Citizens as Scientists

Abstract

Public participation in science is burgeoning, yet little is known about factors that influence potential volunteers. We present results from a national survey of 1,145 marine users to uncover the drivers and barriers to a sightings-based, digital marine citizen science project. Knowledge of marine species is the most significant barrier and driver for participation. Many marine users perceive that they have insufficient knowledge of marine species to contribute to the project, yet they expect to learn more about marine species if they were to participate. Contributing to scientific knowledge is also a strong driver for many marine users to participate.

Keywords

citizen science public participation science engagement theory of planned behavior

Introduction

Across the globe, public involvement in research is actively encouraged in many scientific fields. This method of doing “science by the people” (Silvertown, 2009) is often called citizen science (although a number of terms exist). The extent of public responsibility, control, and involvement reflects a number of different approaches to public participation in scientific research, giving rise to typologies of citizen science (Shirk et al., 2012). In what Shirk et al. (2012) describe as contributory projects, scientists request public assistance often to collect or analyze data. Some projects attempt to answer scientific questions (e.g., the astronomy-based Galaxy Zoo project [www.galaxyzoo.org] asks volunteers to classify images of galaxies to understand how they were formed), while other projects conduct long-term environmental monitoring (e.g., www.reeflifesurvey.com), transcribe historical documents (e.g., www.weatherdetective.net.au), or contribute to scientific databases of life on earth (e.g., www.questagame.com).

Several projects, such as Galaxy Zoo, have had success in attracting large numbers of volunteers, but most achieve more modest levels of public assistance. The number of volunteers needed is dependent on the nature of the activities (e.g., whether volunteers are required to be on site, need training, or simply contribute in an opportunistic way) and the temporal and spatial scale of the project (e.g., whether the project is local, regional, national, or global and whether it is conducted as a single event or as ongoing research). Projects in which contributions are made to existing databases (without volunteer training) can be termed opportunistic citizen science. In these projects, volunteers contribute data as they encounter the objects of research interest (e.g., birdwatchers are encouraged to send sighting details to the eBird project via www.ebird.org). Opportunistic citizen science has been greatly enabled through technology such as digital cameras and mobile devices. We consider this method of citizen science to present the greatest opportunity to involve large numbers of the public in scientific research, no matter their background. In addition, opportunistic citizen science may gain volunteers who would not normally contribute to scientific research but may engage through a shared interest outside of science (e.g., fishing). This article investigates factors that influence public contributions to opportunistic citizen science. Insights into these factors are essential for the design and promotion of citizen science to wider audiences.

The challenge for citizen science projects, particularly those operating at large temporal and spatial scales, is to develop effective ways of recruiting volunteers. This is made more difficult by the lack of information and understanding of the drivers and barriers to public participation, which is essential for the long-term success of citizen science projects (Measham & Barnett, 2008). Recent citizen science studies show volunteer motivations include the following: making a contribution to scientific research, learning new skills, interaction with others, environmental concern, altruistic reasons, personal satisfaction, public recognition, education of others, or simply that a project aligns with their interests (Crowston & Prestopnik, 2013; Curtis, 2015; Johnson et al., 2014; Land-Zandstra, Devilee, Snik, Buurmeijer, & van den Broek, 2016; Raddick et al., 2013; Thiel et al., 2014). It is difficult to ascertain which of these motivations are most important due to inconsistencies in the research questions and methodologies. One commonality among these studies is all respondents have volunteered in citizen science. This leaves a gap in our knowledge about the barriers and drivers for potential volunteers in opportunistic citizen science projects, that is, those who have never participated before. Here we apply the theory of planned behavior (TPB; Figure 1), developed by Ajzen (1991) and Fishbein and Ajzen (2010), to explore predictors of Australians’ intentions to contribute to an opportunistic citizen science project. We focused on marine citizen science since it is thought that marine science will inspire the public to become more engaged with science generally, through the strong associations Australians have with the beach and ocean (Department of Industry, Innovation and Science, 2010; Department of Industry, Innovation, Science, Research and Tertiary Education, 2012).

Figure 1.

Theory of planned behavior.

Specifically, we used TPB measures (explained below) to examine public willingness to submit photographs of uncommon marine species to a hypothetical marine citizen project, which was modeled on Redmap (the Range Extension Database and Mapping, or Redmap, project asks the public to submit photographs of marine species that are potentially shifting their normal range, hence are “uncommon” in a particular area. The information will help determine which species are shifting their normal distribution in response to changes occurring in the marine environment (e.g., warming waters around Australia); see www.redmap.org.au). Although Redmap is a national-level project, it is not yet widely known among marine users in all regions and sectors. To avoid distraction from the core questions about willingness to participate, the hypothetical project was presented to respondents rather than explaining Redmap in detail. This approach was also expected to remove any bias from other influences on the decision to participate (e.g., imagery, design, which institution was behind the research, etc.). According to TPB, a behavior should be defined in terms of four key elements: action, target, context, and time. The definition of the behavior we are investigating is as follows: a person logging (action) a sighting (target) of an uncommon marine species (context) sometime in the next 12 months (time).

By uncovering factors that influence people’s intention to submit a sighting, we can suggest strategies for effective communication and project design likely to result in greater volunteer recruitment. Of particular interest are the beliefs held by the target audience about the behavior, and differences in the strength of these beliefs between people who have contributed in the past, and those who have not. These beliefs and differences identify drivers and barriers for recruitment and retention of volunteers.

Theoretical Framework

The TPB is one of the most widely used theories for understanding human behavior (Figure 1). It has been applied in many fields, particularly in health sciences, in public safety, and for encouraging proenvironmental behavior (Armitage & Conner, 2001; Darnton, 2008; Fishbein & Ajzen, 2010). TPB has also been used in science communication research to understand scientists’ willingness to become actively involved in public engagement activities (Poliakoff & Webb, 2007).

To illustrate the theory in further detail, we provide an example of an environmental communication campaign to encourage recycling at a university food hall (Table 1). According to TPB, a person’s intention can predict a specific behavior. The theory acknowledges that a positive intention does not necessarily result in performance of the behavior, and other factors (e.g., infrastructural constraints) may prevent the intended behavior. However, intention has been consistently found to be the strongest predictor of behavior in TPB studies (Armitage & Conner, 2001).

Table 1.

Illustration of the Theory of Planned Behavior (TBP).

TPB dimensions	TPB subdimension	Example
Context of example	An environmental communication campaign to encourage recycling at a university food hall
Desired behavior	Staff and students sort their rubbish into the recycle bins during term
Intention	n/a	In our example, this is the intention to sort rubbish into the appropriate recycle bin.
Attitude toward the behavior	Instrumental attitude	For example, a student may think that sorting his or her rubbish into the correct bin is a good thing for environmental health.
Attitude toward the behavior	Experiential attitude	Instrumental and experiential attitudes can sometimes be inversely related. For example, although people think sorting rubbish is a good thing (positive instrumental attitude), they may also think it will be unpleasant removing their food scraps and placing them into a smelly compost bin (negative experiential attitude).
Perceived norms	Descriptive norms	Piles of rubbish left lying around the food hall may lead to the perception that “everyone else” litters.
Perceived norms	Injunctive norms	This might be expressed as “Most people think I should sort my recycling into the correct bin.”
Perceived behavioral control	Capacity	A person might think, “It is easy for me to sort my rubbish into the correct recycling bins.”
Perceived behavioral control	Autonomy	If there are no recycling bins provided people would be unable to recycle. In this example, the degree of autonomy would be very low.
Behavioral beliefs	Belief strength and outcome evaluation	A person might believe their recycling efforts would result in a tidier campus, but that recycling does not actually have a positive impact on environmental health.
Normative beliefs	Belief strength and motivation to comply	Students may believe that staff would approve of their recycling efforts, but other students do not care.
Control beliefs	Belief strength and power of control	These are beliefs about the resources and opportunities to perform the behavior, for example, the belief that more recycling bins in convenient locations may help increase recycling or that signage is inadequate to determine what types of rubbish can be recycled.

