Progression through time: Development of birdwatcher careers based on propensity score matching

Abstract Recreation specialization postulates that people progress through a career when engaging in a leisure activity. This has been studied by cross-sectional and few panel studies, but both methods have some shortcomings. Here, I studied German birdwatchers from the 16 federal states of Germany by employing the propensity score matching method. Data were collected using the online research tool SoSciSurvey between 14 February 2020 and 15 June 2020 include study population, sampling, data collection, and analysis in short. Three career stages of the same size were defined according to birdwatching experience in years. Then, respondents were matched according to age, gender, federal state, and educational level but differed in their leisure career stages by propensity score matching, thus birders of the same age, gender, educational level, and federal state were identified that differed in their birding experience in years. Stage 1 was birders with ≤10 years of birding, stage 2 with 11–34, and stage 3 with ≥35 years. This led to 177 pairs in the stage 1–2 comparison and 113 pairs in the group 2–2 comparison. The recreation specialization construct based on the dimensions behavior, skill, and commitment was applied. There was a highly significant effect concerning skill/knowledge and behavior between stage 1 and 2 (p < .001 for both). However, no differences were found concerning personal and behavioral commitment (p > .5). Skill/knowledge increased again between stage 2 and 3 (F = 4.554, df = 112, p = .035), no significant changes concerning the other three dimensions behavior, personal, and behavioral commitments were found (p > .2). The study provides evidence for a career development in leisure.

Dr. Christoph Randler is a professor in the department of biology at the University of Tuebingen. He earned his PhD in 2004 with a study on cognitive and emotional factors contributing to school science education. His research is mainly focused in the area of biological education with a focus on biodiversity, ornithology, and outdoor recreation, recently working on the link between bird diversity and social sciences (learning, citizen science, and health benefits).

PUBLIC INTEREST STATEMENT
The study shows that people who spend their spare time observing birds undergo a kind of career similar to a career in occupation, with an increase in knowledge and, in part, an increase in behavior, e.g., by spending more time afield. The study of birders is important because they provide ample data for nature conservation, and also educate others by leading field trips.

Recreation specialization framework
Recreation specialization is a framework based on the seminal papers of Bryan (1977Bryan ( , 2000. Bryan (1977) formalized the specialization alongside a continuum of a recreational career where people start as beginners with low knowledge and low involvement and end up as highly specialized participants with a high interest and knowledge. Although it is generally seen as a continuum, afterward usually three or four distinct groups have been categorized (Harshaw et al., 2020;McFarlane, 1994;Randler, 2021b). Each level includes different aspects of behavior, commitment, and knowledge (Bryan, 1977). In addition to the recreation specialization framework (Bryan, 1977;Scott & Lee, 2010), the observation that people have careers in their chosen leisure activities is also proposed in the serious leisure literature (Stebbins, 2007).
Further, studies differ in their operationalization and their constructs of recreation specialization. Some studies were based more on behavioral aspects, others more on attitudinal variables (see, Scott & Shafer, 2001). In the current study, birdwatchers were studied, and the concept of Scott and Shafer (2001) was followed, which includes the behavioral dimensions (e.g., number of birding trips and days in the field), a cognitive dimension (related to skill and knowledge about birds), and the dimension of commitment, which is related to the centrality to lifestyle. Later, Lee and Scott (2004) presented a measurement of birding specialization based on this construct. The dimension of psychological commitment encompassed both a personal commitment and a behavioral commitment (Scott & Shafer, 2001). The measurement approach of Lee and Scott (2004) was followed by using the three dimensions as a construct to measure recreation specialization.

