CHAPTER 2: STUDY DESIGN & METHODS
2.3. Methods
2.3.2. Methods & data for RQ2
My second research question examines the incoming perspective the participant: What experiences, values, beliefs, and behavioral norms do participants bring to the LAIE, and how do learners differ across sites?
Table 2.5: Objectives and data for Q2
Objectives Data Collection
a) Establish participant’s significant life experiences relating to the environment and animals
b) Explore learners’ existing environmental values, beliefs, and norms
c) Understand how participants explain and rationalize their current levels of pro-environmental behavior
d) Uncover how the learner evokes VBN in their life narrative
1) Learner interviews: semi-structured 2) Learner interviews: prompt cards
e) Compare the distribution of attitudes, characteristics, and previous experiences among visitors at the three sites
3) Mixed-method survey (demographics, values scales, themes from interviews)
Learner Interviews
Interviews are good for examining environmental values because they allow the respondent to provide a descriptive, holistic, process-oriented, interpretive description of their environmental values (Weiss 1994). However, memory data can prove challenging to collect due to the difficulty in remembering and verbalizing such concepts (Chawla 1998a). I addressed this issue by conducting two types of interviews: traditional semi-structured interviews (see interview guides in Appendices C and D), and semi-structured interviews using prompt cards. I drew frequently mentioned experiences from the Round 1 interviews to prepare cards participants can use to construct a cohesive narrative and think more about making connections between the types of experiences they have had and how their values and behaviors have been shaped by them. When people are given prompts related to original events, recall is improved in both detail and frequency (Wagenaar 1986). This exercise also improved participants’ ability to mobilize abstract concepts such as values, which tend to be difficult to verbalize without context or
feedback (Burgess, Limb and Harrison 1988b). I allowed participants to organize the cards in whatever way made the most sense to their personal experience. While this meant that
quantifying their responses proved difficult, the purpose of the exercise was mostly to spark conversation and introspection about the shaping influences on one’s environmental values, beliefs, and norms; it proved to be an effective tool for this task.
In both rounds, I recruited interviewees by asking visitors at the beginning of the tours if they’d be willing to volunteer to be interviewed. During Round 1, I was only recruiting for the interviews, and 74% of the people I called were interviewed (23 out of 31; Table 2.6). During Round 2, 378 of the learners who filled out the pre-survey included their contact information (66% of participants). Out of those 378, I contacted 67 people, 24 of which ended up being interviewed (36%). The lower Round 2 response rate is likely attributable to the addition of the survey in Round 2: some participants likely filled out the contact information believing it to be part of the survey (not realizing they would be contacted later), and other participants perhaps became disinterested in participating further after completing the survey.
I aimed to conduct participant interviews within two weeks of the LAIE so that the participant’s memory would be fresh. Most interviews occurred between one and two weeks after the LAIE, although due to scheduling difficulties some happened later; all interviews were completed within one month of the LAIE.
Table 2.6: Summary of interview data by site
CTR DLC NCA Total
Traditional semi-structured 7 9 7 23
Semi-structured with card-
sorting activity 8 8 8 24
Mixed-methods survey
During Round 2, I implemented a mixed-methods survey. This survey primarily consisted of pre-existing VBN metrics and demographic/experiential history information. (Appendix E). I developed the survey using experiences from Round 1 data collection, which helped me decide what data to collect and how to collect it. Most importantly, I learned the majority of the questions for a survey needed to be asked before the program. At the end of the programs people were hot and tired, whereas at the beginning of the programs learners were more alert and engaged. Also, learners were told to arrive anywhere from five to 15 minutes early, so I wanted to take advantage of the loitering time before each LAIE. While the choice to front load the survey had clear benefits, the cost was that my sample did not include people who arrived too close to the program start time. If a party arrived less than five minutes before the program began, I did not administer the survey. Because the survey took five to ten minutes, this choice meant that occasionally people were still filling the survey out when the program started. In these cases, either the guide started a few minutes late (waiting until they were finished), the people finished the pre-survey as the guide was doing the introductory spiel, or they would return incomplete surveys.
My participation rate was high at all three sites (Table 2.7). Across all sites, almost 82% of the adult visitors agreed to participate in the survey; of the 18.4% who were not able to take the survey, the majority of those were people who arrived within 5 minutes of the start of the program. Of the learners who chose not to participate, some needed to watch small children, some had not brought their reading glasses, and some faced language barriers. These issues suggest that parents/grandparents with young children, older adults, and non-native English speakers are underrepresented, but due to the overall high rate of participation, these issues have minimal impact on data quality.
The pre-program survey captured the learner’s pre-existing beliefs and previous
experiences in order to explore whether the incoming populations are substantively different at the outset. The first section of the survey asked participants about motivations for attending, previous experiences with animal/environmental facilities, and the characteristics of their party.
