Meeting volunteer expectations — a review of volunteer motivations in citizen science and best practices for their retention through implementation of functional features in CS tools

Abstract

Citizen Science (CS) projects vary greatly. The aims and goals of a CS project determine the type of citizen involvement and the tools to be used, which in most cases also entail information and communication technology (ICT) that facilitates public participation in scientific research. Resource limitations in CS projects often require adopting suboptimal tools, which, however, may come with hidden costs stemming from poor usability and underwhelming functionality, thus reducing volunteers’ motivation. Meeting the volunteers’ expectations by designing or using existing tools with functional features which fulfill and nurture their motivations, will foster long-term participation and contribute to project sustainability. This paper reviews the types of CS projects, volunteer motivation and retention strategies from the literature and classifies them thematically. This is distilled into guidance that can help CS practitioners to design and implement CS tools and plan and manage CS projects, which better serve their scientific and volunteer-related goals.

Keywords:

• citizen science
• volunteer motivation and retention
• expectation management; sustainability of CS projects
• tools for citizen science

1. Introduction

Citizen Science (CS) refers to the voluntary public engagement in scientific research activities, such as, collecting and processing environmental data and helping to answer real-world scientific questions with their intellectual effort, knowledge of their surroundings or tools and other resources (Bonney, Cooper, et al. 2009; Sanz et al. 2014; Silvertown 2009). Methods to practice CS have undergone a revolution over the last two decades, mainly due to the advances in Information and Communication Technologies (ICT), sensor technologies and social networking, which has resulted in new tools which foster open, efficient and agile systems (Sanz et al. 2014 ; Silvertown 2009; Wald, Longo, and Dobell 2016). Many CS projects rely on surveys completed by volunteers who report objective or subjective observations about their environment; a task made ever-easier using modern technologies such as smartphone apps that facilitate the interaction between scientists and volunteers. Many projects require volunteers to submit data on, e.g. biodiversity and environmental phenomena or conditions using image acquisition, web- and smartphone-based applications and surveys (Couvet and Prevot 2015; Havlik and Schimak 2014; Wallace, Snedigar, and Cameron 2015). More and more CS projects also explore the use of low-cost devices, including handheld or portable sensor systems (Aspuru et al. 2016; Thompson 2016; Uhrner et al. 2014). Data collected are increasingly visualized in web-based portals or via smart phone apps in near real time. Moreover, there is a gradual shift toward participatory, bottom-up and co-design processes, where citizens are not just data collectors and where communication is only one way, as in traditional contributory crowdsourcing projects, but are involved in advanced participatory communication models addressing community needs (Mominó, Piera, and Jurado 2016; Sanz et al. 2014; Shirk et al. 2012). In these more reciprocal and inclusive projects, such as Citizen Observatory (CO) projects, the aim is to directly benefit the citizens and society generally, rather than science alone (Grainger 2017). These projects strengthen social capital, collective intelligence, scientific capacity, and inclusiveness in local decision making (Aspuru, Herranz-Pascual, and Santander 2016; Dickinson et al. 2012; Wehn et al. 2018). They also provide the public with scientific information, which is accessible in forms they can understand and utilize (Golumbic, Fishbain, and Baram-Tsabari 2019).

Making sure that the tools used in CS are of scientific value and provide meaningful information to the volunteers, many authors call for more user involvement in the design of the services (Golumbic, Fishbain, and Baram-Tsabari 2019; Preece 2016; Robinson et al. 2018; Sanz et al. 2014; Skarlatidou et al. 2019). It is frequently reported that the lack of user involvement can cause problems both for the volunteers as well as for the scientists, affecting motivation and data quality, and to this end, it is suggested that Human-Computer Interaction (HCI) research should be incorporated in Citizen Science (Preece 2016). If data collection is too complicated, or too time-consuming, volunteers often lose their initial desire and drive to participate (Robinson et al. 2018; See, Fritz et al. 2016).

The advantages of well-designed ICT products are numerous; for example, they enable CS projects to engage volunteers more efficiently, increase scientific productivity and keep volunteers engaged for a longer time (Wald, Longo, and Dobell 2016). Further, involving users in the product design process increases their satisfaction with the final product (Mahmood et al. 2000). Moreover, understanding volunteers’ motivation and fulfilling their expectations increase their satisfaction, which in turn determines their level of participation (Wright et al. 2015). Consequently, involving the citizens at various design stages and incorporating their feedback increases their expectations and capacity for participation, while conversely, failing to do so inhibits satisfaction and participation (Mahmood et al. 2000).

Despite the apparent benefits of involving citizens in the design process of CS projects and tools, their involvement is not yet the mainstream practice. Resource limitations in projects usually constrain inclusive involvement of citizens, leading to the use of tools which might lack appropriate functionality (Wiggins 2013). Furthermore, limiting crowdsourcing only to a proof of concept level e.g. to test and develop a new CS tool might be de-motivating for the volunteers, who are otherwise altruistic and ready to substantially contribute (Havlik and Schimak 2014). Knowing certain features of tools which are known to work can help to bridge the gap for those who do not have the resources to involve volunteers in the design process, yet acknowledge the benefits. Few authors have attempted to create guidelines to facilitate the overall design process. For example, Jennett and Cox (2014) provided guidelines for virtual CS projects focusing on website interfaces. Similarly, Skarlatidou et al. (2019) summarized best practices for CS applications, with a focus on web interface features based on a systematic literature review. Our paper extends the scope of these technical guidelines covering a broader topic of volunteer motivation, which is the driving force that determines the types of technical features CS tools need to meet volunteer expectations and retain their interest long-term. The aim of the paper is to make a synthesis of current knowledge from the CS literature with respect to i) understanding types of CS related projects and the characteristics of the volunteers involved, including participation patterns, ii) grouping common volunteer motivations according to functional categories, and iii) summarizing strategies for retaining volunteers, with examples of functional features for CS tools.

This literature review draws best practices from inclusive CS projects including Citizen Observatories, since the trend is toward collaboration and co-design in line with the level of understanding and involvement of the volunteers. It aims at helping to manage volunteer expectations better and ultimately supporting the planning and management of successful and sustainable CS projects.