The antecedents to intention are a person’s attitude toward the behavior, his or her perceived norms, and his or her perceived behavioral control (Figure 1). In TPB, attitude is composed of two types (or subdimensions) of attitude: instrumental and experiential. Instrumental attitude is distinguished by the cognitive nature of the construct. Instrumental attitudes reflect a person’s thoughts on whether the behavior is a good (or wise, valuable, etc.) thing to do. Experiential attitudes are more affective in nature, that is, they reflect the person’s evaluation of the experience of performing the behavior. Such attitudes may be described by terms such as painful, exciting, unpleasant, or fun. It is possible for instrumental and experiential attitudes to be inversely correlated.

The next antecedent is perceived norms, which represent the social pressures to perform the behavior. Pressure may come from people who are considered to be important in a person’s life (e.g., family, friends, colleagues), and it extends to opinion leaders and celebrities, or in some situations to people who are absent. Perceived norms can also be divided into two subdimensions: descriptive and injunctive. Descriptive norms refer to what others do are perceived to do in the context of the behavior under investigation. Injunctive norms refer to a person’s perception of what behavior others approve of.

The final antecedent to intention in TPB is perceived behavioral control (PBC), which is the degree of control a person feels he or she has over performing the behavior. Like the other antecedents, Fishbein and Ajzen (2010) identify two subdimensions: capacity and autonomy. Capacity refers to the ease or difficulty a person expects in performing the behavior, and autonomy relates to the person’s perceived degree of control over performing the behavior, that is, to what extent the decision to perform the behavior rests with him or her (and not others, or other factors). The different levels of attitude, perceived norms, and PBC are measured by the researchers using scale items (see Table 2 in Method section), and since these measures are asked as direct questions they are referred to as direct measures.

Table 2.

Survey Item Mean Scores, Standard Deviations, and Reliability of Scales.

Item code ^d	Survey question	M	SD
	Behavioral beliefs: Belief strength (correlation with direct measure composite = .509, p < .001); Question: How likely do you think the following will be if you log an uncommon marine species in the (website/mobile app)? [Scale: 1 = unlikely, 7 = likely]
IABS1	• It will increase scientific knowledge	5.84	1.08
IABS2	• It will increase my own knowledge of marine species	6.10	1.03
IABS3	• It will provide information for the greater good/everyone	5.85	1.02
IABS4	• It increase public awareness of the marine environment	5.63	1.15
IABS5	• It will help to protect/manage the marine environment	5.75	1.19
	Behavioral beliefs: outcome evaluation; Question: How good (positive) or bad (negative) do you think the following things are?[Scale: 1 = a very bad thing, 7 = a very good thing]
IAOE1	• Increasing scientific knowledge is:	6.79	0.55
IAOE2	• Increasing my own knowledge of marine species is:	6.54	0.71
IAOE3	• Providing information for the greater good/everyone is:	6.47	0.80
IAOE4	• Increasing public awareness of the marine environment is:	6.65	0.68
IAOE5	• Helping to protect/manage the marine environment is:	6.69	0.72
	Attitude toward behavior: Instrumental and experiential (Cronbach’s α = .934); Question: For me, logging a sighting of an uncommon marine species onto the [website/mobile app] would be:
DAIN1	• worthless: 1: 2: 3: 4: 5: 6: 7: valuable	6.20	1.15
DAIN2	• unimportant: 1: 2: 3: 4: 5: 6: 7: important	6.21	1.10
DAIN3	• unproductive: 1: 2: 3: 4: 5: 6: 7: productive	6.13	1.14
DAEX1	• boring: 1: 2: 3: 4: 5: 6: 7: interesting	6.16	1.19
DAEX2	• unenjoyable: 1: 2: 3: 4: 5: 6: 7: enjoyable	5.90	1.23
DAEX3	• aggravating: 1: 2: 3: 4: 5: 6: 7: satisfying	6.23	1.11
	Normative Beliefs: Motivation to comply^a,b; Question: If your____ (replace with item below, e.g., “family”) think you should/should not log uncommon species, how much do you want to please them by doing what they think you should do? [Scale: 1 = not at all, 7 = a great deal]
INMC1	• family	2.37	1.72
INMC2	• close friends	2.22	1.64
INMC3	• recreational peers (i.e., other people, not close friends, who do your favorite marine activities with you)	2.15	1.62
INMC3	Perceived (injunctive) normb (Cronbach’s α = .850); Question: How much would your ____ (replace with item below e.g., “family”) approve or disapprove of you logging uncommon marine species? [Scale: 1 = strongly disapprove, 7 = strongly approve]
DNIN1	• family	6.32	1.04
DNIN2	• close friends	6.24	1.06
DNIN3	• recreational peers	6.12	1.11
	Control beliefs: Belief strength (correlation with direct measure composite = .398, p < .001), [Scale: 1 = strongly disagree, 7 = strongly agree]
ICBS1	• An easy and user-friendly website/app will enable me to log an uncommon marine species	6.63	0.77
ICBS2	• Having better knowledge of marine species would help me to log an uncommon species	6.42	1.03
ICBS3	• Having more free time will enable me to log uncommon marine species	5.62	1.62
	Control beliefs: Power of control [Scale: 1 = very unlikely, 7 = very likely]
ICPC1	• I expect the [website/mobile app] will be easy and user-friendly	6.19	0.97
ICPC2	• I will have enough knowledge of marine species to log an uncommon species	5.29	1.51
ICPC3	• I will have enough time to be able to log uncommon marine species	5.72	1.29
ICPC3	Perceived behavioral control: Autonomy^c (Cronbach’s α = .769)
DCAU1	• The number of events outside of my control which would prevent me from logging an uncommon marine species on the [website/mobile app] are: [Scale: 1= numerous, 7 = very few]	5.53	1.53
DCAU2	• How much control do you have over whether you would log an uncommon marine species on the [website/mobile app] or not: [Scale: 1 = no control, 7 = complete control]	6.45	0.93
	Perceived behavioral control: Capacity
DCCA1	• For me to log an uncommon marine species on the [website/mobile app] would be: [Scale: 1 = very difficult, 7 = very easy]	6.22	1.14
DCCA2	• If I see an uncommon marine species, I am certain that I can log it on the [website/mobile app]: Scale: [1 = strongly disagree, 7 = strongly agree]	6.10	1.20
DPCA3	• How capable do you think you are to log an uncommon marine species on the [website/mobile app]? Scale: [1 = very incapable, 7 = very capable]	6.51	0.85
	Intention (Cronbach’s α = .879)
INT1	• If you saw an uncommon marine species, how likely is it (realistically) you would log it on the [website/mob. app]? [Scale: 1 = very unlikely, 7 = very likely]	6.09	1.12
INT2	• I intend to log any uncommon marine species I may see in the next 12 months on the [website/mobile app] [Scale: 1 = definitely not, 7 = definitely will]	6.07	1.13
INT3	I am willing to log any uncommon marine species I may see on the [website/mobile app] [Scale: 1 = strongly disagree, 7 = strongly agree]	6.43	0.91
INT4	• I want to log any uncommon marine species I may see in the next 12 months on the [website/mobile app] [Scale: 1 = strongly disagree, 7 = strongly agree]	6.29	0.97

Indirect measures were removed from the analysis following our concerns with potential measurement issues outlined in the Method section. ^bQuestions are presented in this style for simplicity within the table. They were different in appearance on screen in the online survey. For example, where there are blanks in the question presented here, the survey was designed to ask each question in full (no blanks).^cOne item was removed following the analysis of the interview data, which revealed the third item reduced the scale reliability.^dExpansion of item coding. With the exception of the final four codes, the 1st letter refers to (I)ndirect or (D)irect measures, the 2nd letter refers to (A)ttiude, (N)orm or (C)ontrol-level measures, the 3rd and 4th letters refer to the subdimensions the item belongs to (BS = belief strength, OE = outcome evaluation, IN (in DAIN) = instrumental attitude, EX = experiential attitude, MC = motivation to comply, IN (in DNIN) = injunctive norm, PC = power of control, AU = autonomy, CA = capacity, INT = intention), and the number identifies the specific item.

