TY - JOUR
AU - Jensen, Ulrich, Thy
AB - Abstract Involving volunteers in the production and delivery of public services is a core policy objective of governments around the world. While existing research on volunteer involvement in service production, for example, has focused on advantages and disadvantages of such involvement and different dimensions of volunteer involvement, little is known about service professionals' response to volunteer involvement in public service production. Integrating perspectives from multiple theories, we build a theoretical framework for understanding how and when service professionals come to see volunteers as a threat to the quality of service, the profession's privileged position and monopoly, and professionals' own work tasks and job security. Based on a central distinction between production of core and complementary tasks, we propose that volunteers come to be seen as a threat in the eyes of service professionals when volunteers solve core rather than complementary tasks. Using a survey experiment among health assistants at nursing homes, we find partial support for our argument. Health assistants are more likely to perceive volunteers as a threat to the quality of care when volunteers solve core rather than complementary tasks. The study guides research toward a more nuanced understanding of volunteer involvement in service production in public organizations. Introduction In recent decades, agencies and organizations at all levels of government have increased efforts to get volunteers to assist service professionals with the production and delivery of services. Benefits from volunteer involvement in service production have been argued to include positive effects such as increased cost efficiency and improvement of service quality (Brudney 1993; Brudney and Gazley 2002; Nesbit, Christensen, and Brudney 2018). In times of continuous demands from elected officials and citizens for public organizations “to do more with less,” involving volunteers in service production may therefore represent a real lever for alleviating heavy workloads of service professionals and bringing about greater flexibility in the service production process. While existing research on involving volunteers in public and nonprofit organizations' service production has focused on, for example, understanding the advantages and disadvantages of such involvement (Brudney 1993; Brudney and Gazley 2002; Gazley and Brudney 2005) and different dimensions of volunteer involvement (for a review, see Nesbit, Christensen, and Brudney 2018), less empirical knowledge on volunteer-staff relationships exists (few exceptions are Rimes et al. 2017; Rogelberg et al. 2010; Wandersman and Alderman 1993). This is surprising since scholars have raised concerns that service professionals may view volunteer involvement in service production as threats to their professional standards, profession, and own job (see, e.g., Alford and O'Flynn 2012; Nesbit et al. 2016). Moreover, recent empirical studies reveal that volunteers today undertake tasks previously performed by paid staff (Chum et al. 2013; Handy, Mook, and Quarter 2008). Despite this development, we lack empirical research about what may cause professional staff to perceive volunteer involvement in service production as a threat to their job or the quality of services. In this article, we respond to this shortcoming in two ways. First, we combine insights from coproduction theory (Brandsen and Honingh 2016), the sociology of professions (Andersen and Pedersen 2012; Freidson 2001; Roberts and Dietrich 1999), and group conflict theory (McLaren 2003) to develop theoretical arguments for how and when service professionals come to perceive volunteer involvement in service production as a threat. Integrating these theoretical perspectives, we propose that service professionals are more likely to perceive volunteers as a threat to the quality of service, the profession's privileged position and monopoly, and their own work tasks and job security, when volunteers solve tasks identified by professionals as core rather than complementary. Second, we test our theoretical propositions using a survey experimental vignette design. In the experiment, subjects were randomly assigned to either a vignette text describing volunteers solving a core task, volunteers solving a complementary task, or to a baseline group receiving only an introductory text. Subsequently, we measured service professionals' perceptions of volunteers as threats. The study is conducted among health assistants working at nursing homes in Denmark. This service area is chosen because Danish nursing homes rely extensively on volunteers to assist with various tasks (FOA 2016), and volunteers solve a variety of tasks, including tasks perceived as core and as complementary by health assistants themselves. Our empirical analyses show that health assistants are more likely to perceive volunteers as a threat to the quality of care provided to elderly residents when volunteers solve a core task in contrast to a complementary task. Our data do not lend support to an effect on other outcome