The work was conducted within the H2020 SMURBS/ERA-PLANET project¹ under a wider literature review on Citizens Observatories, which covered 352 articles found in SCOPUS, Web of Science and PubMed in July 2018. Detailed methodology can be found in Robinson (2019). For the purpose of this manuscript with a narrower focus on citizen involvement, only articles which address inclusive citizen science were considered. Additional articles were reviewed for the purpose of specific chapters to improve readability and to cover a joint discussion of motivations, volunteer experiences and suggestions for design.

2. Characteristics of CS projects and volunteers

What is common to all CS projects is that they involve both competent communities, i.e. the scientists and the volunteers who are participating as citizen scientists. The experts and CS practitioners involved in the projects include, but are not limited to, environmental scientists, social scientists, hardware and software specialists, and data analysts. Depending on the type of CS project, the target group of volunteers and other stakeholders might involve locals living nearby who are affected by the issue at hand (Grossberndt and Liu 2016; Uhrner et al. 2014), schools (Hunt et al. 2015; Kobernus et al. 2013; Mominó, Piera, and Jurado 2016), representatives of local officials (Mietlicki, Gaudibert, and Vincent 2012) and do-it-yourself (DIY) communities (Busch et al. 2016). In addition, external stakeholders such as scientists from other domains, NGOs, and government agencies may want to harness the data for various purposes, such as complementing their own data sets (Cooper et al. 2017; Ferster and Coops 2013; Hunt et al. 2015; Wehn and Evers 2015). It is observed that people who take part in CS projects are generally highly educated and belong to middle and upper socioeconomic classes (Soleri et al. 2016), are retired (Wright et al. 2015) or have an already established interest in the topic (Mazumdar et al. 2017).

The specific aims and goals of a Citizen Science project are to determine what kind of volunteers it seeks to engage and what kind of tools are best suited for its purpose (Wiggins and Crowston 2011). To meet the expectations of volunteers and CS practitioners, who typically consist of experts with various backgrounds (some without previous experience of CS), communicating the objectives early on is crucial to the process (Grossberndt and Liu 2016). These are commonly written using technical terms and need to be translated into a common language to be easily understood by the layperson (Sanz et al. 2014). Understanding and using the correct terminology can help define the aims already outlined in the project proposal at the design stage while reducing misunderstanding later on by clarifying participants’ roles, responsibilities and limitations (Grossberndt and Liu 2016; Wiggins and Crowston 2011). There is currently no single definition of CS, but there are a plethora of different types of CS projects that reveal the dynamics of this evolving research field (Sanz et al. 2014). The terminology used in the literature is summarized in Table 1. The definitions provide additional information about the level and type of participation.

Table 1. Citizen Science terminology.

Term	Definition	Reference
Citizen Science (CS)	The umbrella term of the voluntary involvement of the wider public in scientific research from data collection to research design	(Bonney 1996; Bonney, Cooper, et al. 2009; Hand 2010; Silvertown 2009)
Citizen Observatory (CO)	CO is an information ecosystem where citizens are empowered in community-based environmental monitoring with the capability to collect, report, access and share environmental information and connect and interact with other citizens, scientists and stakeholders to discuss, intervene and make decisions with the help of ICT enabled technologies.	(Degrossi et al. 2014; Grainger 2017; Lanfranchi et al. 2014; Liu et al. 2014; Miorandi et al. 2013; Mominó, Piera, and Jurado 2016; Montargil and Santos 2017a; Tapia et al. 2014; Zaman and Meuter 2015)
Community Remote Sensing (CRS)	Another umbrella term which “combines remote sensing with citizen science, social networks, and crowd-sourcing to enhance the data obtained from traditional sources. It includes the collection, calibration, analysis, communication, or application of remotely sensed information by these community means.”	(“IGARSS” 2010; Williamson 2009)
Community-Based Monitoring (CBM)/ grass root sensing/ civic or community science	Similar to participatory sensing, but a bottom-up process where concerned citizens, in collaboration with scientists, public authorities and further stakeholders monitor, track and respond to issues that arise from common community concern	(Breen et al. 2015; Burke et al. 2006; Haklay 2015; Whitelaw et al. 2003)
Crowdsensing	A sub-category of crowdsourcing in which individuals with sensing and computing devices collectively share data and extract information to measure and map phenomena of common interest	(Ganti, Ye, and Lei 2011; Miorandi et al. 2013)
Crowdsourcing	Outsourcing work to the crowd	(Howe 2006; Howe 2008)
Participatory sensing (PS) (also called citizen-or mobile (crowd)sensing, crowdtasking or participatory mapping)	Volunteers acting alone or in groups observing their environment with smart mobile devices or sensors forming interactive, participatory sensor networks that enable public and professional users to gather, analyze and share local knowledge	(Burke et al. 2006; Haklay 2015; Jennett et al. 2016)
Passive sensing/ opportunistic sensing/ passive crowdsensing/ Ambient Geographic Information (AGI)/ social geographic data (SGD)/ Involuntary volunteered geographic information (iVGI)	Automatic data collection through a volunteer’s resources without the active engagement of volunteers. E.g. social media harvesting. Passive sensing can also refer to a volunteer agreeing to host a sensor device in his/her backyard.	(Dell'Acqua and Vecchi 2017; Haklay 2015; Lane et al. 2010; Loukis and Charalabidis 2015; Stefanidis, Crooks, and Radzikowski 2013)
Public Participation in Scientific Research (PPSR), crowd science, crowd-sourced science, civic science, or networked science	Synonyms for CS	(Bonney, Ballard, et al. 2009; Hand 2010; Liu and Kobernus 2017)
Public Participatory Geographic Information Systems (PPGIS)	The use of geographic information systems (GIS) in community participation in a plethora of social and environmental contexts.	(Craig, Harris, and Weiner 2002; Newman et al. 2010; Sieber 2006)
Volunteer computing	Volunteers share the unused processing power of their computers and device and allow scientists to run complex computer models when the device is not used.	(Haklay 2015; Jennett et al. 2016)
Volunteer thinking	Making use of the volunteers’ cognitive resources in parallel with leisure activities or explicitly on their free time where volunteers, for example, classify or identify species or land cover types through a web application.	(Haklay 2015; Jennett et al. 2016)
Volunteered Geographic Information (VGI)	Spatially referenced data provided voluntarily by individuals	(Goodchild 2007a; Goodchild 2007b; Gómez-Barrón et al. 2016; See, Mooney et al. 2016)