Table 2.7. Survey counts and response rates by site.
CTR DLC NCA Total
Matched pre-post 130 133 139 402
Match rate 60.7% 72.7% 74.3% 69.2%
Only pre survey 28 18 13 59
Only post survey 56 32 22 110
Total 214 183 174 571
Response rate (total) 83.2% 77.5% 84.1% 81.6% Response rate (matched only) 55.0% 56.3% 67.2% 58%
The second section of the survey consisted of the Animal Attitudes Scale, or AAS (Herzog, Betchart and Pittman 1991). This metric captures sensitivity to animal welfare issues. This metric does not correspond directly to a variable in the VBN framework, but I chose to include it because the relationship between animal welfare attitudes and environmentalism is not well-studied (Herzog and Golden 2009), but it is possible that animal welfare issues activate a different subset of the population than general environmental topics. The AAS asks participants to rate their level of agreement with 20 animal-rights statements (e.g. “The use of animals in rodeos and circuses is cruel”) on a 5-point Likert scale.
The third section included three different metrics intended to capture different concepts in the VBN framework. First, I used a values orientation scale by de Groot and Steg (2007b)
because it succinctly measures the three dimensions of environmental values (Schultz 2013). De Groot and Steg’s metric is based on Schwartz’s values scale, which consists of 50 or 60 items, but the shortened one only focuses on values dimensions shown to be associated with
three altruistic, and four egotistic, including items like influential, helpful, and unity with nature) and asks participants to rate them on a 9-point valuation scale.
The New Ecological Paradigm (NEP) scale served as the general worldviews metric (Dunlap et al. 2000). The NEP scale is one of the most frequently used environmental metrics in the literature (Stern et al. 1995, Dunlap 2008). While the NEP has been reworked into various different scales, I used the 15-item scale, which is the most reliable scale according to meta- analysis (Hawcroft and Milfont 2010). The items make statements about the relationship between humans and the environment (e.g. “We are approaching the limit of the number of people the earth can support”), and participants are asked rate their level of agreement on a 5- point Likert scale. This scale has four separate dimensions, including the belief in the balance of nature, the idea of ecological limits to growth, fear of ecological catastrophe, and the hierarchy of man over nature (Albrecht et al. 1982, Hawcroft and Milfont 2010).
The third metric in the third section focused on the norm activation theory component of VBN theory (Schultz et al. 2005). Norm activation theory includes awareness of consequences, ascription of responsibility, and behavioral norms (Stern 2000). Thus, I used a metric that covered all of those together. In this metric, participants rate the severity of problems and their responsibility for those problems. The metric I chose had very general environmental issues, with the ‘targets’ being things like water pollution, air pollution, and climate change. While some scholars have argued that the compatibility principle requires the item focus to remain constant throughout a survey (Ajzen and Fishbein 1977), others have found that method biases overrule the need to ensure that the items are consistently focused (Kaiser, Schultz and Scheuthle 2007). In other words, asking repeated questions about the same focal object will mean that
participants will not view each question as substantively different from the rest. Thus, in my survey design, I did not aim for compatibility across metrics.
I originally planned to include an environmental behavior metric on the pre-program survey, but I eliminated it for two reasons: first, the whole survey was too long during pretesting, and second, the available behavioral metrics were either too long to administer in the time before the program (e.g. Kaiser and Wilson 2004), or they were too short, causing construct validity issues (e.g. Dutcher et al. 2007, Schultz and Zelezny 1998) Thus, I decided not to include a behavior metric on the pre-survey.
The final section of the survey collected demographic information. While this study is not hypothesis-driven, learner demographics are likely to have some explanatory power over the VBN variables, so this effect must also be captured. Past research has shown individuals who are young, female, liberal, educated, white, and higher socioeconomic status tend to score higher on a variety of psycho-social environmental measures (Blocker and Eckberg 1997, Van Liere and Dunlap 1980, Jones and Dunlap 1992, Honnold 1984, Palmer and Suggate 1996).
At the end of each LAIE I administered the post-program survey, which was one page long. I marketed the post-program survey as being “much shorter than the first one,” and participation in the post-program survey was higher than the pre-program survey. The post survey included a list of 11 behaviors that might have been suggested or inspired by the LAIE, including items such as donating money, volunteering, or seeking ways to reduce one’s impact on the environment. I generated this list using the behaviors discussed during Round 1 LAIEs and interviews. I also had three free-response questions: “What are the three most important things you learned during your tour?” “Do you think it’s important for this facility to exist? Why or why not?” and “Do you think it is important to conserve these animals in the wild? Why or
why not?” These questions were meant to get a sense of how learners ranked the urgency and importance of the information they received on the LAIE, and how perceptions of animal conservation and institutional activities differed across sites.