The three antecedents to intention (attitude, perceived norms, and PBC) are determined by a person’s underlying beliefs (behavioral, normative, and control beliefs). The three types of beliefs are measured using two components: the belief itself and the strength of the belief. Behavioral beliefs refer to the consequences (positive and negative) of the behavior (called outcome evaluation) and the strength of the belief in the likelihood of the outcome. Normative beliefs refer to the extent that others would approve or disapprove of the behavior (belief strength), and how motivated they are to do what these people expect (motivation to comply). Control beliefs refer to the extent to which the person believes certain factors will facilitate or inhibit their performance of the behavior (belief strength) and the power he or she has to perform the behavior (power of control). Belief-based measures are developed using open-ended questions during formative research with a subsample of target population, usually through interviews or focus groups.

Researchers assess the effect of attitude, perceived norm, and PBC on intention to determine the most important influences on behavior. Once understood, the basic premise of TPB is to look at belief differences between compliers (those who are performing the behavior) and noncompliers (those who are not performing the behavior). This is central to behavior change interventions, particularly since past behavior has been found to influence future behavior (Conner & Armitage, 1998; Ouellette & Wood, 1998). From a communication perspective, important beliefs and belief differences can also be instructive for communication strategies.

The TPB was developed from Ajzen and Fishbein’s (1980) earlier Theory of Reasoned Action, which led to the original formulation of TPB by Ajzen (1991). In this first model of TPB, the predictors were unidimensional. Since then, the theory has been refined to include two subdimensions in each predictor, which has strengthened the predictive power of the model (Fishbein & Ajzen, 2010). Most assessments of TPB have analyzed the original version of the theory (Ajzen, 2011; Armitage & Conner, 2001; Carmi, Arnon, & Orion, 2015; Kaiser, Hübner, & Bogner, 2005). Nevertheless, these studies have shown TPB to predict behavior, and especially intention, relatively well when the theory is applied appropriately.

The key shortcoming of TPB is the level of specificity required to describe the behavior (i.e., stipulating the action, target, context, and time). This makes it inappropriate for understanding generalized behavior (e.g., “conservation behavior” or “volunteering behavior”). As with other “intention-behavior” models of human behavior, it is often difficult to measure both intention and actual behavior (if the latter is measured it is more common to see self-reports of behavior than measures taken by observations). In many cases, intention is used as a proxy for behavior, with acknowledgment that there often remains a gap between intention and performance of the behavior (Carmi et al., 2015; Fishbein & Ajzen, 2010).

Another theory that has seen extensive use to predict behavior in an environmental context is the values-beliefs-norms theory (Stern, 2000; Stern, Dietz, Abel, Guagnano, & Kalof, 1999). While this theory has had success in predicting behavior, several studies have found TPB to be the stronger model of the two (Aguilar-Luzón, García-Martínez, Calvo-Salguero, & Salinas, 2012; Kaiser et al., 2005; López-Mosquera & Sánchez, 2012). Despite its stronger predictive capability, TPB has rarely been used in its full application in the marine environment. More often, researchers have used it to guide the direction of their research, and as such it has been used to investigate diver behavior around sharks (Apps, Lloyd, & Dimmock, 2014), turtle conservation (Kamrowski, Sutton, Tobin, & Hamann, 2014), and public awareness of marine wildlife entanglement in marine debris (Pearson, Mellish, Sanders, & Litchfield, 2014). Several other marine-based studies simply refer to TPB among other theories used in their research (Maguire, Rimmer, & Weston, 2015; Nursey-Bray et al., 2012; Wyles et al., 2013). The goal of our study is to determine factors that influence people’s intention to contribute to opportunistic marine citizen science. Using TPB as a framework, key objectives are the identification of (1) important beliefs and (2) belief differences between those who have contributed to a marine citizen science program and those who have not. We aim to use this information to provide an evidence-based approach to the development of effective public engagement strategies for marine citizen science.

Method

Our research comprised two phases for the application of TPB (1) a qualitative phase to uncover salient beliefs using face-to-face interviews and (2) the quantitative phase during which a national survey was conducted and analyzed to determine important influences on people’s intention to contribute to marine citizen science. The national survey asked respondents a broad range of questions, of which the TPB questions formed one section (albeit a substantial section). This article reports on the TPB results from second phase, while the first phase is reported in full elsewhere (Martin, Christidis, Lloyd, & Pecl, 2016). A brief summary of the interviews conducted in the first phase is provided below.

National Survey Development and Recruitment

Our TPB questions were developed according to Fishbein and Ajzen (2010). In Phase 1, belief-based measures arose from face-to-face interviews with 110 marine users in four regions of Australia. Responses to the open-ended questions on behavioral, normative, and control beliefs were coded into the key themes, checked for intercoder reliability between three coders using Krippendorff’s alpha with a cutoff of .80 (Krippendorff, 2004), then used as the belief items in the national survey (Table 2). The most salient behavioral beliefs about participation in marine citizen science were that it would increase knowledge (either scientific or their own), provide information and raise awareness for the community, and help protect/manage the marine environment. The most salient control beliefs (i.e., things that would help them to log a sighting) were an easy and user-friendly website/mobile app, having better knowledge of marine species, and having more free time.

During the interviews, we also tested the direct measures for use in Phase 2. It became evident that descriptive norms were problematic. To apply TPB correctly in our study, we attempted to ask respondents about the behavior of others who have observed an uncommon species. Very few interviewees knew someone who had seen an uncommon species, and those who had assumed that the others did not log a photograph. This problem raised the issue of relevance of the descriptive norm questions since this is not a commonly observed behavior of others and therefore has little influence on behavior of the individual. For this reason, descriptive norms were removed from the national survey questions.

We assessed the validity of the remaining direct measure scales (Table 2) by means of a confirmatory factor analysis (CFA) in SPSS Amos 22.0. The results indicated that one PBC (autonomy) item (It is mostly up to me whether I log an uncommon marine species on the website/mobile app) had a low factor loading (.34), and so this item was removed, leaving five PBC measures.

The national survey in Phase 2 included questions on demographics and the quantitative TPB measures developed through the earlier phase. Demographic questions were based on the most recent population census (2011) by the Australian Bureau of Statistics. They included age, gender, Australian state residence, and education variables. All TPB measures used a 7-point scale, which were presented as either unipolar scales (1 to 7) or bipolar scales (−3 to +3), where appropriate (a bipolar scale was used for items with negatively worded responses, then converted to unipolar for the analysis).