In addition to the terminologies, CS projects can be classified according to different levels of engagement, e.g. contributory, collaborative and co-designed CS projects (Bonney, Ballard, et al. 2009; Shirk et al. 2012). The involvement of the public and the amount of control over different steps of the project increases in this classification. In contributory projects the volunteers are merely involved in data collection, where as in collaborative projects, in addition to data collection, they can also analyze and disseminate data and participate in the project design. Co-designed projects present the highest degree of engagement including the co-design phase, as well as active involvement in most or all aspects of the research itself. While some projects aim at long term volunteer commitment (e.g. birding communities [Sullivan et al. 2009]), others might rely on event-driven volunteer contributions, e.g. monitoring ash fall (Wallace, Snedigar, and Cameron 2015), earthquakes (Dell'Acqua and De Vecchi 2017), and invasive species (Crall et al. 2011).

The level and pattern of volunteer participation, as well as satisfaction, is known to vary and evolve (Ferster and Coops 2013; Grainger 2017; Houghton et al. 2019; Jennett et al. 2016; Wright et al. 2015). Some contribute consistently, others in bursts of activity, while others contribute more sporadically (Ferster and Coops 2013). Moreover, as Grainger (2017) points out, some of the observing activities are regular and thus advocates that the volunteers’ activities need to become institutionalized in the sense of having reoccurring patterns. However, in reality this might be disrupted due to unforeseen incidents such as an illness or other causes which hinder routine. Volunteers’ contribution levels might indeed plummet due to loss of interest or time, or because of changing prioritizations (Jennett et al. 2016). The decision of volunteers to contribute is influenced by whether the particular activity fits with the volunteers’ needs and goals (Clary and Snyder 1999; Wright et al. 2015). That is why understanding volunteer psychology, e.g. their motivations, will promote the efficient gathering of quality data while optimizing the benefits for the volunteers (Wright et al. 2015). Familiarity with the theories that explain participation patterns, e.g. the Technology Acceptance Model (TAM), Theory of Diffusion of Innovations (DIT), Theory of Planned Behavior (TPB) and the 90-9-1 rule for participation inequality, will help CS practitioners to understand the different roles and characteristics as well as the information needs of citizen scientists (Haklay 2016; Lai 2017). The TAM by Davis in 1986 focuses on acceptance of information technologies explaining user behavior. It is comprised of the Perceived Usefulness (PU) and Perceived Ease of Use (PEU) components. The DIT Theory, developed by Rogers in 1995, helps researchers understand the acceptance and adoption of innovations among individuals and organizations over time. Rogers’ famous S-shaped curve encapsulates the innovators, early adopters, early majority, late majority and laggers. The TPB was developed by Ajzen in 1991 and theorizes the behavioral intention of the person’s attitudes toward a particular behavior, broken down into attitude, subjective norms and perceived behavioral control. Participation inequality or the 90-9-1 rule described by Nielsen in 2006 is the phenomenon where a small percentage of volunteers contribute a significant proportion of information to the total output, while the majority are less involved. To give an example, in the well-known Zooniverse project, only 4–7% of volunteers actively contribute over long periods (Sauermann and Franzoni 2015), which implies that even the most successful projects need robust mechanisms to increase and sustain participation. The upside however is that these “super volunteers” are most likely able to contribute consistently and submit good quality data (Jennett et al. 2016).

Similarly, several authors have identified different roles of volunteers and participation characteristics. For example Cooper et al. (2017), divides volunteers into two categories, data collectors and data consumers, whereas Mominó, Piera, and Jurado (2016) introduce the so called “makers”, “observers” and “analyzers” with different levels and roles of participation. More specifically, the makers describe people with abilities and interests in the technologies with the capacity to build DIY observation devices. The observers collect observations and the analyzers interpret the data. Following a different classification Kobernus et al. (2013) have identified six citizen roles. The Observer makes the initial observation, the Publisher makes the observation discoverable by others, the Discoverer finds an existing data resource, the Service Provider makes the information accessible to others to use and download, and finally, the Service Orchestrator who combines existing services for a distinct purpose i.e. a smartphone application, and the Decision maker who exploits the data to make an informed decision.

3. Motivation of CS volunteers

Motivation drives people to take action (Kaufmann, Schulze, and Veit 2011). The higher the motivation, the more the volunteers are satisfied with the project outcomes and are willing to continue to contribute (Clary and Snyder 1999; Wright et al. 2015). The nature of volunteerism can be multi-motivational, a volunteer having both altruistic and egoistic reasons to participate e.g. both wanting to selflessly help as well as desire to benefit oneself (Clary and Snyder 1999). Motivation can be classified into intrinsic and extrinsic motivation. Intrinsically motivated individuals might, for example, participate due to the project being fun, while extrinsically motivated individuals might be motivated due to short or long term pay offs, such as direct compensations or improved skills (Kaufmann, Schulze, and Veit 2011). Intrinsically motivated people are more likely to stay committed for longer (See, Fritz et al. 2016; Wright et al. 2015), while using incentives, such as rewards might have the opposite effect on volunteerism, due to the perception of external control versus personal control (Clary and Snyder 1999; Gharesifard and Wehn 2016). Clary and Snyder (1999) indeed observed that volunteerism can be induced on the short term with incentives, i.e. external control, yet the stronger perception of personal control to volunteerism, i.e. participating from altruistic reasons, is the drive for contribution in the long run. The level of participation, as well as the goals and attitudes driving the participation, can vary over time (Kaufmann, Schulze, and Veit 2011; Wright et al. 2015). Initial interest to participate might be triggered by e.g. curiosity, interest in science and desire to contribute to research, which are factors that may not necessarily justify ongoing and sustained engagement (Jennett et al. 2016).

