Conservation concern among Australian undergraduates is associated with childhood socio-cultural experiences

1. Fostering widespread concern for conservation problems requires a robust understanding of the life experiences that positively contribute to an individual's conservation attitudes and behaviours. However, few studies have assessed a com prehensive range of social and experiential predictors of conservation concern. 2. Using survey responses from 391 undergraduate students enrolled in various course disciplines across Australia, we describe the relationships between five major constructs of early-life experiences and two measures of conservation con cern: a preference for a conservation career and positive conservation attitudes. 3. We find that conservation career preferences are positively associated with childhood preferences for nature-related books, movies and school subjects, bio spheric family value orientations and environmental volunteering. Conservation attitudes were positively associated with biospheric family value orientations and environmental volunteering. 4. Both constructs were negatively associated with egoistic family values and child hood experiences of nature were not a significant factor in either of our models. 5. This suggests that limited nature experiences do not necessarily impede the de velopment of conservation concern in young Australians, and that family values and experiences with environmental organizations, nature-related storylines and school subjects may help to foster greater conservation concern.

In this study, we explore the role of five major life experience dimensions for predicting conservation concern, drawn from the environmental psychology and significant life experience literature. The first of these is Childhood nature experiences, which are the physical experiences of play or immersion in nature that are consistently cited as salient influences in an environmentalist's life course (Bixler et al., 2002;Chawla, 1999;Easton et al., 2009;Ewert et al., 2005;Palmer et al., 1999;Vadala et al., 2007). Second is Nature-related cultural narratives, which describe the emotional connections that children develop to storylines communicated through books and media. A childhood connection to cultural narratives that communicate value for, and awareness of, the natural world may be one ingredient in developing conservation concern (Chawla & Cushing, 2007;Deruiter, 2002;Finger, 1994;Hsu, 2009).
The third dimension is childhood involvement in Environmental organizations (Chawla, 1999;Hsu, 2009;Place, 2004). In adults, membership in environmental groups is associated with readiness to take pro-environmental action (Fielding, McDonald, & Louis, 2008;Olli, Grendstad, & Wollebaek, 2001), and it is possible that childhood memberships could serve a similar function. Fourth, we consider the role of formal Environmental education in schools, in particular student perceptions of environmental problems as defined in part two of the 1977 Tbilisi Declaration (McComas, 2014;UNESCO, 1978). Environmental education is an important variable, as it is known to produce measurable increases in nature connectedness and conservation attitudes in children (Ballantyne, Fien, & Packer, 2001;Garner, Taft, & Stevens, 2015;Leisher et al., 2012;Liefländer, Fröhlich, Bogner, & Schultz, 2013).
Lastly, the environmental psychology literature points to the importance of values as predictors of pro-environmental behaviour (Finger, 1994;Fransson & Garling, 1999;Stern, Dietz, Abel, Guagnano, & Kalof, 1998). Value orientations, which are defined as guiding principles in a person's life, fall into one of three categories: biospheric, altruistic and egocentric (Schultz, 2001). Biospheric values are based on the core belief that nature is intrinsically valuable, and prior research has discovered that biospheric value orientations are positively associated with conservation behaviours, beliefs and intentions (De Groot & Steg, 2008;Stern, 2000;Stern & Dietz, 1994), whereas egocentric values are linked to less pro-environmental action (De Groot & Steg, 2008;Munroe, 2003;Stern, 2000;Stern & Dietz, 1994). Although not tested previously, it is possible that Family value orientations, which are the value orientations held by an individual's close family, might influence a child to express high levels of conservation concern in adulthood.
Despite the significant research interest in the relationship between early-life experiences and adult environmentalism, no study has yet evaluated the full range of possible predictors of conservation concern, whilst simultaneously accounting for common sources of bias in study design. For example, many significant life experience studies are based on open-ended retrospective reports of persons with a mean age over 30 years (Chawla, 1999;Hsu, 2009;Tanner, 1980;Wells & Lekies, 2006). This is problematic because some researchers strongly caution against the use of retrospective reports (Henry, Moffitt, Caspi, Langley, & Silva, 1994) because people tend to recall their past attitudes as being the same as their current ones, even though they change significantly over time (Koriat, Goldsmith, & Pansky, 2000). To limit the influence of retrospection bias in this study, we have selected a young sample group with a mean age of 21 years. Social desirability bias ( Van de Mortel, 2008) can also influence the honesty of responses when participants are aware of the hypothesis of the study (Vining & Ebero, 2002), as is the case when studies sample only environmentalists and ask open questions that reduce the transparency of the research, such as 'what are the childhood experiences that influenced you to take action for the environment?' (Chawla, 1999;Hsu, 2009;Li & Chen, 2014;Tanner, 1980). In the current study, we attempt to reduce social desirability bias by including distracter questions to reduce study transparency (John, Edwards-Jones, Gibbons, & Jones, 2010). Finally, since many prior studies have limited their samples to environmentalists, we consider it important to consider the life experiences of persons from a variety of career interests who may be engaging in pro-environmental behaviour (Tanner, 1998).
Here we seek to determine the relative contribution of different early-life experiences to predicting (a) conservation concern in early adulthood and (b) intention to follow a conservation related career path. We examine the role of family values, alongside the physical and educational experiences that are more commonly studied. This integrative approach helps unite psychological and phenomenological bodies of literature and explain their relative influence on conservation concern.