The questions were entered into the Qualtrics online survey platform and pretested by 12 people in different locations across the country on a variety of devices. This resulted in some minor changes to wording for clarity. Since we were presenting a hypothetical project based on Redmap, we wanted to ensure respondents understood the terms being used to describe the behavior, that is, logging a sighting of an uncommon marine species sometime in the next 12 months. The objectives of the project (without naming Redmap) and the terms used to describe what is sought from volunteers were described in the text along with a diagram of the process, as well as an explanatory video. To promote the survey nationally, we used snowball sampling strategies (Bryman, 2012) in a large-scale communication campaign. The promotion targeted mainstream media (newspapers, magazines, and radio) and social media (through interest groups on Facebook, online forums, and to a lesser degree, Twitter), and direct e-mails were sent to over 1,350 relevant groups around the country (including dive clubs fishing clubs, magazines, marine-related organizations, museums, boating/sailing groups, underwater photography contacts, surf life savers, sea scouts, industry, fishing/outdoor shops dive shops, and interviewees from the previous work). In addition, a convenience sample of staff and students at the lead researcher’s university were e-mailed a request to promote the survey among their marine-related networks. The method for recruiting respondents is described in greater detail in Martin, Christidis, and Pecl (2016).

The survey was open for 8 weeks from February to April 2015. During that time, 1,375 people commenced the questions, of which 1,145 were determined to be fully complete and valid after data cleaning and hence could be included in the final analysis. This represents a completion rate of 83.3%. Respondents took approximately 30 minutes to complete the survey. The survey was designed to ask the TPB questions only if the respondents had access to technology that would allow them to take digital photos and load them on the Internet or through a smartphone app. This reduced the number of respondents for the TPB questions to 1,076. Any negatively worded questions were reworded and reverse coded to facilitate comparisons of the results.

Analysis

The analysis of the TPB questions from the survey proceeded in four steps (1) assessment of direct measure scale reliability, (2) computation of composite scores and correlations between direct and indirect measures, (3) structural equation modeling of direct measures and their influence on intention, and (4) computation of differences in belief measures between past contributors and non-contributors. SPSS 22 was used for Steps 1, 2, and 4 and SPSS AMOS 22 for Step 3. The first two steps are assessments of the data, while the third and fourth steps produce results.

The first step in the analysis examined the scale reliability of the direct measures using Cronbach’s alpha (Table 2). All scales exceeded the minimum reliability coefficient of .70 (Pallant, 2013). This means the scales were reliable measures of the direct measure construct.

In the second step, we computed composite scores for the direct and indirect measures. Items in each of the direct measures (attitude, injunctive norm, and PBC) were summed. This resulted in the following maximum composite scores: 42 for attitude (6 items × 7 points on scale), 21 for injunctive norms (3 items × 7 points on scale), and 35 for PBC (5 items × 7 points on scale).

The indirect measure composite scores were calculated for behavioral and control beliefs. Behavioral beliefs were determined by multiplying the outcome evaluation by the strength of each belief (maximum score = 49 from the two 7-point scales) and the resulting five belief scores were summed to create the composite score with a maximum of 245 (49 × 5). Similarly, control beliefs were calculated by multiplying the power of control by the strength of each belief (maximum score = 49 from the two 7-point scales) and the resulting three scores summed to generate the composite score with maximum = 147 (49 × 3).

Composite scores were checked for any differences in direct and indirect measure correlations arising from the use of bipolar scoring (−3 to +3) or unipolar scoring (1 to 7). Fishbein and Ajzen (2010) recommend selecting the scale that results in the strongest correlation. There was little difference in the correlations between attitude and behavioral belief composites (Spearman’s ρ = .568 for unipolar scale, and .557 for bipolar scale, p < .001 for both correlations), and between PBC and control belief composites (Spearman’s ρ = .363 for unipolar scale, and .347 for bipolar scale, p <. 001 for both correlations), so the unipolar scale was maintained throughout the analysis.

The resultant composite scores were used to check for correlations between (1) behavioral beliefs and attitude toward the behavior and (2) control beliefs and PBC to ensure the underlying beliefs are a function of the direct measures (Table 2). The correlation coefficients indicate (1) large and (2) moderate effects, respectively (Field, 2013). A composite score for intention was calculated by summing the scores from the four intention items.

The third step in the analysis was the structural equation model. To assess the model, we used a two-step procedure as outlined in Byrne (2010). Since our data are multivariate nonnormal (most scales were negatively skewed, which affects the reliability of the model, and in particular may result in an inflated χ² likelihood ratio test of model fit), we used bootstrapping procedures with 1,000 samples during the analysis. First, we examined the validity of scores using a second-order CFA, which was performed in SPSS AMOS 22. According to Hagger and Chatzisarantis (2005), incorporating the subdimensions of the higher order TPB factors into the CFA model enables distinctions to be made at the subdimension level. The model was assessed using the following model fit indices, all of which fell into acceptable parameters (Byrne, 2010; Hu & Bentler, 1999): χ² = 447.111, degrees of freedom = 125, p < .001, root mean square error of approximation (RMSEA) = .049, comparative fit index (CFI) = .975, standardized root mean square residual (SRMR) = .0347.

The factors in the CFA were examined for convergent and discriminant validity using the following measures, all of which fell into acceptable parameters (Hair, Black, Babin, Anderson, & Tatham, 2010): composite reliabilities were all >.7, average variances extracted (AVE) were all >.5, maximum shared variances and the average shared variances for all factors were less than the AVE for corresponding factors, and the square root of AVE was greater than interconstruct correlations.

Reliability of the model was examined for common method bias using the common latent factor (CLF) method (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003), which showed no large differences between the model with a CLF and the model without a CLF (the largest difference was .02, which is well below the standard cutoff of .2). Measurement model invariance was also tested for configural and metric invariance. The model passed the configural variance test using two groups (males and females) to test for differences in model fit. For the metric invariance test, which relies on the χ², the sample size had to be reduced to a random sample of 400 (200 in each of two groups) due to the sensitivity of χ² to large sample sizes. This test showed the two models being compared were invariant. In summary, the CFA confirmed that we could proceed with the structural model without modification. The structural model (presented below) was also assessed for validity and reliability using the same measures and was found to pass all of the tests.

The fourth and final step in the analysis was to investigate belief differences between compliers and noncompliers. For this, we used several questions to identify those who had participated in an opportunistic marine citizen science project in the past. First, we asked whether people were aware of any projects similar to the hypothetical project and then asked them to name the project if they could. Redmap is the only citizen science project in Australia asking people to log sightings of uncommon marine species, and it matches our definition of the behavior. Next, we asked, Have you ever contributed observations, data, or photos to any of the projects you mentioned? People who said they (1) had made a contribution and (2) mentioned Redmap specifically were recoded into a group called “contributors” (N = 88, or 8% of the sample), while those who had not made a contribution in the past were coded as “non-contributors” (N = 988).

We encountered further issues with the questions on normative beliefs, primarily on the motivation to comply items, which obtained low mean scores (averaging 2.25 on the 7-point scale). This means respondents felt other people cannot significantly influence their behavior. While we acknowledge that, in this context, it is possible respondents could have a low desire to “do what others think I should do,” we suspect this result may be due more to the social desirability to say, “No one can influence my behavior.” Yet, if an important other (say, a child) is particularly enthusiastic about sending a photograph of an uncommon species to marine scientists, it is likely the child will be able to persuade the parent to log the sighting. We also acknowledge the considerable problems identified in behavioral research when it comes to measuring norms accurately (Armitage & Conner, 2001; Manning, 2009; Osborne & Waters, 2002; Rivis & Sheeran, 2003). Recognizing that our results for perceived norm may be inaccurate, we have removed indirect normative beliefs from further analysis and highlight this topic as an area requiring more research.

Due to the difference in sample size and variance of composite behavioral and control belief scores for contributors and non-contributors, a Mann–Whitney U test was used to determine whether there were significant differences in the mean composite scores (Field, 2013; Pallant, 2013).