The motivational factors reported in the literature are thematically grouped and summarized in Table 2. Clary and Snyder’s (1999) and Wright et al.’s (2015) Volunteer Functions Inventory (VFI) was used to classify the different factors into five categories: (i) Values (ii) Personal development (iii)) Career and recognition (iv) Social, and (v) Recreation. Each are provided with an example that binds them with the concept of CS. Volunteer motivation can be studied through stakeholder and volunteer surveys as well as focus groups and analyzed with qualitative content analyses e.g. with coding, or with quantitative research methods using statistical tests (Clary and Snyder 1999; Kaufmann, Schulze, and Veit 2011; Robinson et al. 2018; Wright et al. 2015). Surveys should be designed together with scientists experienced in social sciences, and if necessary, approved by the researchers’ ethical committee (Druschke and Seltzer 2012).

Table 2. Summary of volunteer motivations.

Function	Conceptual example in environmental CS	Reference in CS literature
Values	The wish to contribute to science or society (i.e. civic responsibility) while helping the environment	(Bruyere and Rappe 2007; Dickinson et al. 2012; Grainger 2017; Grossberndt and Liu 2016; Hunt et al. 2015; Jennett et al. 2016; Montargil and Santos 2017b; Robinson et al. 2018; Wright et al. 2015; See, Fritz et al. 2016)
Personal development	Learning opportunities (about the scientific subject area and data collecting and analyzing techniques, environmental awareness, familiar places)	(Bruyere and Rappe 2007; Devictor, Whittaker, and Beltrame 2010; Dickinson et al. 2012; Grainger 2017; Havlik and Schimak 2014; Hunt et al. 2015; Jennett et al. 2016; Montargil and Santos 2017b; Robinson et al. 2018; See, Fritz et al. 2016)
Career and recognition	To gain relevant experience related to their career interests and to gain recognition and other personal benefits for their input	(Grainger 2017; Havlik and Schimak 2014; Hunt et al. 2015; Jennett et al. 2016; Kotovirta et al. 2015; See, Fritz et al. 2016; Wright et al. 2015)
Social	Social interaction and being part of a community of likeminded individuals	(Bruyere and Rappe 2007; Dickinson et al. 2012; Grainger 2017; Montargil and Santos 2017b; See, Fritz et al. 2016; Wright et al. 2015; Jennett et al. 2016)
Recreation	To have fun and to undertake new activities as part of existing recreational activities while commonly being outdoors	(Grainger 2017; Havlik and Schimak 2014; Jennett et al. 2016; See, Fritz et al. 2016; Tserstou et al. 2017; Wright et al. 2015)

Intrinsic motivation drives many people to volunteer. Protecting and improving the environment and making it a better place to live are amongst the most well-documented motivations in environmental CS projects (Bruyere and Rappe 2007; Havlik and Schimak 2014; See, Fritz et al. 2016). Many feel a civic responsibility to address certain issues affecting local communities (Hunt et al. 2015; Montargil and Santos 2017b; See, Fritz et al. 2016) and like to see the impact of their contribution to policy making, fulfilling them with the feeling of empowerment (See, Fritz et al. 2016). Contributing to science is also a strong motivation (Dickinson et al. 2012; Grainger 2017; Robinson et al. 2018; See, Fritz et al. 2016; Wright et al. 2015).

Citizen science projects can offer informal- (Hunt et al. 2015) and indirect community- (Jennett et al. 2016) learning opportunities and contribute to robust learning outcomes (Dickinson et al. 2012). Some volunteers want to learn more about familiar places, i.e. their neighborhood (Devictor, Whittaker, and Beltrame 2010; Grainger 2017; Robinson et al. 2018) and the natural environment (Bruyere and Rappe 2007), while at the same time enabling them to gain valuable and relevant information about the topics addressed in the project (Grainger 2017; Montargil and Santos 2017b). Others are interested in the science behind the project (See, Fritz et al. 2016) and the technicalities of data collection and analysis, which also enables them to make new “discoveries” while exploiting the collected data (Grainger 2017; Robinson et al. 2018). Curiosity may also drive many volunteers to participate (Devictor, Whittaker, and Beltrame 2010; Havlik and Schimak 2014).

In addition to informal learning opportunities, CS projects can help people gain valuable knowledge useful for their current or future career, especially in Science, Technology, Engineering and Mathematics (STEM) (Hunt et al. 2015). The volunteers also expect to gain some recognition for their input, e.g. through feedback and interaction with the CS community and scientists (Grainger 2017; Havlik and Schimak 2014; Kotovirta et al. 2015; See, Fritz et al. 2016). Some common incentives are monetary rewards (Kotovirta et al. 2015) and prizes (See, Fritz et al. 2016). However, Clary and Snyder (1999) and Gharesifard and Wehn (2016) warn about their demotivating effect because of their perceived external control, and counter that a greater perceived personal control will lead to stronger motivations and greater intentions to volunteer. Recognition of volunteers’ contribution can also be accomplished by providing them with an opportunity to perform more complex tasks, which in turn also requires more responsibility allowing them to achieve “expert status” in the project (See, Fritz et al. 2016). An example of this approach is providing experienced participants the role of moderating discussion forums (Jennett et al. 2016). For others, it is important to feel like an active participant and co-owner in the project (See, Fritz et al. 2016). Bruyere and Rappe (2007) and Wright et al. (2015) emphasize the need to recognize the significance of volunteers’ contributions, which will satisfy and enhance the volunteers’ intrinsic motivation.

Participating in activities in a community of likeminded people is an important social motivation (Bruyere and Rappe 2007; Dickinson et al. 2012; Grainger 2017; Montargil and Santos 2017b; See, Fritz et al. 2016; Wright et al. 2015). The social interaction with people with similar interests can be perceived as an important means of socializing and making new friends (Bruyere and Rappe 2007; See, Fritz et al. 2016).

