A k-Means clustering procedure was used to identify 5 groups of students, from both Phase II (national cross-section) and Phase III (youth from disadvantaged areas) of this research. This was done after the data was weighted to bring the "disadvantaged" student population in-line with its true percentage of the population, about 17%. Since the samples were randomly selected and approximated the real population of students, we can reasonably assert that our 5 clusters represent natural groups in the student population as a whole.
Our approach will be from a path analytic perspective. Path analysis, due to Wright (1934)1, is a technique to assess the direct causal contribution of one variable to another in a non-experimental dataset. The problem, in general, is one of estimating the coefficients of a set of linear structural equations, representing the cause and effect relationships hypothesized by the investigator. The system involves variables of two kinds: independent (or cause) variables and dependent (or effect) variables.
In this instance, we intend to examine the causal relationship among the system of variables delineated below:
Basically, the model works as follows. Environmental education (Q.6 and Q.7) is hypothesized to lead to environmental knowledge (Q.5) [pathset-1]; demographic factors (region, urbanicity, and disadvantaged status) also influence environmental knowledge (Q.5) [pathset-2]. Environmental knowledge (Q.5) is hypothesized to lead to environmental concern (Q.8a-s) [pathset-3]. Environmental concern (Q.8a-s) is hypothesized to lead to a desire for more environmental knowledge (Q.12a-s) [pathset-4] and an interest in involvement in environmental activities (Q.28) [pathset-5].
On the one hand, the desire for more environmental knowledge (Q.12a-s) is hypothesized to lead to increased environmental knowledge (Q.5) [pathset-6], while interest in involvement in environmental activities(Q.28) is hypothesized to lead to actual involvement in environmental activities(Q.25,27), particularly when there are few obstacles standing in the way (Q.23,24,26) [pathset-7].
Table 2, (See print version for this), provides a detailed presentation of all the path coefficients involved in the model summarily described above. These coefficients were derived using the maximum likelihood simultaneous equation solution from LISREL 7 (Joreskog and Sorbom, 1989).
Thus, for example, in Pathset 1, the hypothesized relationship between parental environmental knowledge (Q.6) and a student's own knowledge (Q.5) is strong (0.221), stronger in fact than the relationship between environmental education in school (Q.7) and a student's own environmental knowledge (Q.5), which has a path coefficient of 0.078.
At the same time, another strong relationship exists between disadvantaged/non- disadvantaged status and overall environmental knowledge (Q.5), part of Pathset 2. The path coefficient for disadvantaged students and environmental knowledge is -0.012, indicating that these students tend to know less about the environment than students from non-disadvantaged areas, whose path coefficient for these two variables is 0.04.
Pathset 3, the hypotheisized relationship between environmental knowledge (Q.5) and environmental concern (Q.8a-s) is strong. In other words, students who profess to a good deal of environmental knowledge also tend to exhibit high levels of concern about various environmental issues and problems. Thus, the path coefficient between overall knowledge and concern about endangered plants, animals, insects is 0.098 and the path coefficient between overall knowledge and destruction of the rainforest is 0.095. This relationship is significant for sixteen of the nineteen issues, though the coefficients are lower.
In turn, concern about an environmental issue (Q.8a-s) often leads to the desire for further knowledge of that same issue (Q.12a-s) [Pathset 4]. The issues for which these two variables relate most closely include endangered plants, animals, insects (0.185), damage to the ozone layer (0.144), lead poisoning from water or old paint (0.121) and global warming (0.115). The relationship between concern and desire for further knowledge is weaker for other issues.
The next step in the path analysis was to discern if environmental concern (Q.8a-s) leads to interest in involvement in environmental activities (Q.28). As seen in Pathset 5, concern about various environmental problems often ties significantly to interest in involvement in environmental activities. The strong path coefficient for problems such as littering of trash and garbage (0.342), endangered animals, plants, insects (0.340), damage to the environment from mining or cutting down trees (0.304) and pollution of water from pesticides or fertilizers used in farming (0.183) indicate that concern for these problems leads many students to report interest in involvement in environmental groups and clubs. For other environmental problems, though, the relationship is either weaker or non-existent, such as not having enough energy (-0.122) and pollution of ocean waters or beaches (-0.090).
In Pathset 6, the desire for further environmental knowledge (Q.12a-s) was hypothesized to lead to greater environmental knowledge overall (Q.5). From the data, however, there is little support that a relationship exists between these two variables, and for nine of the nineteen problems, the relationship is negative, indicating that further knowledge of an issue may actually turn some students away from an increase in their overall knowledge. In the ten cases where the relationship is positive, none are significant, an indication that these two variables do not effect each other, at least in the terms of this path analysis.
Finally, in Pathset 7, we see a strong and significant relationship between interest in joining an environmental group (Q.28) and actual involvement in environmental activities (Q.25, Q.27). Thus, students who report an interest in joining a group often report joining groups both at school (Q.25, path coefficient of 0.102) and in the community (Q.27, path coefficient of 0.115).
In addition, we examined this relationship through a third variable, the existence of obstacles to involvement, generated by responses to Q.23, how easy or hard students feel it would be for them to be personally involved. As might be expected, when there is a low level of obstacles to personal involvement ("very easy" or "sort of easy"), interest in being involved is linked more closely with actual involvement (path coefficients of 0.112 for groups at school and 0.138 for groups in the community) than when students feel a high level of obstacles to personal involvement ("sort of hard" or "very hard"), path coefficients for which are 0.107 for groups at school and 0.081 for groups in the community. In other words, while participation in environmental groups is related strongly to interest in involvement, students who feel there are few obstacles to their personal involvement exhibit a greater likelihood of actual involvement than students who report a high level of obstacles to involvement.
Overall, then, a distinct path leads from environmental knowledge to environmental concern, from environmental concern to interest in involvement in environmental activities, and finally from interest in involvement in environmental activities to actual involvement in environmental groups, especially if there are few obstacles to involvement.