| Participants and procedure
Almost a third of all respondents (29.2%; N = 114 students) were enrolled in conservation related courses such as wildlife biology or environmental management, and the remainder (70.8%) were studying in non-conservation fields. Only 15.3% (N = 60) of our sample population spent most of their childhood outside Australia.
The first section related to demographics and our first dependent variable: conservation career choice. The subsequent five sections evaluated level of exposure to the life-experience constructs, and the last section concerned our second dependent variable: conservation attitudes (measured by the Nature Relatedness-Perspective sub-scale).
Six-point Likert-type scales were chosen to measure the three constructs of early-life experiences: nature experiences, environmental education and family values, for their increased sensitivity, elimination of the neutral point and normal distribution (Leung, 2011). Cultural narratives, environmental organizations and favourite subjects/ teachers were measured either by multiple choice or open-ended questions.
To verify the content and face validity of the survey instrument, three experts in survey design were consulted and the instrument was modified according to their suggestions. A pilot test at a local wildlife sanctuary (N = 10) helped us assess the length and clarity of the questionnaire and resulted in slight improvements to the wording.  (Nisbet, Zelenski, & Murphy, 2009). The NR-perspective sub-scale evaluates an eco-centric worldview and is assessed with seven items. The mean of these items had adequate internal consistency (Cronbach's α = 0.71).