Being able to connect the CS activity with a volunteer’s existing recreational activities will nurture the volunteer’s motivation to keep it fun (Grainger 2017; Havlik and Schimak 2014; See, Fritz et al. 2016). These activities can take place outdoors while performing observations in the natural environment, e.g. in biodiversity-related CS projects, or by performing tasks using a computer at home (online CS projects) (Grainger 2017; Tserstou et al. 2017; Wright et al. 2015).

4. Retention of CS volunteers

Retaining motivated volunteers is crucial for the sustainability of a CS project (See, Fritz et al. 2016). If citizens do not see any value and usefulness in their participation, loss of interest and abandonment of effort is likely (Ferster and Coops 2013; Wright et al. 2015) as is a lack of initial engagement (Gharesifard and Wehn 2016; Wright et al. 2015). In addition to the common constraint of lack of time, other barriers for sustained involvement include the tasks’ complexity, being mundane, or even apprehension for submitting incorrect data (Jennett et al. 2016). See, Fritz et al. (2016), and Dickinson et al. (2012) suggest communication strategies which take into account both volunteer recruitment and retention. In order to get the word out in recruiting the volunteers, Dickinson et al. (2012) suggest using well-timed press-releases that are picked up by national and local media. Newsletters, blogs, and social-networking groups help to create a sense of community, in addition to active forms of communication, such as contents, badging systems and methods to recognize the effort of the volunteers (Dickinson et al. 2012). Providing rapid feedback to volunteers and regular communication on their contributions have also shown to work in terms of retaining a community (See, Fritz et al. 2016). Clary and Snyder (1999) emphasize that those recruitment strategies, which address the specific motivational functions underlying behavior and attitudes, are the most successful. Montargil and Santos (2017a) underline the critical balance between the participation effort and return benefits. Many projects lack motivation mechanisms and rely on volunteers’ goodwill and interest to assist in research (Kotovirta et al. 2015). However, several approaches have been established to motivate volunteers. These approaches are summarized in Table 3, with examples of functional features.

Table 3. Summary of approaches to retain volunteers.

Feature	Explanation/Example of a functional feature	Sources
Fun and cool	The project tasks should be cool and tools fun to use while nurturing curiosity	(Castell et al. 2015; Grainger 2017; Havlik and Schimak 2014; Kaufmann, Schulze, and Veit 2011; See, Fritz et al. 2016; Simpson, Page, and De Roure 2014)
Simple, easy, versatile and user-friendly tools	Tools adapted to different user groups according to their skills	(Druschke and Seltzer 2012; Gharesifard and Wehn 2016; Grainger 2017; Havlik and Schimak 2014; Kobernus et al. 2013; Kotovirta et al. 2015; Mazumdar et al. 2017; Newman et al. 2010; Prakash et al. 2004; See, Fritz et al. 2016; Wallace, Snedigar, and Cameron 2015)
Provide relevance and meaning to volunteers’ everyday lives and interests	Address environmental issues, which affect the direct environment of the volunteers	(Ferster and Coops 2013; Grossberndt and Liu 2016; Guillaume et al. 2016; Hunt et al. 2015; Kotovirta et al. 2015; Mominó, Piera, and Jurado 2016; Montargil and Santos 2017b; Williamson 2009)
Rapid feedback on volunteers’ contributions	E.g. visualization of the measurements on a map	(Druschke and Seltzer 2012; Grainger 2017; Gutiérrez et al. 2016; Kotovirta et al. 2015; Liu and Kobernus 2017; Montargil and Santos 2017a; See, Fritz et al. 2016; Zaman and De Meuter 2016)
Communication and recognition of the importance of volunteers’ contribution and scientific outcomes	Certificates of recognition, communication on how results are being used	(Aspuru, Herranz-Pascual, and Santander 2016; Cohn 2008; Dickinson et al. 2012; Druschke and Seltzer 2012; Grainger 2017; Grossberndt and Liu 2016; Havlik and Schimak 2014; Kotovirta et al. 2015; Mackay et al. 2015; See, Fritz et al. 2016; Sullivan et al. 2009; Wright et al. 2015)
Provide access to volunteers’ contributions and the possibility to analyze and share them	Enable volunteers to make discoveries from their data	(Busch et al. 2016; Charvat et al. 2018; de Assis et al. 2018; Gharesifard, Wehn, and van der Zaag 2017; Grainger 2017; Hunt et al. 2015; Jirka, Remke, and Bröring 2013; Liu and Kobernus 2017; Mackay et al. 2015; Marantos et al. 2017; Mazumdar et al. 2017; Mietlicki, Gaudibert, and Vincent 2012; See, Fritz et al. 2016)
Rewarding and gamification	Reputation ranking, competition between volunteers, prices	(Castell et al. 2015; Dickinson et al. 2012; Gutiérrez et al. 2016; Havlik and Schimak 2014; Kotovirta et al. 2015; Mominó, Piera, and Jurado 2016; See, Fritz et al. 2016)
Create a sense of community	E.g. discussion forums, social networking groups	(Cooper et al. 2017; Dickinson et al. 2012; Druschke and Seltzer 2012; Kotovirta et al. 2015; Mackay et al. 2015; Mazumdar et al. 2017; See, Fritz et al. 2016; Sullivan et al. 2009; Williamson 2009)
Enable feeling of co-ownership and control over the scientific process	Actively involve the volunteers in the design, implementation and dissemination of the project	(Breen et al. 2015; Gharesifard, Wehn, and van der Zaag 2017; Grossberndt and Liu 2016; Jones et al. 2010; Liu and Kobernus 2017; Mackay et al. 2015; Mazumdar et al. 2017; See, Fritz et al. 2016)
Actively remind volunteers to contribute	Pop-up features in smartphone apps and using geofencing functions near desirable locations	(Kotovirta et al. 2015; Mazumdar et al. 2018)
Organize training	Adequate information on the data collection procedure either in forms of training or user guides as well as receiving help on demand	(Aspuru, Herranz-Pascual, and Santander 2016; Busch et al. 2016; Degrossi et al. 2014; Druschke and Seltzer 2012; Grainger 2017; Grossberndt and Liu 2016; Hunt et al. 2015; Jones et al. 2010; Kotovirta et al. 2015; Liu and Kobernus 2017; Mietlicki, Gaudibert, and Vincent 2012; Mominó, Piera, and Jurado 2016; See, Fritz et al. 2016; Wehn et al. 2015)