Leisure career
A career progress is postulated in different theories and concepts (Scott & Shafer, 2001;Stebbins, 2007). Recreation specialization imposes that some people remain at a given level without further development of their career and some may even give up their leisure activity (Bryan, 2000;Scott & Shafer, 2001). However, when turning to data and studies, Scott and Shafer (2001) reported findings from various sources suggesting that progression is not a typical career path pursued by recreation participants but rather an exception. As different methodological approaches have been used in previous studies, a closer look at the differences is needed (see below).
Progression through time or development of a leisure career has rarely been tested in the previous work. One possibility is to assess the number of years a person is engaged in a specific activity and to relate this to different career stages. Although such cross-sectional analyses cannot infer causality, they give hints on career development. Previous work has shown mixed results, e.g., concerning boaters (Donnelly et al., 1986;Kuentzel & Heberlein, 1997). Studies about birdwatchers revealed changes in motivations across specialization levels, rendering the career hypothesis likely (McFarlane, 1994).
More salient from the research approaches are panel studies that follow the trajectories of individuals and their leisure careers (Kuentzel & Heberlein, 2008) because cross-sectional studies are either correlational or retrospective. Kuentzel and Heberlein (2006) examined the progression hypothesis using a panel of boaters. In contrast to the hypothesis, they found that progression was the exception rather than the rule among the boaters. Moreover, changes in specialization level were often related to life course changes, such as marriage or childbirth (Kuentzel & Heberlein, 2008). Watkins (2010) showed that students engaged in a leisure program changed meanings during the study and that meanings evolved in a logical progression from less to more developed understandings. Scott and Lee (2010) studied American birdwatchers using a panel survey with a five-year wave (between 1997 and 2002). The authors similarly showed that stability or decline were the case, rather than progression. However, panel studies can show trajectories across a given time span in individuals, but they also may not represent the gold standard because of high dropouts or panel mortality. Further, all changes can be simply assigned to age effects, e.g., changes in birding behavior (less or more field trips) may just reflect a decrease in abilities or an increase after retirement (see Discussion in Scott & Lee, 2010).

Birdwatching in modern societies
Birdwatching is an activity that is related to searching for, spotting, and identifying birds (Oddie, 2002). Although considered an individual leisure activity, birdwatchers contribute to the thrive of a modern society. For example, birders contribute to the data collection for citizen science programs by submitting their observations to web-based platforms. Here, the data can be used for the monitoring of breeding birds, and, in turn, for nature conservation (Sullivan et al., 2014). Given some special data collection programs, Bonney (1991) showed that citizens contributed data in an amount worth millions of dollars. If these data had been collected by professional scientists, such projects could not be funded (Bonney, 1991). In addition, birders serve the society because they present lectures to the public or lead field trips, in many instances as unpaid volunteers (Randler, 2021c). Thus, studying birders and their career progress may have an impact on society.

The current study
The goals of this study are to assess progression through time during a leisure career, with birdwatchers as an example that might be generalized. Further, due to the importance of this leisure activity for biodiversity conservation (by data collection) and its impact on education (e.g., leading field trips), the study of birdwatchers seems worthwhile. In addition to addressing shortcomings of previous methods in assessing progression in a leisure career, propensity score matching as a third method was applied in this study. This has not been applied in the previous work in leisure research. When comparing groups of career stages, they may differ in several aspects from another (e.g., age, gender, and others). Therefore, propensity score matching focuses on a subset of data where individuals from two different groups or stages were matched based on age, gender, education, and residence (Dehejia & Wahba, 2002). Here, the participants of a large cross-sectional study on German-speaking birdwatchers (Randler, 2021a) were analyzed based on propensity score matching. Germany was chosen as an example from Europe because most studies have been carried out on the North American continent and only few in Europe. Germany is comparable to the USA at the start of formal scientific ornithology (German Ornithological Society: foundation in 1850; American Ornithologists' Union: 1883) but differs a lot in the development of birding as leisure activity, with Germany being some decades retarded (Cordell & Herbert, 2002;Randler, 2021b). For example, the Christmas Bird Count in the USA was established in 1900 (Larson et al., 2020), and a similar count within the frame of a citizen science project was established in 2011 in Germany (NABU, 2023).

Data collection
Data were collected using the online research tool SoSciSurvey via an open link. On the first page, the aims of the study were explained, and formal consent was requested. Participants had to actively click "yes" to start the study. They were also able to stop and leave at any time. The aim was to cover a wide variety of birdwatchers from different organizations in Germany, from people preferring backyard birdwatching to highly specialized birdwatchers (Randler, 2021a(Randler, , 2021b. This study was granted permission by the ethics committee of the Social Science and Economic Faculty of the University of Tuebingen . The data that support the findings of this study are available on request from the corresponding author.