| Nature and animal play
Firstly, respondents were asked about the frequency of their childhood experiences with ten different urban and natural play environments on a 6-point Likert-type scale (1 = never, 6 = most days). Nature play items were adapted for an Australian audience from an established scale by Bixler et al. (2002), to include relevant landscapes such as backyards, beaches and bushland. Principle axis factoring using the varimax rotation revealed that three nature play items ('Paddock', 'Bush' and 'Pond') positively loaded onto Factor 1, which had an eigenvalue of 2.54 and accounted for 36.3% of the variance.
Factor 2 comprised three items representing Urban play areas, including 'Skate park', 'Laneway' and 'Park'. 'Indoors' was removed due to lack of variation, and 'Courtyard' excluded because it loaded onto two factors. Since the final Urban play scale was discovered to have low internal reliability (Cronbach's α = 0.59), we discounted this construct from further analysis.
Although we originally conceptualized nature and animal play as separate variables, the two scales were highly correlated. We therefore combined them to form one scale assessing the extent to which respondents engaged in nature and animal play in their childhood.
Factor 1 comprising Nature play items was combined with the Animal interactions scale, which measured the frequency of childhood interactions with 5 different types of animals (e.g. native wildlife, farm animals; 1 = never, 6 = most days). Three items, 'Beach', 'Backyard' and 'Zoos', were removed from this analysis because they either lacked variance, were highly skewed or did not load with other items in the factor analysis (respectively). The mean of the final six items formed our Nature and animal play scale, that had adequate internal reliability (Cronbach's α = 0.87).
Family holiday experiences were measured with ten items, assessing the frequency of family visits to environments that were predominantly urban (e.g. 'City', 'Theme park') and natural (e.g.  Cultural narratives were assessed using two measures called Nature books and Nature movies. The first measure, relating to television shows and movies, requested that respondents identify three categories of media which they watched most frequently prior to 10 years of age. Of ten possible choices, two items related to the environment or wildlife ('Natural world' and 'Pets and farm animals') which were used to assess preferences for nature-related storylines. Using the same format, we assessed favourite books and bedtime stories, using two of eight possible items to identify a preference for nature-based cultural narratives ('Kids adventuring in nature' and 'Books about animals'). For both measures, respondents were free to offer alternatives using the 'Other' response option. All responses were coded so that nature-based preferences were designated as 1 and non-nature-based preferences were coded as 0. We then computed new variables by summing response scores, the final measure therefore ranging from 0 to 4.
Environmental organizations were assessed using two items. The first assessed involvement in Outdoors groups and asked about clubs and societies that participants were involved in during childhood.
Respondents selected all applicable options out of ten possible activities that included three focal items reflecting nature orientation ('Guides and scout groups', 'Environment clubs' and 'Wilderness skills').
The second question, assessing Environmental volunteering, asked respondents to self-report any long-term volunteering they had undertaken across seven categories of organizations, with the focal items being 'Animal shelters' and 'Environmental groups'. For both measures, the choice of an 'Other' option was available and these open-ended responses were later coded into appropriate categories. A score of 1 was given to each environmentally related group, yielding a total score that could range from 0 to 3 for Outdoors groups, and 0 to 2 for Environmental volunteering.
Environmental education was assessed by asking respondents to self-assess the quality of their formal school education relating to 'International politics', 'Humanitarian issues', 'Environmental problems' and 'Science & technology'. These items were measured using a sixpoint Likert-type scale (1 = extremely inadequate, 6 = extremely adequate). Of these four categories, 'Environmental problems' was our key focus (UNESCO, 1978), whilst the remaining three subject areas were included to reduce transparency about the aims of the study.
Participants were also asked to report their Favourite subject and the subject taught by their Favourite teacher (Place, 2004 (Schultz et al., 2005), only these two value orientations (and not altruistic values) are considered in our analyses.
We found that the internal reliability of both the Biospheric family values and Egoistic family values sub-scale measures were satisfactory (Cronbach's α = 0.92, 0.72, respectively). In addition to family value orientations, we also measured injunctive and descriptive family social norms in our survey instrument. However, we did not include these measures in the final analyses, as there was a high correlation between the injunctive norm, descriptive norm and biospheric family value orientation measures (r(390) = 0.713, p < 0.001). We chose to include values rather than norms, since the former are expected to be more stable and more likely to have a formative influence.
Typical demographic control variables such as age, gender and country of childhood residency were collected at the beginning of the survey, since these factors are likely to influence environmental attitudes and behaviour (Fransson & Garling, 1999).

| Statistical analysis
Pearson's correlations were computed to explore bivariate relationships between our dependent variable of conservation attitudes (NR-perspective), and the 16 measures of early-life experiences.
This allowed us to correct for any instances of multicollinearity, or high correlations between our independent variables (Pallant, 2013 Prior to conducting these analyses, we ensured that both models satisfied standard parametric assumptions, including those of normality, linearity and homoscedasticity. We confirmed that the tolerance of all variables was above 0.10, and variance inflation factor was below 10 (Pallant, 2013). Statistical analyses were completed in 2018 using IBM SPSS Statistics 25.