The above-listed features match with the volunteers’ motivational functions, which CS practitioners should nurture. For example, Havlik and Schimak (2014) argue that motivation determines the fate of the built infrastructure and therefore an application designed to be used by the volunteers should be fun, nourish curiosity and enable the user to make a difference, in order for them to stay motivated. Table 3 can function as a checklist for CS practitioners, first during the design phase of the project to include as many features as possible, and secondly, to evaluate the project, whether these points were met to allow for correcting actions or a post-evaluation of the user-centricity of the CS project. The analysis, however, should take into account certain tradeoffs between the features of Table 3 and other requirements of the project, such as robustness and scientific integrity of measurements as well as quality assurance (QA) and quality control (QC) procedures.

Resource limitations in CS projects often require adopting suboptimal ICT tools which come with hidden costs from poor usability and lack of appropriate functionality (Wiggins 2013). In order to adapt the tools to meet the various volunteers’ motivations and needs, they should be useful, easy to use and versatile (Grainger 2017; Havlik and Schimak 2014; Kobernus et al. 2013; Kotovirta et al. 2015; Mazumdar et al. 2017; Prakash et al. 2004; Wallace, Snedigar, and Cameron 2015). The first step to achieve this is to understand and adapt to the citizens’ needs as it is not enough to provide them with some arbitrary tools (Lawton et al. 2011; Liu and Kobernus 2017). Secondly, it is important to provide data in an easy to comprehend format. As Grainger (2017) points out, data becomes information only after someone understands the method of data generation, knows how to interpret the data and knows how to use the data meaningfully. If data is presented only as lists of figures or spreadsheets that only experts can interpret, the project will lose part of its audience (Kobernus et al. 2013). Golumbic, Fishbain, and Baram-Tsabari (2019) suggest to include Human-Computer Interaction (HCI) and User Centered Design (UCD) in the design process, since there is no universal user, and hence no universal interface. In this way, the design process becomes more inclusive and will increase the overall attractiveness of the interface. Nevertheless, in practical terms, many CS projects simply lack the resources to involve users in the design process, or it might not be their aim or priority, in which case, a good tradeoff would be to use preexisting tools, which are already deemed fit for purpose. If the project explicitly aims at developing a new tool, the volunteers should be involved in the design process, where as if the aim is to collect scientific data, it is more cost-efficient to use existing CS tools proven to be user-friendly.

Druschke & Seltzer (2012) emphasize that it is critical to provide access to the data the volunteers collect, even preliminary ones, as soon as possible, in order to allow them to provide feedback and inquire on an issue that might interest them, while at the same time providing them with a feeling of connectedness to the study and demonstrating the value of their involvement. Providing rapid feedback might even involve utilizing suboptimal tools, but still, this expectation can sometimes be difficult to meet depending on analytical demands combined with time constraints.

See, Fritz et al. (2016) emphasize how user needs can change, and there is a need to understand the volunteers’ skills, expectations and interest to adjust the developed tools accordingly, which might mean that data analysis, modeling and presentation of results must be adjusted for distinct user groups (Gharesifard, Wehn, and van der Zaag 2017; Grainger 2017; Kobernus et al. 2013). Golumbic, Fishbain, and Baram-Tsabari (2019) suggest a multilayer information display to address different user needs. For example, scientists might be interested in the raw data, while the general public might prefer off-the-shelf data visualization products (Prakash et al. 2004). An example of adaptation to different user needs in data visualization platforms is given in Table 4 and is based on the findings of Prakash et al. (2004). However, many features depend on the level of desired participation, e.g. in virtual projects where participants analyze data vs. investigative projects where volunteers submit data, as well as the project goals (e.g. conservation vs. investigation), and physical location (e.g. local, global or virtual) (Wiggins and Crowston 2011).

Table 4. Adapting to user needs after Prakash et al. (2004).

User groups	Example of visualization
The general public and young children	Simple visualization of processed data
Decision and policy makers	Retrieve data for different scenarios in order to make decisions
Students	Possibility to change input parameters and explore the data to fulfill curiosity
Web and tool developers	Access to archived and near real-time data for web interface and tool development purposes
Researchers	Access to raw and original data for research purposes

Citizen science campaigns should allow volunteers to match their interests with those environmental issues that directly affect their environment and provide relevance and meaning to their daily lives (Grossberndt and Liu 2016; Kotovirta et al. 2015). They can do this by involving communities, such as schools and other local populations directly affected by the issue (Ferster and Coops 2013; Grossberndt and Liu 2016). While the CS project will provide opportunities for informal training of the volunteers, the scientists will also gain valuable information as these locals will hold valuable knowledge of the study area and related issues and who can be actively involved in solving them (Ferster and Coops 2013; Guillaume et al. 2016; Hunt et al. 2015; Mominó, Piera, and Jurado 2016; Montargil and Santos 2017b; Williamson 2009). Mobile applications should be used both for extracting and disseminating information. The data should be visible to the end user since it can affect their willingness to participate and is a critical success factor (Montargil and Santos 2017a). The volunteers should receive instant qualitative feedback by, for example, displaying data on a live map (See, Fritz et al. 2016).

The desire by the volunteers to contribute to science and society means CS practitioners should frequently communicate the importance of their contribution and how their data is being utilized (Kotovirta et al. 2015; Mackay et al. 2015; See, Fritz et al. 2016; Wright et al. 2015). According to Grossberndt and Liu (2016) to foster meaningful participation, it is essential to reassure the volunteers that the results are being used to solve real issues. Similarly, Aspuru, Herranz-Pascual, and Santander (2016) noticed the importance of participants reaffirming that their observations can affect the decision-making process. It is also important to acknowledge their input, e.g. rewarding them with certificates of recognition (Dickinson et al. 2012; See, Fritz et al. 2016).