Results

Demographics of Respondents

Although our survey was not intended to be representative of the Australian population, it is nevertheless useful to compare our sample with census data to understand the types of people who responded. A detailed description of the respondents is provided in Martin, Christidis, and Pecl (2016); however, for the purposes of this article we summarize the background of respondents below.

Survey responses came from every state and mainland territory of Australia, differing from the census by no more than 4% except for an overrepresentation from the state of New South Wales (41.3% vs. 32.0% in the census) and underrepresentation from Victoria (8.3% vs. 24.9% in the census). The sample represented a broad range of age-groups between 15 and 84 years, two thirds of which fell between 25 and 54 years. Compared to the census data (49.8% males, 50.2% females), we had a slight overrepresentation of males (54.1%) and fewer females (45.9%). The most striking difference between our sample and the census population was the level of higher education. A little over one fifth of our sample (22.4%) had attained postgraduate degrees, compared to 5.2% of the general population.

Intention

The majority of respondents indicated a strong intention to log a sighting of an uncommon marine species if they see one. All mean scores for the four intention measurement items are greater than 6 on the 7-point scales used (Table 2). A Mann–Whitney U test revealed that past contributors to Redmap have stronger intentions to contribute sightings than those who had not contributed. The composite intention scores for past contributors (Mdn = 27.00, n = 88) is higher than for those who have not contributed (Mdn = 26.00, n = 988), U = 33,619, z = −3.593, p < .001; however, the magnitude of the differences in the means was relatively small at r = .11 (Pallant, 2013).

The mean item scores for the antecedents to intention (attitude, injunctive norms, and PBC) were all above 5.50 and most above 6.00 (Table 2). This means that respondents have a positive attitude toward submitting a sighting of an uncommon marine species and feel they have a strong degree of control over whether they do so or not. In addition, important people in their lives (family, friends, and other recreational peers) would support their decision to log a sighting.

The model of the antecedents and their influence on intention (Figure 2) was found to be acceptable (χ² = 447.111, degrees of freedom = 125, RMSEA = .049, CFI = .975, SRMR= .0347). The three predictor variables attitude, injunctive norms, and PBC explain 69% of the variance in behavioral intention. Perceived behavioral control (β = .495) plays the greatest role in determining intention, followed by attitude (β = .374) and, to a lesser extent, injunctive norms (β = .138; all β are the standardized regression weights after bootstrapping). All of the regression coefficients are significant at p < .01 using the bias-corrected percentile method. In addition, the three predictors moderately correlate with each other between .39 and .50.

Figure 2.

Structural equation model.

Beliefs

The respondents’ beliefs about logging an uncommon marine species are (in descending order of importance) that it would (1) increase their own knowledge, (2) increase scientific knowledge, (3) help protect/manage the marine environment, (4) provide information for the greater good/everyone, and (5) increase public awareness of the marine environment (Table 3). The most prominent control belief was that website/mobile app would be easy and user-friendly, which would help them log a sighting. This belief received a much higher score than the other two (better knowledge of marine species and having more free time).

Table 3.

Composite Belief Scores.

	Overall (N = 1,076)			Contributor (N = 88)		Non-contributor (N = 988)
	Maximum	M	SD	M	SD	M	SD
Behavioral beliefs (BB)
BB1: Increase scientific knowledge	49	39.78	8.416	40.88	8.42	39.68	8.41
BB2: Increase my own knowledge	49	40.20	9.086	40.41	9.00	40.18	9.10
BB3: Provide information for the greater good/everyone	49	38.27	9.283	38.28	8.68	38.27	9.34
BB4: Increase public awareness of the marine environment	49	37.68	9.273	37.28	9.17	37.71	9.29
BB5: Help protect/manage the marine environment	49	38.77	9.636	37.36	9.22	38.90	9.67
Control beliefs (CB)
CB1: Easy and user-friendly website/app	49	41.23	8.644	42.52	7.55	41.11	8.73
CB2: Better knowledge of marine species^a	49	33.79	11.056	37.66	10.00	33.45	11.08
CB3: More free time	49	31.51	10.996	33.33	9.92	31.35	11.08

Differences between contributors and non-contributors belief scores are statistically significant, p < .001.

The Mann–Whitney U test revealed that the only significant difference in beliefs between past Redmap contributors and non-contributors is one control belief CB2: better knowledge of marine species. Past contributors to Redmap believe more strongly that they have enough knowledge of marine species to be able to submit a sighting compared to the beliefs of non-contributors (contributors: Mdn = 42.00, n = 88; non-contributors: Mdn = 35.00, n = 988; U = 33,065.50, z = −3.76, p < .001), although the effect size is small (r = .12).

Discussion

The results highlight important considerations for the design and communication of marine citizen science projects, particularly those that are opportunistic in nature. Below we discuss influences on people’s intention to participate, and examine underlying beliefs about participation. Finally, we consider how marine citizen science projects can use this information for enhancing volunteer engagement.

Influences on Intention

The structural equation model showed the strongest influence on intention is their perception of control over submitting a sighting. The stronger their feelings of control, the higher the likelihood they will log a sighting. However, the majority of marine users in our survey feel they have the capacity and autonomy to submit sightings if they choose to do so.

Perceived behavioral control has also been found to be a strong predictor of intention and/or behavior in studies of pro-environmental behaviors (de Leeuw, Valois, Ajzen, & Schmidt, 2015; Howell, Shaw, & Alvarez, 2015; Le, Yamasue, Okumura, & Ishihara, 2013). In many ways, contributing to a citizen science project could be considered a pro-environmental behavior, which, in essence, is behavior that minimizes environmental impacts or benefits environmental health (Steg & Vlek, 2009). Most marine citizen science projects are concerned about negative impacts and aim to improve the health of the marine environment either through provision of data for science and management or directly through volunteer restoration work (Thiel et al., 2014). Our results emphasize the important role that contextualized “control” factors play in encouraging people to act pro-environmentally.

The second factor influencing participation is attitude toward the behavior, that is, the more positive a person’s attitude about logging a sighting with the marine citizen science project, the higher his or her intention to do so. Overall, marine users in this survey have a favorable attitude toward our hypothetical project, considering it to be a worthwhile and positive experience. These results likely reflect the generally positive attitude toward science in the Australian community (Department of Business and Innovation, 2012; Searle, 2014). When combined with the high social value of marine environments in Australia (Department of Industry, Innovation, Science, Research and Tertiary Education, 2012; Tobin et al., 2014; Voyer, Gollan, Barclay, & Gladstone, 2015), it appears likely there are many potential volunteers for marine citizen science. Our results indicate that supportive attitudes are already in place for future participation in marine citizen science. They may also help explain why Brossard, Lewenstein, and Bonney (2005) found no change in citizen scientists’ attitudes toward science and the environment, since the types of people most likely to contribute to citizen science already hold these attitudes in a positive light.

The third factor (injunctive norms) plays a minor role in determining a person’s intention to contribute, despite people feeling they would be well supported by others (family, friends, etc.). As mentioned earlier, measurement issues arose for the normative questions, so it remains to be seen whether this result is valid in this particular context. We suspect that, in a “real-life” scenario, social norms will play a more significant role in encouraging people to submit sightings than our results suggest, particularly as there is recognition in the behavioral literature of the important role social influence has on environmental behavior (Abrahamse & Steg, 2013). This issue may become more relevant in the future as the number of marine citizen science projects increase and the behavior becomes more frequent among marine users. Some citizen science projects (e.g., Redmap and QuestaGame) already encourage social norms through promotion of submitted sightings in social media and their websites.