Of our predictor variables, only Urban family experiences and
Environmental education in school were not significantly related with conservation attitudes as measured by NR-perspective (Table 2)

| Conservation attitudes (NR-perspective)
The results of the hierarchical multiple linear regression are presented in Table 4. Prior to running the model, preliminary analyses were conducted to ensure the data conformed to the assumptions of normality, multicollinearity, linearity and homoscedasticity. Ten residual outliers which exceeded the maximum Mahalanobis distance or Cook's distance were removed in this process.
A summary of the proportion of variance explained at each step is shown in  Table 3 for the full list of predictor variables in each model.

TA B L E 5
The model fit summary of the hierarchical linear regression model F I G U R E 1 A summary model of the early-life experiences associated with the two aspects of conservation concern, with unstandardized beta coefficients (B)

Conservation concern
Conservation attitudes Conservation career  . Notably, we found that social experiences had a far stronger relationship with conservation concern than childhood nature and animal play and family nature experiences.

| D ISCUSS I ON
Prior research has reported that physical experiences in nature predict adult environmentalism (Chawla, 1999;Wells & Lekies, 2006), pro-environmental behaviour (Li & Chen, 2014;Palmer et al., 1999;Tanner, 1980) awareness (Palmer et al., 1998), and career choice (James, Bixler, & Vadala, 2010;Place, 2004;Sward, 1999). However, among the comprehensive range of variables we considered, the frequency of physical experiences with animals and nature, including family nature holidays, were not significant predictors of either measure of conservation concern. This finding is reflected in some previous studies (Olli et al., 2001;Reibelt, Richter, Rendigs, & Mantilla-Contrearas, 2017), including a comprehensive examination of over a thousand young Americans, which found that growing up surrounded by nature did not foster stronger environmental attitudes or intellectual interests (Bixler et al., 2002). A recent and comparable study of French undergraduates (N = 919; Prévot et al., 2018) found that childhood experiences in natural environments were only a significant factor before all other variables in their model were considered. Similarly in a Taiwanese study, Hsu (2009) demonstrated that childhood nature experiences were one of the most significant factors in t test comparisons, but not in regression analysis alongside a range of other variables. In mediation analyses performed on German university students (N = 131), Pensini, Horn, and Caltabiano (2016) likewise showed that childhood nature exposure was only indirectly related to ecological behaviour and connectedness to nature, through its relationship to other variables.
This evidence suggests that relationships between childhood nature exposure and adult conservation concern might actually be rather small, and perhaps indirectly mediated by other factors.
If childhood experiences of nature are not a critical factor in the development of conservation concern, this would be a promising finding, given that 91% of the Australian population and 68% of the world's populace are predicted to reside in highly urbanized areas by 2050 (UNESA, 2019). However, further research with different populations and in different settings is needed to confirm this possibility. Although we attempted to measure childhood interactions with urban nature including backyards, parks and beaches, we found that frequency was not a strong indicator as our responses on these items were highly skewed towards 'most days', suggesting that most children passively experience these environments on a regular basis. This is a considerable limitation in our current study, given the growing body of literature highlighting that a lasting love and affinity for nature can develop from childhood interactions in public green spaces and private gardens Whitburn, Linklater, & Milfont, 2019). In future, it will be important to go beyond measuring the frequency of nature experiences by also measuring the quality of a diverse range of nature experiences, including active nature experiences such as gardening, tree planting and observing wildlife.
In our analysis, the strongest predictor of conservation concern (represented by both career choice and environmental attitudes) was environmental volunteering with animal shelters and environmental groups. Intriguingly, childhood membership in clubs and societies, specifically environmental clubs, scouts and other wilderness groups, was not a significant factor in either analysis. Given this, we hypothesize that the duration of involvement could be a defining factor, as our question specified 'long-term' involvement in environmental volunteering, but not for clubs and societies. A strong association between conservation concern and environmental volunteering was anticipated, as involvement in environmental organizations has been previously identified as an important driver of conservation concern (Fielding et al., 2008;Olli et al., 2001). Several mechanisms explain the positive effects of membership in environmental organizations, including the increased sense of normative support (Gockeritz et al., 2010), personal efficacy (Lubell, 2002) and connectedness to nature (Guiney & Oberhauser, 2009) that it provides. Given the distinction between childhood nature experiences and outdoors groups in our study, it is important to note organizational experiences can also provide opportunities to physically interact with and experience nature. That is perhaps why a significant relationship exists between these constructs in our Pearson's correlation analysis (R < 0.18, p < 0.01). Additionally, it should be noted that volunteers can be motivated to join conservation organizations because of existing values and beliefs that developed earlier in life (Ryan, Kaplan, & Grese, 2001). Nevertheless, our findings hint at the strong association between environmental volunteering and conservation concern, suggesting that the combination of outdoors experiences, environmental education and social opportunities these organizations provide could be important.