Providing access to individual contributions is rewarding for volunteers (Sullivan et al. 2009) and will boost their curiosity and personal developmental goals when they are given the opportunity to analyze their data (de Assis et al. 2018; Gharesifard, Wehn, and van der Zaag 2017; Hunt et al. 2015; Marantos et al. 2017). Sharing the data is also an integral part of establishing trust, fairness and value for those who contribute, while it increases comprehension of environmental issues by the general public (Ferster and Coops 2013; Mietlicki, Gaudibert, and Vincent 2012; Miorandi et al. 2013).

Rewarding volunteers is an effective way to encourage and support participation (See, Fritz et al. 2016). On the one hand, Clary and Snyder (1999) acknowledge the efficiency of external rewards, yet on the other, they caution that once the external pressures to perform observations is removed, the motivational force to participate might shift and thus advocate for using methods which foster intrinsic motivation. In addition to providing access to individual data, which were classified separately, providing different levels of progression or reputation ranking and introducing elements of competition between volunteers including contests, incentives, and badging systems, might be rewarding to some of the volunteers (Dickinson et al. 2012; Kotovirta et al. 2015; See, Fritz et al. 2016). Increasingly CS projects are incorporating elements of gamification, for example, by asking users to conduct micro-tasks, e.g. classifying images, which is a type of volunteer thinking (See, Fritz et al. 2016). Gamification can help to solicit and maintain contributions from wider, already existing, communities, e.g. online communities (Simperl et al. 2018). It transforms the crowdsourcing procedure and makes it more engaging and fun (Gutiérrez et al. 2016). Gamification can also incorporate educational functions which can both enhance participation and learning (Mominó, Piera, and Jurado 2016). Some volunteers can indeed find playing a game rewarding (Jennett et al. 2016).

To support a volunteer’s desire to feel part of a community of likeminded people, adding social networking options to the crowdtasking might boost their motivation (Dickinson et al. 2012; See, Fritz et al. 2016). Creating a sense of community through discussion forums and social networking groups will enable social interactions with people of similar interests, e.g. to be able to connect and share experiences with peer contributors (Kotovirta et al. 2015). Enabling volunteers to interact might also lead to scientific discoveries that would have been otherwise overlooked by scientists, the project forum being a key space for such discussion between fellow citizen scientists (Jennett et al. 2016). Connecting people to frame and solve issues together will also create a sense of being part of a community (Williamson 2009). In some cases, merely watching submitted online data from other contributors can be enough for others to feel part of a group (See, Fritz et al. 2016).

Stakeholders need to achieve sufficient levels of trust, ownership and continuity to achieve the desired outcomes of social learning and engagement (Wehn et al. 2018). Volunteers should be offered ways to feel inspired, involved, and active in order to affect educational, attitudinal, behavioral, and scientific outcomes (Druschke and Seltzer 2012). Volunteers need to be actively and iteratively involved in the different aspects of the project, e.g. in the design, implementation and dissemination phases to foster the feeling of co-ownership in the project (Breen et al. 2015; Gharesifard, Wehn, and van der Zaag 2017; Grossberndt and Liu 2016; Mazumdar et al. 2017). Being involved will also give the volunteers a feeling of control over the scientific process (Grossberndt and Liu 2016). Furthermore, the more aspects of the project the volunteers are involved in, the more likely they are to retain interest; this includes especially the possibility to engage in various activities beyond the main task, such as discussion forums in online CS projects (Jennett et al. 2016). Yet, Wright et al. (2015) caution that there is a fine line between overextending and supporting participation, and a balance needs to be struck and managed for equitable participation.

With regards to trust, Barcellos et al. (2016), Gharesifard and Wehn (2016) and Golumbic, Fishbain, and Baram-Tsabari (2019) suggest providing information about the origin of the data and metadata in the platforms and using disclaimers. The metadata should include, for example, information about the observations, i.e. under what conditions the data were acquired, type and accuracy of the equipment used and how often equipment is calibrated (Gharesifard and Wehn 2016). The disclaimer should inform the user about the reliability of the data (Gharesifard and Wehn 2016).

As discussed in section 2, user patterns and levels of participation are known to vary. To keep users active, or to revitalize their activity, the users can be reminded to contribute, for example, with the help of the geofencing approach (Kotovirta et al. 2015; Mazumdar et al. 2018). Geofencing can be used to trigger a volunteer to act near a pre-defined location where they will receive a notification on their smartphones and can decide to act or disregard the message.

Volunteers should also receive training and sufficient information before data collection campaigns (Mietlicki, Gaudibert, and Vincent 2012; See, Fritz et al. 2016). They need to be provided with adequate information on the data collection procedure and to receive help on demand (Aspuru, Herranz-Pascual, and Santander 2016; See, Fritz et al. 2016). The information can be passed on remotely by providing volunteers with user-guides, instruction sheets, guidelines and protocols on the use of the measuring device and observation recording, or face-to-face guidance during a workshop or training session (Grainger 2017; Hunt et al. 2015; Liu and Kobernus 2017; See, Fritz et al. 2016; Wehn et al. 2015). Trained volunteers can then use their gained knowledge to train more users (Kotovirta et al. 2015). Training of volunteers will create win-win situations; the volunteers will increase their skills, knowledge and experience while the scientists will receive higher quality data (Aspuru, Herranz-Pascual, and Santander 2016; Jones et al. 2010; Liu and Kobernus 2017; Mietlicki, Gaudibert, and Vincent 2012; See, Fritz et al. 2016).

Investing time and effort to involve citizens does not guarantee success. Reporting on a project that appeared to be a success, at least regarding scientific outcome, Druschke and Seltzer (2012) warned other CS practitioners that while superficially it may seem that all aspects of CS engagement have been taken into account, it is not enough to provide user-friendly info and tools and to set up a community communication medium. The volunteers need to be actively involved from the design phase onwards. Otherwise the apparent efforts towards engagement of volunteers in scientific data collection does not guarantee dual benefits e.g. positive learning outcomes for the participants.