Beliefs About Participation

This study found that respondents’ belief about their knowledge of marine species is the most influential barrier to participation in the hypothetical project. Changing this belief among potential (rather than current) volunteers could increase public contributions to marine citizen science. People who are already contributing to marine citizen science feel more confident in their knowledge, and have higher intentions to submit sightings, than those who have never contributed. This may be a consequence of past experience in citizen science, which has been shown to have a positive effect on volunteer knowledge in other contexts (Bonney, Phillips, Ballard, & Enck, 2015; Brossard et al., 2005; Crall et al., 2013).

There were no other significant belief differences between contributors and non-contributors. Nevertheless, it is useful to look at other underlying beliefs of all respondents to understand important considerations for volunteer engagement. These issues emphasize best practice for project design, communication, and recruitment of volunteers. For instance, while time considerations are less of a barrier than species knowledge, this may be an issue for newcomers to opportunistic citizen science. Unless it is well communicated, many may not be aware of the actual time commitment (which is often minimal). Logging an observation with Redmap, for example, takes approximately 2 minutes.

The behavioral belief results highlight potential volunteers’ expectations that participation will bring about increases in (1) the individuals’ knowledge of marine species and (2) scientific knowledge. In other words, marine citizen science volunteers expect they will learn something and will be able to make a tangible contribution to scientific understanding of the marine environment. Studies on the motivations of citizen scientists have also found that contributing to science is an important reason behind volunteer effort (Curtis, 2015; Haywood, 2015; Land-Zandstra et al., 2016; Raddick et al., 2013). While this is also an important expectation for our respondents, our study shows they believe a more likely outcome is that they will learn more about marine species. With this in mind, it is important to remember that volunteer motivations are likely to change over time as they continue to make contributions to the project (Rotman et al., 2014).

Many respondents also believe their contributions may help protect or manage the marine environment or increase public awareness about the marine environment. These may also be important drivers for participation, particularly for divers, who tend to have a strong desire to assist conservation outcomes (Hammerton, Dimmock, Hahn, Dalton, & Smith, 2012). The results from the additional beliefs (beyond simply the motivation to contribute to science) increase our understanding of reasons why people are likely to assist marine research in the future.

The study shows the most important control belief is the user-friendly design of websites or apps. This was also a key issue for marine users in our interviews (Martin, Christidis, Lloyd, et al., 2016). Most Australians are very familiar and comfortable with using digital and mobile technology and have high levels of access to the Internet (Internet World Stats, 2015). This means they are likely to have experienced good and poor design of websites and apps and understand the value of a user-friendly design that is quick, simple, and free from errors. Since potential volunteers feel that this is an important issue, and expect good design, citizen science projects should not underestimate the value of developing these interfaces with end users in mind.

Implications for Citizen Science Projects

This study found that many people are in favor of making contributions to marine citizen science, despite the fact that 92% of the respondents had never participated in a project similar to the one we described. The barriers to participation appear to be relatively minimal, although this will depend on good project design and thoughtful communication aimed at targeting the beliefs held by marine users in relation to their potential contributions.

The interventions most likely to lead to new volunteers submitting sightings are those that increase people’s perception that they know enough about marine species to be able to do so. We recommend project managers implement mechanisms within their project to build volunteer knowledge (e.g., ID charts, running workshops, providing training materials, etc.) and clearly communicate the level of knowledge required to participate (which may be less than volunteers presume since many projects do not require expert knowledge). The findings in this study suggest that doing so will help marine users overcome the hurdle of uncertainty in their ability to participate, which would increase the number of sightings reported.

The responses to other belief questions also provide valuable information on important elements of project design from volunteers’ perspectives, which form the basis for our additional recommendations below. Although many of these issues should be part of any effective engagement strategy (e.g., user-friendly design of websites and apps), our observation is that some projects deal with these issues better than others, and some do not consider them at all.

Our next recommendation is to ensure potential volunteers understand the time commitment required, especially since a lot of opportunistic citizen science projects take very little time to submit a sighting (often a matter of minutes). Many projects we have looked at make no mention of the time it takes to add a record. Given volunteers expect they will learn something, we also suggest that project managers communicate stories about actual participants to demonstrate that this is a realistic outcome. Additionally, volunteers expect their contributions to have real impact on scientific knowledge or conservation. Demonstrating these outcomes in a public space (e.g., on a website, through newsletters, etc.) will confirm and strengthen the beliefs of potential volunteers. Engaging volunteers also requires effective website and mobile app design, not only to recruit volunteers but also to retain them. Good design should help address the issue of volunteer retention in large-scale collaborative projects (Crowston, Jullien, & Ortega, 2013).

Finally, we encourage citizen science practitioners to conduct further research on their target audience since certain beliefs are likely to differ in other contexts. We have demonstrated the usefulness of a theory-based approach to understanding drivers and barriers of public contributions to opportunistic marine citizen science. However, there are some important limitations of this study to consider. First, it is (by necessity) focused on a particular project (the Redmap model). While this project shares similarities with other observation-based digital marine citizen science projects such as Eye on the Reef (www.gbrmpa.gov.au) for which some of the findings will be informative, similar research will need to be conducted in other types of citizen science projects to understand whether public perceptions differ between particular projects. For example, differences may be more acute for projects that require a greater level of volunteer involvement such as attending training, conducting data collection activities at a particular time, and so on.

Second, the problems encountered with the normative questions leaves a gap in knowledge about the influence other people can have over volunteering behavior in citizen science. There may be a very strong influence of “others” in this setting since one of the strongest predictors of volunteering to formal organizations is simply being asked (Sundeen, Raskoff, & Garcia, 2007). Of particular importance will be the question of who is doing the asking, since certain people in an individual’s life will have more influence in some situations than others (Fishbein & Ajzen, 2010). Who is behind the project also matters because potential volunteers are more willing to share data with some organizations than with others (Martin, Christidis, & Pecl, 2016). We suggest that a qualitative approach to this issue in future research will help uncover social norms in greater depth.

Citizen science has seen remarkable growth in recent years (Kullenberg & Kasperowski, 2016) and is being actively encouraged by governments across the globe (Department of Industry, Innovation and Science, 2016; Holdren, 2015; Museum für Naturkunde, 2016). The value in growing the number of productive collaborations between marine scientists and the public is the potential to increase the spatial and temporal scale of data collection (Theobald et al., 2015) and to facilitate more timely data collection to inform policy and decision making (Danielsen, Burgess, Jensen, & Pirhofer-Walzl, 2010). This information is urgently needed to increase understanding of the considerable changes occurring in our oceans now and into the future.

Footnotes

Acknowledgements

The authors would like to thank Martin Thiel and an anonymous reviewer for the valuable comments on the original manuscript. Special thanks go to the countless individuals who assisted the recruitment of respondents and to those who completed the survey. We would also like to extend our gratitude to James Gaskin, Brigham Young University, for the provision of additional statistical advice.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Victoria Martin was supported by an Australian Postgraduate Award. Gretta Pecl was supported by an ARC Future Fellowship. This research was conducted with ethics approval from Southern Cross University (Approval Number ECN-14-041).

Author Biographies

Victoria Martin is a PhD researcher in the School of Environment, Science & Engineering, at Southern Cross University, Australia. Her research is investigating the connections between science and society, and the role that public participation in marine science might play in strengthening this relationship.

Liam Smith is the director of BehaviourWorks Australia, an applied behavior change research unit based in Monash University’s Sustainable Development Institute, which he founded in 2011. BehaviourWorks has grown significantly and is helping governments around Australia design and test interventions to encourage behavior for individual and social good. Focal areas include preventative and clinical health, biosecurity, education, workplace safety, state revenue, and environment.

