample population are listed in
Table 1.
Table 1.
Description of the sample.
Table 1.
Description of the sample.
Variable | n | Descriptive |
---|
Population and households | | |
Gender (% men) | 342 | 21.3 |
Mean age ± SD (years) | 332 | 45.6 ± 12.2 |
Average number of persons/household | 340 | 2.7 |
Dwelling (% of total sample) | | |
Detached house | 174 | 52.1 |
Semidetached house | 58 | 17.4 |
Flat/apartment | 99 | 29.6 |
Other | 3 | 0.9 |
Workplace (% response rate) | | |
Hospital | 227 | 59.7 |
University | 97 | 50.2 |
Construction site | 20 | 40.0 |
The three workplaces included in the sample were chosen based on a previous study [
3]. The sites differ substantially with respect to waste intensity (kg waste per employee and year) and the share of handled hazardous waste in relation to total waste amount, with the university having the lowest and the construction site having the highest total amount of waste per employee. However, owing to the low number of completed questionnaires returned from the construction site, it was not possible to draw reliable conclusions regarding the differences in these factors between the groups. Therefore, the 20 respondents from the construction site were excluded from the analyses in the present report.
2.2. Questionnaire
The survey was devised to examine employees’ source separation behaviours at their workplace and household. Emphasis was placed on understanding the reasons and psychological mechanisms for recycling in the workplace in relation to the adoption of EMS, and on elucidating source separation behaviour at work has a spill-over effect on source separation behaviour at home.
On the first page of the questionnaire, participants were informed that the aim was to investigate source separation behaviour and opinions regarding waste management at different workplaces. Participants were informed that their responses would be treated anonymously.
The survey comprised 22 forced-choice questions and statements that required the respondent to indicate the level of agreement. In the present study, although the reported values for several variables are high, we have no reason to believe that the responses to some items are subject to a greater degree of social desirability bias than the responses to others. Therefore, there is no hindrance to comparing and analysing the data. The questionnaire items that are relevant to the present study are described below.
The questionnaire asked general questions about the workplace, as well as the respondents’ environmental concern and perceptions of the efficacy of source separation. Participants were asked to categorise their workplace as being within the healthcare or education sector. They then indicated their awareness of whether their workplace is engaged in an EMS by choosing one of the following response alternatives: (1) ‘Yes, the EMS work has been on-going during a long period’; (2) ‘Yes, the EMS work has just started’; (3) ‘Not yet’; or (4) ‘I don’t know’. Other questions asked participants to rate to what extent they view themselves as environmentally concerned, whether source separation requires too much time and effort, and if it is still worth the effort. The responses were made on a 7-point scale, which ranged from (1) ‘strongly disagree’ to (7) ‘strongly agree’.
Respondents were also asked to rate the perceived difficulty of performing source separation at their workplace. Another question assessed participants’ beliefs regarding the efficiency of source separation in terms of decreasing energy consumption in society, and the contribution of source separation to conservation of resources. Respondents’ personal norms were assessed by asking the participants to state their degree of personal responsibility to source separate their waste at work. Social norms were evaluated by asking to what extent respondents believe their colleagues’ feel personal responsibility to source separate (prescriptive social norm), and whether their colleagues actually engage in source separation behaviour at work (descriptive social norm). All these items were rated on the 7-point scale described above. Respondents were also asked to define the proportion of the total amount of waste at their workplace they source separate and to state the extent to which they have the possibility to de-crease this amount on a 5-point scale, ranging from (1) ‘small part’ to (5) ‘large part’.
Source separation behaviours within the household were also investigated. One question asked to what extent the respondent source separates their waste, and two other questions asked the respondent to rate the extent of source separation before and after their workplace adopted EMS, respectively. The items were rated on a 7-point scale, which ranged from (1) ‘not at all’ to (7) ‘very much’. Questions about demographics (age, residence, gender, and level of education) concluded the questionnaire.
2.3. Measures
One way of stabilising statistical results is to use indices that are constructed from a number of items instead of single items. These indices can be created if the variations in responses to a number of items largely reflect the variations in a single latent variable. A statistical method that describes variability among observed variables is factor analysis, which is commonly used in behavioural sciences. In the present study, we used a form of factor analysis, called principal component analysis (PCA), to cluster items that measure the same constructs. By averaging the items loading on the same factor, indices were constructed. Items with factor loadings less than 0.45 were not included in the indices. The indices of single items were used in the subsequent analyses.
The items that captured environmental concern in everyday behaviour were subjected to PCA with Varimax rotation, which resulted in a two-factor solution that explained 72.3% of the variance (KMO = 0.70; Bartlett’s test, approximate χ2 = 76.65, p < 0.001). The first factor, which was labelled “environmental concern”, consisted of the items that captured whether respondents are environmentally conscious, act in environmentally benign manners, and do what they can to reduce the amount of waste (Cronbach’s α = 0.77). The second factor consisted of two items, i.e., that source separation requires too much time and commitment, and that the net contribution of waste separation is not worth the effort (Cronbach’s α = 0.67). This factor was not included in the subsequent analysis.
PCA was also performed on the items that captured perceived difficulty associated with waste separation in the workplace. A one-factor solution that explained 56.3% of the variance (KMO = 0.61; Bartlett’s test, approximate χ
2 = 108.76,
p < 0.001) consisted of the following three items: the belief that the recycling site is too distant in order for the respondent to engage in source separation at the workplace; that there are limited possibilities for source separation at the workplace; and a reversed item stating that it is easy to source separate (Cronbach’s α = 0.60). The factor was labeled ‘Perceived difficulty’. To create an index of the items that corresponded to beliefs about the efficiency of source separation, an additional PCA was conducted. The result was a one-factor solution that explained 57.2% of the variance (KMO = 0.54; Bartlett’s test, approximate χ
2 = 148.84,
p < 0.001) and consisted of the following three items: ‘Source separation contributes to decreasing energy consumption in society’; ‘Source separation contributes to conservation of environmental resources’; and the reversed item ‘Source separation does not have a large impact on the decrease of environmental stress’ (Cronbach’s α = 0.55). The factor was labelled ‘Beliefs’. The Cronbach’s α-values for the indices of ‘perceived difficulty’ and ‘beliefs’ are low. According to George and Mallery [
31], α-values of about 0.60 are questionable. However, we argue that the indices still can be used in the analysis of theoretical terms.
The items that captured the normative components were submitted to PCA, which resulted in a two-factor solution that explained 66.6% of the variance (KMO = 0.71; Bartlett’s test, approximate χ2 = 311.38, p < 0.001) and consisted of the items: ‘measured personal responsibility to source separate at the workplace’; ‘feeling a personal obligation to contribute to handling waste in an environmentally friendly way’; and ‘not engaging in source separation creates a feeling of bad conscience’. These items were loaded on a single factor, which was labelled ‘Personal norms’ (Cronbach’s α = 0.77). Two items that captured social norms were loaded on the second factor but due to a low Cronbach’s α value of 0.34 the index of these items was not used in the subsequent analyses. Instead, one item that corresponded to the belief that the respondents’ colleagues thinking that the respondent should source separate represented a social norm.