Survey instruments
Sociodemographic data included age, gender, birding initiation age, education, and the federal state of Germany. Years of birdwatching were calculated by subtracting the birding initiation age from the actual age. Following the construct of Lee and Scott (2004), birding specialization was assessed concerning the three dimensions of skill/knowledge, behavior, and psychological commitment. Skill/knowledge was based on self-reporting of the number of species a person is able to identify by appearance and by song without a field guide; self-assessed knowledge was rated from 1 (novice) to 5 (expert). The behavior scale was related to the number of birding trips taken last year (at least 2 km away from home), number of days spent for birding last year, number of bird species on life list, number of bird books owned, replacement value of the total equipment, and number of species on a national list. Psychological commitment was measured with constructs based on a) personal commitment and b) behavioral commitment. Personal commitment was measured with three questions, e.g., "Other leisure activities don't interest me as much as birding." Behavioral commitment was based on three items, such as "If I couldn't go birding, I am not sure what I would do." All items were measured on a five-point Likert scale. Details of the full measurements and of the origin of the items are depicted in Randler (2021b).

Statistical analysis
Using three time frames (novice, advanced, and specialized birder) following the previous segmentation study that revealed three clusters of birders (Randler, 2021b), birding experience over the years was also divided into three groups. First, three career stages of the same size were defined according to birdwatching years by percentile split (33/66). This led to stage 1 with 10 years of birding or less, stage 2 with 11-34 years of birding, and finally, stage 3 with 35 years of birding and more. Propensity score matching was applied to two comparisons separately, the group 1 versus group 2 scores, and the group 2 versus group 3 scores.
Based on the initial data set (Randler, 2021a), only participants were retained were age, gender, birding initiation age, educational level and federal state in Germany have been available. Then, propensity score matching was applied with birding experience as dependent variable and age, gender, educational level, and federal state as independent predictors in a binary logistic regression. The group probabilities were saved as propensity scores. Then, individuals with the same propensity score but differences in birding experience were matched by the nearest neighbor algorithm. This led to 177 pairs in the stage 1-2 comparison and 113 pairs in the group 2-3 comparison ( Table 1). The pairs provided an exact match, with χ2-values of 0 and p values of 1 for all comparisons. This means that the distribution was identical with the pairs, e.g., pair 1 compared career stage 1 with 2, but both respondents were identical in age, gender, residence (federal state), and education. Afterward, pairs were treated as matched in a linear mixed model, with pair ID linking the subjects and birding experience as repeated measures. First-order autoregression was used as covariance type.
The main effect of time on skill/knowledge between stage 2 and 3 was significant (F = 4.554, df = 112, p = .035; Figure 3), and the autoregression coefficient rho was also significant (Wald Z = 2.120, p= .034). Thus, birders progress in skill and knowledge from stage 2 to stage 3, and in combination with the results above, from stage 1 to 3. However, there were no changes concerning the other three dimensions: behavior (F = 0.333, p = .565), personal commitment (F = 0.648, p = .422), and behavioral commitment (F = 1.296, p = .257). This means that psychological commitment and behavior remain stable throughout the career stage 2 to 3.