5. Sustainability

Projects face many challenges concerning their long term sustainability. In CS projects, sustainability is guaranteed when both volunteers and the infrastructure e.g. data platform is maintained over the long term, and therefore retaining motivated volunteers is not the only aspect of sustainability (Ferster and Coops 2013; See, Fritz et al. 2016; Tapia et al. 2014; Wright et al. 2015). Several previously discussed points affect the sustainability of the project, such as creating tools and services which are usable, versatile, appealing, engaging, easy to use and intuitive (Aspuru, Herranz-Pascual, and Santander 2016; Barcellos et al. 2016; Botteldooren et al. 2013; Busch et al. 2016; Castell et al. 2015; Havlik and Schimak 2014; Kotovirta et al. 2015; Simpson, Page, and De Roure 2014). However, in order to sustain the infrastructure and other elements of the project, actions should be taken to foster a broader and more longer-lived influence beyond involved volunteers. In addition to providing the volunteers with access to the data, providing access to the general public (Ferster and Coops 2013; Montargil and Santos 2017a) and other interested parties, such as third-party developers (Ferster and Coops 2013; Miorandi et al. 2013) in the form of open APIs that extend the project’s reach to other platforms and applications, can foster its continuation and long term success. For this, it is important to use open standards for interoperability to complement and re-use existing datasets (Botteldooren et al. 2013; Busch et al. 2016; Liu and Kobernus 2017). By using standardized parameters and methods, data can be accepted more widely into other data repositories and to reach a wider user base (Busch et al. 2016), while interoperability becomes imperative when fusing data from different platforms and developing “universally” operated applications.

Creating a business model, e.g. selling a software license or finding organizations interested in the data or its user groups (Havlik and Schimak 2014), while making sure the data is preserved, curated and documented in a data management plan (Mietlicki, Gaudibert, and Vincent 2012; See, Fritz et al. 2016), can sustain the project features beyond its lifetime. Commercially driven projects and cross-financing can help to sustain a project financially (Gharesifard, Wehn, and van der Zaag 2017; Vincent et al. 2011). Close collaboration with well-known existing organizations and integrating the interactive features of CS using their infrastructure, such as enabling citizens to contribute observations through their webpage, can further promote the longevity of the project (Cooper et al. 2017; Grainger 2017; Sanz et al. 2014; Wehn and Evers 2015). This is both an efficient way to reach out to the target audience and at the same time, assure a communication network that can be used for information and support.

Reaching wider audiences via mass participation and scaling up the project geographically will increase both the number of observations and their geographical coverage (Gutiérrez et al. 2016). Other means to increase a broader uptake are, for example, the promotion of tools and services (Liu and Kobernus 2017) as well as ensuring they provide relevance and meaning to various stakeholders, such as enabling them to influence governance (Montargil and Santos 2017b; Vincent et al. 2011). Finally, Grossberndt and Liu (2016) call for close project documentation of both successes and failures for other projects as a mean to learn from, while Sanz et al. (2014) emphasize the need for evaluation of the projects.

6. Conclusions

This paper provides a typology of CS projects, and introduces a joint discussion on the motivation and expectations of volunteers, together with functional features of CS tools to retain them, including approaches which help to sustain a CS project over the long term. The broad application areas and types of CS projects make it difficult to describe a one-size-fits-all solution, which is why knowledge and suggestions summarized here should be considered and used as a map and reference point to inspire and guide CS practitioners in adopting specific aspects tailored to their projects. This is further supported, as emphasized previously by Freitag and Pfeffer (2013), by the fact that different CS projects perceive success differently, which also depends on the particularities of the local setting. As the field of inclusive CS continues to grow, the joint discussion presented about volunteer motivation, retention strategies and design aspects is unambiguously of high value for the CS community. The value cross-cuts scale from the local to the global. Both scholarly and practice-based communities can adopt good practices presented here and in conjunction with the continuous technological innovations make the difference in incorporating CS practices in local use cases, while, given the increasing role of local views for addressing global issues, enhance the international relevance of CS i.e. in support of the Sustainable Development Goals (SDG) frame (Fraisl et al. 2020; Fritz et al. 2019).

In summary, this paper makes the following recommendations:

1. Clarity of project aims as well as managing expectations early on in the process are vital elements of success.

2. Recruitment of volunteers and stakeholders should be based on relevance criteria and after thorough collection and comprehension of their motivations and needs.

3. Volunteers should be actively involved in all stages of the project and its products.

4. Functional features should match the volunteers’ expectations and for this knowledge gained during their implementation should be fed back.

5. The sustainability of projects is better served when relevant factors are addressed early in the design phase and potential exploitation scenarios are foreseen.

Importantly, citizens should have a central role in CS projects. Due to the transdisciplinary nature of CS projects, the focus should not only be on scientific outcomes, but also on the volunteers themselves and the participatory process needed for their efficient involvement. Examining and taking into account volunteers’ motivation and needs during the project design stage will result in longer-term participation retention, as long as their motives are fulfilled and a win-win situation is accomplished. Practitioners can achieve this by providing features within the tools used in CS projects which enable the matching of volunteers’ expectations and motivation to their benefit. A dynamic balance between co-benefit of volunteers and scientific objectives should be established for best-fit-for-purpose. Overall, for the optimization of CS applications, volunteers and infrastructure should be regarded as one integral ecosystem, whose good health significantly drives long term sustainability and success.

Acknowledgements

The authors would like to thank all SMURBS project partners who contributed in reviewing the CS and CO literature as part of the preparation used for SMURBS/ERA-PLANET project deliverable 3.5 “Citizen Observatories (COs) implementation”, a work which inspired the synthesis of this manuscript. We also thank the anonymous reviewers for providing feedback that greatly improved this manuscript.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was funded by Slovenian Research Agency/Javna Agencija za Raziskovalno Dejavnost Republike Slovenije (ARRS) program P1-0143 “Cycling of substances in the environment, mass balances, modelling of environmental processes and risk assessment and the and the ARRS Young researchers programme. This work was partly funded by ERA-PLANET (www.era-planet.eu), trans-national project SMURBS (www.smurbs.eu) (Grant Agreement No. 689443) under the EU Horizon 2020 Framework Programme.

Notes

1 Smurbs.eu