Cultural narratives were significantly linked to career choice, since Australian undergraduates that preferred conservation careers were almost twice as likely to have favoured storylines that feature animals and nature as children. This was expected, as environmental educators are known to be strongly influenced by their childhood preferences for fictional nature stories (Freestone & O'Toole, 2016).
However, cultural narratives did not significantly contribute to conservation attitudes in our study. This is an intriguing result, as prior research has found that watching nature films and reading about the environment is strongly associated with the development of biospheric values (Eagles & Demare, 1999) and pro-environmental beliefs (Ewert et al., 2005). Our findings instead reflected those of Hsu (2009), who showed that books and authors only exerted a significant effect when environmental activists were compared to an apathetic group. Although the current evidence base lacks satisfactory explanations for this nuance, we hypothesize that while conservation attitudes are likely to develop from complex socialization experiences and understanding of environmental problems, career choices may be motivated by desires to emulate a character or setting that was prominently featured in a treasured childhood storyline.
Preference for conservation career, but not conservation attitudes, was significantly associated with the subjects and teachers our undergraduate study group preferred in school. As it has been previously shown that students have greater aptitude in their favourite subjects (Raza & Shah, 2011), it is logical that students may choose careers based upon disciplines they excel in, or that they might be inspired by a favourite teacher to follow a certain career path. However, it is harder to explain why adult conservation attitudes were not associated with quality of formal school education on environmental problems, since previous studies have reported strong effects of environmental education (Ballantyne et al., 2001;Leisher et al., 2012). One potential explanation of this finding is that our measure assessed formal environment education in school, which is standardized through the curriculum. Future studies could consider less formal sources of environmental education such as social media or television (Deruiter, 2002;Lee, 2011;Vadala et al., 2007), to better reflect the variety of ways young people receive environmental education.
Our findings also showed that the conservation attitudes and career preferences of Australian undergraduates were strongly influenced by the value orientations of their immediate family. As previously demonstrated by Schultz et al. (2005), we found that biospheric values positively predicted conservation concern, while egoistic values was negatively related. The strength of this relationship was unsurprising given the known importance of parental attitudes and behaviours on conservation concern (Evans et al., 2007), which has also been confirmed by one of the few longitudinal studies in this field (Evans, Otto, & Kaiser, 2018). This evidence supports the idea that biospheric ethics may develop in childhood from socialization experiences within the immediate family unit. Likely, these are experiences that deepen connection, understanding and appreciation of nature, as opposed to simply increasing the proportion of time of spent in nature (Martin & Czellar, 2017). Our findings complement the widely accepted theory that conservation concern is driven by values (Fransson & Garling, 1999), and that families demonstrating biospheric values will increase their child's likelihood of developing conservation concern (Broom, 2017;Deruiter, 2002).
Our research suggests that conservation concern may develop during childhood, from a varied and interrelated sequence of social, audio-visual and educational experiences of nature. We find that childhood play experiences with animals and in nature were not significant predictors of conservation concern in Australian undergraduates. We consider that these results may reflect the issue of generational change in significant life experience research, likely driven by global declines in the amount of time that children spend outdoors (Louv, 2008;Pergams & Zaradic, 2008). In Australia, there is strong evidence of a generational shift away from unstructured outdoor play (Laird, McFarland, & Allen, 2014), but despite this, levels of environmental concern in Australia appear to have grown over time (Markus, 2019). Based on our findings, we attribute this growth to the critical role of socio-cultural experiences, and increased use of innovative audio-visual tools to promote conservation concern.