Discussion
This study uses propensity score matching to analyze the progression through time hypothesis in German birdwatchers. Conform to the hypotheses, birders increased their skill and knowledge through stages 1-3, and their behavior changed from stage 1 to 2. In contrast, psychological commitment, both personal and behavioral commitment, did not change throughout the birding career. The advantage of the propensity score matching approach is that age as a confounder is removed and that the career stages cover a long time of birding careers compared to the panel studies focusing on 5-10 years.
Similar to this study, Scott and Lee (2010) found changes in skill/knowledge also with an increasing knowledge in relation to a higher career stage. An increase of birding experience with career stage is probable because an accumulation of knowledge on diagnostic traits of birds happens as well as an accumulation of birding trips, which, in turn, should increase the number of bird species seen and heard. Although behavior changed in both studies, it was in a contrasting manner. Fewer trips and fewer days in the field were reported by Scott and Lee (2010), while in this study, behavior increased. Such a decrease in behavior, as observed by Scott and Lee (2010), may be a result of some natural attrition because their sample naturally grow older and respondents in the second wave of their study were 5 years older, and the mean age in the 2002 survey was about 60 years. Differences in behavior may also be owed to the assessment of behavior, which was broader in the present study, covering a wider range of items, in addition to field trips and field days. Using the propensity score matching method, the age effect can be removed in the present study, which may be the reason why behavior increases from stage 1 to 2. The propensity score matching method allows us to compare two persons who have exactly the same age in years but differ in their career stage. All three methods (cross-sectional, panel, propensity score matching) taken together may shed light on progress, stability, or decline of leisure activities, but yet the results remain inconclusive.  From a life-course perspective, the current study could be related to previous work that showed that important life-course events may be related to a change in leisure activities, such as the cessation of an activity due to marriage and childbirth (Kuentzel & Heberlein, 2008). The previous studies that showed a regression rather than a progression (Kuentzel & Heberlein, 2008, Scott & Lee, 2010 were mainly based on already existing participation in an activity. Therefore, it would be an important aspect to study people that take up a new leisure activity. Here, life-course changes were considered only marginal as a trigger of the start of birding (Randler & Marx, 2022). In retrospect, social influence, nature experience, bird-centered triggers, and education (formal/ informal) have been mentioned more often than any life-course events (Randler & Marx, 2022). Thus, life-course changes may be triggers to cease and activity but not to initiate one.
From a general viewpoint, the study supports the view that career progression is taking place in birding, and this may be representative of other leisure activities. Thus, it is suggested that this research is an example for other leisure careers. With societal impact, birding is a leisure activity that produces valuable data about bird species, bird abundance, and seasonal occurrence that can be used for citizen science data projects (Randler, 2022;Sullivan et al., 2014). Within the framework of citizen science, there are different programs concerning birds that are related to different complexities, such as simple tasks like counting birds during 1 hour at the feeder (Alexandrino et al., 2022) and much more complex tasks, such as breeding bird census that require a higher skill/knowledge and more days out in the field (Randler, 2022). Thus, the career progress can be directly related to specific citizen science activities, and the career stage model on which this study is based is reflected in different citizen science programs. Knowing the career stage of individual participants may help tailor citizen science projects more precisely toward participants (Parrish et al., 2019). Further, the career progress model may help in the analysis of retaining people in citizen science programs (Fischer et al., 2021). Finally, the data gathered from birders can be generalized to other nature-related activities, but also to leisure activities themselves in analyzing career paths of volunteers.
Apart from data collection, birders also contribute to the education of the public by giving lectures and presentations and by organizing and leading field trips (Randler, 2021c). Birders usually start leading field trips when they have gained at least basic knowledge (Randler, 2021c), and thus, as shown in this current study, have made some career progress. As biodiversity is declining at a fast pace (Lees et al., 2022), raising awareness for biodiversity conservation is an important task for society. This is implemented in the sustainable development goals of the United Nations (United Nations Department of Economic and Social Affairs, 2023). The goals address lifelong learning (goal 4 quality education), and birders serve their part when leading field trips. Further, citizen science participants contribute to the goal 15 (life on land) because their data can be used for the monitoring of biodiversity and for conservation.
Further studies may therefore be organized into some kind of panel studies that predominantly may focus on beginners of birdwatching to challenge the specialization career framework. This would add to the previous panel studies (Scott & Lee, 2010) because they have focused on already active birders, while research focusing on beginners would be an important addition, which in turn may also help to understand why people drop-out from a leisure activity.

Conclusion
The model of a career process during a leisure activity has been supported by empirical data in this current study. This may help in expanding onto other areas of leisure research. Based on the results of this current study and on previous ones, a career stage model could be developed (Scott & Lee, 2010;Scott & Shafer, 2001). This model shows an increase in skill/knowledge during the career stages from 1 to 3 as shown in this current research (Scott & Lee, 2010). Behavior, however, may be increasing during the stages 1 and 2, but then may decline again (between stage 2 and 3; Scott & Lee, 2010). Commitment remains more or less constant (Figure 4). This can be explained by the fact that commitment is easiest to change, followed by behavior and then by knowledge. Thus, commitment might be already high in stage 1 so that it cannot develop further, while knowledge is rather low at the beginning of a leisure career and has the highest potential for an increase. It is suggested that this model can underlie future studies on career progression in a wide range of leisure activities. Apart from the individual career development, studying this progress is also important when addressing citizen science projects and educational purposes. Future studies may help to identify critical points during a leisure career when citizen science needs more support or possibilities to carry on with a lower engagement level rather than dropping out of the activity. As citizen scientists provide a wealth of data, e.g., for nature conservation (Bonney, 1991), this has an impact on society, and there is a need for a policy that does not only identify characteristics to involve people into citizen sciences but also tools and means to keep them integrated in such projects.
From a research focus, the method of propensity score matching might be introduced in measuring progression through time when only cross-sectional samples are available, but also more longitudinal studies are needed, especially to accompany people if and when they engage in a new leisure activity.