Alison Bowling is a lecturer in psychology at Southern Cross University, teaching statistical methods to third- and fourth-year psychology students. Her research primarily involves cognitive change with ageing, using eye-tracking methodology. She has a keen interest in advanced statistics and provides statistical support to students at Southern Cross University.

Les Christidis is director of the National Marine Science Centre, Southern Cross University, Australia. Previously he was in the museum sector where he led the development of several major science-based exhibitions, including planetarium productions.

David Lloyd is an associate professor of protected area management in the School of Environment, Science & Engineering at Southern Cross University, Australia, and has a long history of working in national and marine park management.

Gretta Pecl is the deputy associate dean of Research at IMAS and an ARC Future Fellow. Her research focuses on species responses and adaptation of fisheries systems to climate change and on citizen science approaches to ecological monitoring.

References

Abrahamse

Steg

(2013). Social influence approaches to encourage resource conservation: A meta-analysis. Global Environmental Change, 23, 1773-1785.

Aguilar-Luzón

M. C.

García-Martínez

J. M. Á.

Calvo-Salguero

Salinas

J. M.

(2012). Comparative study between the theory of planned behavior and the value–belief–norm model regarding the environment, on Spanish housewives’ recycling behavior. Journal of Applied Social Psychology, 42, 2797-2833.

Ajzen

(1991). The theory of planned behavior. Organizational Behavior and Human Decision Processes, 50, 179-211.

Ajzen

(2011). The theory of planned behaviour: Reactions and reflections. Psychology & Health, 26, 1113-1127.

Ajzen

Fishbein

(1980). Understanding attitudes and predicting social behavior. Englewood Cliffs, NJ: Prentice Hall.

Apps

Lloyd

Dimmock

(2014). Scuba diving with the grey nurse shark (Carcharias taurus): An application of the theory of planned behaviour to identify divers beliefs. Aquatic Conservation, 25, 201-211.

Armitage

C. J.

Conner

(2001). Efficacy of the theory of planned behaviour: A meta-analytic review. British Journal of Social Psychology, 40, 471-499.

Bonney

Phillips

T. B.

Ballard

H. L.

Enck

J. W.

(2015). Can citizen science enhance public understanding of science? Public Understanding of Science, 25, 2-16.

Brossard

Lewenstein

Bonney

(2005). Scientific knowledge and attitude change: The impact of a citizen science project. International Journal of Science Education, 27, 1099-1121.

10.

Bryman

(2012). Social research methods. Oxford, England: Oxford University Press.

11.

Byrne

B. M.

(2010). Structural equation modeling with AMOS: Basic concepts, applications, and programming. New York, NY: Routledge.

12.

Carmi

Arnon

Orion

(2015). Seeing the forest as well as the trees: General vs. specific predictors of environmental behavior. Environmental Education Research, 21, 1011-1028.

13.

Conner

Armitage

C. J.

(1998). Extending the theory of planned behavior: A review and avenues for further research. Journal of Applied Social Psychology, 28, 1429-1464.

14.

Crall

A. W.

Jordan

Holfelder

Newman

G. J.

Graham

Waller

D. M.

(2013). The impacts of an invasive species citizen science training program on participant attitudes, behavior, and science literacy. Public Understanding of Science, 22, 745-764.

15.

Crowston

Jullien

Ortega

(2013). Sustainability of open collaborative communities: Analyzing recruitment efficiency. Technology Innovation Management Review, 3, 20-26.

16.

Crowston

Prestopnik

N. R.

(2013). Motivation and data quality in a citizen science game: A design science evaluation. In Proceedings of the 2013 46th Hawaii International Conference on System Sciences (pp. 450-459). Washington, DC: IEEE Computer Society. Retrieved from http://dl.acm.org/citation.cfm?id=2470776&picked=prox&cfid=627826005&cftoken=77930876

17.

Curtis

(2015). Motivation to participate in an online citizen science game: A study of foldit. Science Communication, 37, 723-746.

18.

Danielsen

Burgess

N. D.

Jensen

P. M.

Pirhofer-Walzl

(2010). Environmental monitoring: The scale and speed of implementation varies according to the degree of peoples involvement. Journal of Applied Ecology, 47, 1166-1168.

19.

Darnton

(2008). GSR behaviour change knowledge review: Reference report: An overview of behaviour change models and their uses. London, England: HMT.

20.

de Leeuw

Valois

Ajzen

Schmidt

. (2015). Using the theory of planned behavior to identify key beliefs underlying pro-environmental behavior in high-school students: Implications for educational interventions. Journal of Environmental Psychology, 42, 128-138.

21.

Department of Business and Innovation. (2012). Community interest and engagement with science and technology in Victoria 2011. Melbourne, Victoria, Australia: Department of Business and Innovation, Government of Victoria.

22.

Department of Industry, Innovation and Science. (2010). Inspiring Australia: A national strategy for engagement with the sciences. Canberra, Australia.

23.

Department of Industry, Innovation and Science. (2016). Crowdsourcing research assistants. Retrieved from http://minister.industry.gov.au/ministers/pyne/media-releases/crowdsourcing-research-assistants

24.

Department of Industry, Innovation, Science, Research and Tertiary Education. (2012). Marine science: A story for Australia. Canberra, Australia.

25.

Field

A. P.

(2013). Discovering statistics using IBM SPSS statistics and sex and drugs and rock ‘n’ roll. London, England: Sage.

26.

Fishbein

Ajzen

(2010). Predicting and changing behavior: The reasoned action approach. New York, NY: Taylor & Francis.

27.

Hagger

M. S.

Chatzisarantis

N. L. D.

(2005). First- and higher-order models of attitudes, normative influence, and perceived behavioural control in the theory of planned behaviour. British Journal of Social Psychology, 44, 513-535.

28.

Hair

J. F.

Black

Babin

Anderson

R. E.

Tatham

R. L.

(2010). Multivariate data analysis. Upper Saddle River, NJ: Pearson/Prentice Hall.

29.

Hammerton

Dimmock

Hahn

Dalton

S. J.

Smith

S. D. A.

(2012). Scuba diving and marine conservation: Collaboration at two Australian subtropical destinations. Tourism in Marine Environments, 8, 77-90.

30.

Haywood

B. K.

(2015). Beyond data points and research contributions: The personal meaning and value associated with public participation in scientific research. International Journal of Science Education, Part B, 1-24.

31.

Holdren

J. P.

(2015). Addressing societal and scientific challenges through citizen science and crowdsourcing: Memorandum to the heads of executive departments and agencies. Washington, DC: Office of Science and Technology Policy.

32.

Howell

A. P.

Shaw

B. R.

Alvarez

(2015). Bait shop owners as opinion leaders: A test of the theory of planned behavior to predict pro-environmental outreach behaviors and intentions. Environment & Behavior, 47, 1107-1126.

33.

L.-T.

Bentler

P. M.

(1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1-55.

34.

Internet World Stats. (2015). Internet usage and 2015 population in Oceania. Retrived from http://www.internetworldstats.com/stats6.htm#oceania

35.

Johnson

M. F.

Hannah

Acton

Popovici

Karanth

K. K.

Weinthal

(2014). Network environmentalism: Citizen scientists as agents for environmental advocacy. Global Environmental Change, 29, 235-245.

36.

Kaiser

F. G.

Hübner

Bogner

F. X.

(2005). Contrasting the theory of planned behavior with the value-belief-norm model in explaining conservation behavior. Journal of Applied Social Psychology, 35, 2150-2170.

37.

Kamrowski

R. L.

Sutton

S. G.

Tobin

R. C.

Hamann

(2014). Potential applicability of persuasive communication to light-glow reduction efforts: A case study of marine turtle conservation. Environmental Management, 54, 583-595.