| Limitations and future research
While we cannot demonstrate causality in this correlational study, we have attempted to shed light upon the underlying predictors of conservation concern measured through career choice and environmental attitudes. In this study we considered a comprehensive range of psychological and phenomenological predictor variables in our analysis, which explained a greater proportion of variance in conservation attitudes (52.5%) and career preferences (37%-54%) than many other published studies (e.g. 14% in Ewert et al., 2005, 13% in Wells & Lekies, 2006. By selecting a younger sample group and masking the intent of our research, our survey methodology also minimized the issues of retrospection and social desirability biases, which may have improved the reliability of our findings considerably. Importantly, we note that our research remains subject to two primary limitations, which are limited generalizability and the scope of our measures.
One limitation of our research stems from our narrow sample group, which is a significantly female biased Australian undergraduate cohort. Though female bias is common in the life course literature (Arnold, Cohen, & Warner, 2009;Hsu, 2009;Nisbet et al., 2009;Pensini et al., 2016;Villacorta et al., 2003), it has potential to affect the external validity of our results, as women are known to have more positive environmental beliefs than men (Broom, 2017;Dietz et al., 1998;Leppänen, Haahla, Lensu, & Kuitenun, 2012;Prévot et al., 2018). It is also possible that the views and values of the undergraduate cohort may differ from the wider youth population, particularly socioeconomically disadvantaged cohorts. Finally, some studies have shown that the factors implicated in conservation concern are similar but not identical across cultures (Palmer et al., 1998(Palmer et al., , 1999, and so it is unlikely that the physical and cultural context of our study will generalize to the global population at large. Consequently, we identify cross-cultural comparison as a fruitful opportunity for future study. In addition to replicating our study with a more diverse sample, we also recommend that researchers adapt our survey instrument to include a broader scope of nature experiences and environmental education experiences. As our survey participants frequently experienced urban nature as children, some of our measures lacked response variance, resulting in our final Nature and animal play scale comprising a particular and limited set of experiences in natural environments (Table 1). Though our Family nature experiences scale was considerably broader, a key limitation of our study is that it could not adequately capture nature play experiences that occur in environments such as backyards, urban parks and zoos. In future, we recommend assessing active nature experiences (Lohr & Pearson-Mims, 2005) that may occur in these common environments, for example gardening, bird-watching or bug-spotting, which may reflect a higher quality of nature interaction than passive nature exposure alone. In addition, we suggest expanding the environmental education measures beyond the core curriculum, to include alternative mediums such as social media. Finally, we recommend the addition of a dependent variable measuring pro-environmental behaviours (PEBs) in childhood and later life. This is important as conservation behaviours are not always strongly aligned with conservation attitudes (Broom, 2017), and understanding of how we can better promote desirable behaviours will greatly assist us in alleviating conservation problems (Nilsson, Fielding, & Dean, 2020).

| Conclusion
We have shown that among Australian undergraduates, conservation attitudes and career preferences are linked to a varied and interrelated sequence of experiences in early life. Among the dimensions of childhood experiences we explored, socio-cultural experiences of nature, namely biospheric family values and long-term environmental volunteering, were the strongest determinants of conservation attitudes and careers. Limited childhood experiences of natural environments and animals did not prevent participants from having high levels of conservation concern. Our findings suggest that limited access to nature experiences might not necessarily reduce the level of conservation concern in a population, so long as we continue to promote biospheric values and engage young people in an alternative range of educational, organizational and audio-visual nature encounters.

CO N FLI C T O F I NTE R E S T
Nothing to declare.