38.

Krippendorff

(2004). Content analysis: An introduction to its methodology. Thousand Oaks, CA: Sage.

39.

Kullenberg

Kasperowski

(2016). What is citizen science? A scientometric meta-analysis. PLoS One, 11, e0147152.

40.

Land-Zandstra

A. M.

Devilee

J. L. A.

Snik

Buurmeijer

van den Broek

J. M.

(2016). Citizen science on a smartphone: Participants’ motivations and learning. Public Understanding of Science, 25, 45-60.

41.

H.-L.

Yamasue

Okumura

Ishihara

K. N.

(2013). Analysis of intentions to recycle electronic waste (e-waste) using the theory of planned behavior: A case study in urban areas of Vietnam. In Yao

(Ed.), Zero-Carbon Energy Kyoto 2012 (pp. 73-79). Tokyo, Japan: Springer.

42.

López-Mosquera

Sánchez

(2012). Theory of planned behavior and the value-belief-norm theory explaining willingness to pay for a suburban park. Journal of Environmental Management, 113, 251-262.

43.

Maguire

Rimmer

Weston

(2015). Stakeholder knowledge of threatened coastal species: The case of beach-goers and the Hooded Plover Thinornis rubricollis. Journal of Coastal Conservation, 19, 73-77.

44.

Manning

(2009). The effects of subjective norms on behaviour in the theory of planned behaviour: A meta-analysis. British Journal of Social Psychology, 48, 649-705.

45.

Martin

V. Y.

Christidis

Lloyd

Pecl

(2016). Understanding drivers, barriers and information sources for public participation in marine citizen science. Journal of Science Communication, 15. Retrieved from http://jcom.sissa.it/sites/default/files/documents/JCOM_1502_2016_A02.pdf

46.

Martin

V. Y.

Christidis

Pecl

G. T.

(2016). Public interest in marine citizen science: Is there potential for growth? BioScience. Advance online publication. doi:10.1093/biosci/biw070

47.

Measham

T. G.

Barnett

G. B.

(2008). Environmental volunteering: Motivations, modes and outcomes. Australian Geographer, 39, 537-552.

48.

Museum für Naturkunde. (2016). Research for all! [Press release]. Retrieved from http://naturkundemuseum.berlin/en/press/press-releases/research-all

49.

Nursey-Bray

Pecl

G. T.

Frusher

Gardner

Haward

Hobday

. . . van Putten

(2012). Communicating climate change: Climate change risk perceptions and rock lobster fishers, Tasmania. Marine Policy, 36, 753-759.

50.

Osborne

J. W.

Waters

(2002). Four assumptions of multiple regression that researchers should always test. Practical Assessment, Research, and Evaluation, 8(2). Retrieved from http://www-psychology.concordia.ca/fac/kline/601/osborne.pdf

51.

Ouellette

J. A.

Wood

(1998). Habit and intention in everyday life: The multiple processes by which past behavior predicts future behavior. Psychological Bulletin, 124, 54-74.

52.

Pallant

(2013). SPSS survival manual: A step by step guide to data analysis using IBM SPSS. Crows Nest, New South Wales, Australia: Allen & Unwin.

53.

Pearson

Mellish

Sanders

Litchfield

(2014). Marine wildlife entanglement: Assessing knowledge, attitudes, and relevant behaviour in the Australian community. Marine Pollution Bulletin, 89, 136-148.

54.

Podsakoff

P. M.

MacKenzie

S. B.

Lee

J.-Y.

Podsakoff

N. P.

(2003). Common method biases in behavioral research: A critical review of the literature and recommended remedies. Journal of Applied Psychology, 88, 879-903.

55.

Poliakoff

Webb

T. L.

(2007). What factors predict scientists’ intentions to participate in public engagement of science activities? Science Communication, 29, 242-263.

56.

Raddick

M. J.

Bracey

Gay

P. L.

Lintott

C. J.

Cardamone

Murray

. . . Vandenberg

(2013). Galaxy zoo: Motivations of citizen scientists. Astronomy Education Review, 12(1). Retrieved from http://portico.org/Portico/#!journalAUSimpleView/tab=PDF?cs=ISSN_15391515?ct=E-Journal%20Content?auId=ark:/27927/pgg3ztfcv7h

57.

Rivis

Sheeran

(2003). Descriptive norms as an additional predictor in the theory of planned behaviour: A meta-analysis. Current Psychology, 22, 218-233.

58.

Rotman

Hammock

Preece

Hansen

Boston

Bowser

(2014, February). Motivations affecting initial and long-term participation in citizen science projects in three countries. Paper presented at the iConference, Berlin, Germany.

59.

Searle

(2014). How do Australians engage with science? Preliminary results from a national survey. Australian National Centre for the Public Awareness of Science (CPAS). Canberra, Australia: Australian National University.

60.

Shirk

J. L.

Ballard

H. L.

Wilderman

C. C.

Phillips

Wiggins

Jordan

. . . Bonney

(2012). Public participation in scientific research: A framework for deliberate design. Ecology & Society, 17(2), Article 29. Retrieved from http://www.ecologyandsociety.org/vol17/iss2/art29/

61.

Silvertown

(2009). A new dawn for citizen science. Trends in Ecology & Evolution, 24, 467-471.

62.

Steg

Vlek

(2009). Encouraging pro-environmental behaviour: An integrative review and research agenda. Journal of Environmental Psychology, 29, 309-317.

63.

Stern

P. C.

(2000). Toward a coherent theory of environmentally significant behavior. Journal of Social Issues, 56, 407-424.

64.

Stern

P. C.

Dietz

Abel

T. D.

Guagnano

G. A.

Kalof

(1999). A value-belief-norm theory of support for social movements: The case of environmentalism. Human Ecology Review, 6, 81-97.

65.

Sundeen

R. A.

Raskoff

S. A.

Garcia

M. C.

(2007). Differences in perceived barriers to volunteering to formal organizations: Lack of time versus lack of interest. Nonprofit Management & Leadership, 17, 279-300.

66.

Theobald

E. J.

Ettinger

A. K.

Burgess

H. K.

DeBey

L. B.

Schmidt

N. R.

Froehlich

H. E.

. . . Parrish

J. K.

(2015). Global change and local solutions: Tapping the unrealized potential of citizen science for biodiversity research. Biological Conservation, 181, 236-244.

67.

Thiel

Penna Díaz

Luna Jorquera

Salas

Sellanes

Stotz

(2014). Citizen scientists and marine research: Volunteer participants, their contributions, and projection for the future. Oceanography and Marine Biology, 52, 257-314.

68.

Tobin

R. C.

Bohensky

Curnock

Goldberg

Gooch

Marshall

. . . Stone-Jovicich

(2014). The Social and Economic Long Term Monitoring Program (SELTMP) 2014, recreation in the Great Barrier Reef (Report to the National Environmental Research Program). Cairns, Australia: Reef and Rainforest Research Centre. Retrieved from http://elibrary.gbrmpa.gov.au/jspui/bitstream/11017/2961/1/NERP-TE-PROJ-10.1-SELTMP-2014-RECREATION-Complete.pdf

69.

Voyer

Gollan

Barclay

Gladstone

(2015). “It’s part of me”: Understanding the values, images and principles of coastal users and their influence on the social acceptability of MPAs. Marine Policy, 52, 93-102.

70.

Wyles

K. J.

Pahl

White

Morris

Cracknell

Thompson

(2013). Towards a marine mindset: Visiting an aquarium can improve attitudes and intentions regarding marine sustainability. Visitor Studies, 16, 95-110.