measures of threat perceptions. The results have important implications for practice. Perceptions of volunteers as threats have been shown to lead to tension and conflict between volunteers and service professionals (Kreutzer and Jäger 2011; MacDuff 2011; Romanofsky 1973). Nesbit and colleagues echo this concern, noting “… professionals often view volunteers as a threat to their professional standards and approaches, which creates tensions between the two groups …” (2016, 170). Service professionals might also be reluctant to or resist working with volunteers if they perceive volunteers as threats to their own work tasks, job security, or to the quality of services (Kreutzer and Jäger 2011; MacDuff 2011; Romanofsky 1973). This can manifest in poor volunteer-staff relationships, and be detrimental to the quality of the services produced if volunteers' roles are being limited by service professionals and/or if volunteers do not feel welcomed and accepted by the professional staff (Nesbit, Christensen, and Brudney 2018). The insights from this article are thus highly relevant for government, agencies, and public managers interested in designing volunteer involvement initiatives in ways that help mitigate threat perceptions among service professionals and lead to increased cost efficiency and/or improvements of service quality. The article proceeds as follows. First, we develop our theoretical model and propositions. Next, we outline our research design and report our empirical findings. Finally, we draw a conclusion, discuss our empirical results, and outline their implications for practice and future research. Theoretical Framework What Tasks Do Volunteers Solve in Public Organizations? As suggested by Nesbit and colleagues (2016), public and nonprofit organizations may use volunteers for different purposes—for example, helping service professionals with developing and designing services, and/or assisting service professionals with producing and delivering services. Drawing on a study by Parks and colleagues (1981) it can be argued that the collaboration between service professionals and volunteers on the production and delivery of services—which is the focus of our study—may either occur directly, involving coordinated efforts in the same production process, or occur indirectly through independent yet related efforts of service professionals and volunteers (Parks et al. 1981). An example of indirect volunteer involvement in service production is when volunteers go for a walk with an elderly living at a nursing home; an example of direct volunteer involvement is when a volunteer assists a teacher with practical tasks in the classroom. In order to further classify the type of tasks volunteers may contribute to in the service production phase, we draw on a recent study on coproduction by Brandsen and Honingh (2016) focusing on citizen involvement in the production and delivery of public services. According to Brandsen and Honingh (2016), it is important to be aware of the extent to which citizens are involved in core tasks, that is, tasks that are part of the organization's core services. In the area of education, for instance, a core task is teaching and educating students (Brandsen and Honingh 2016). An example of volunteer involvement in the production of a core task in schools is when volunteers are directly engaged in teaching and educating students as a teaching assistant in the classroom. On the other hand, an example of volunteer involvement in production of a complementary task is when volunteers help renovate the school's playground (Brandsen and Honingh 2016). Here, volunteers do not contribute to a core service of the school but rather help produce a service that indirectly benefits—or complements—students' experience of the physical environment of the school. It is, however, not always easy to classify whether volunteers are directly engaged in producing core tasks in the organization, and Brandsen and Honingh (2016) therefore suggest that classifications should be performed on a case-by-case basis. While multiple stakeholders such as policy makers, advocacy groups, leaders, and service users can have a legitimate say in defining what constitutes core or complementary tasks within a given service area, we follow the sociology of professions' focus on the member of a profession. Members of a profession share specialized, theoretical knowledge acquired from a formal educational program alongside a set of professional norms (Andersen and Pedersen 2012; Freidson 2001). Specialized refers to the fact that only members of the profession are in possession of the knowledge, while theoretical means that the knowledge base includes understandings of concepts and causal relationships relevant to the field (Andersen and Pedersen 2012, 47). Professional norms prescribe which actions are required, permitted, or prohibited in a specific situation (Andersen and Pedersen 2012, 48). Given that specialized knowledge is not immediately available to outsiders, information asymmetry makes it difficult at best for others to evaluate the appropriateness of a specific procedure, standard, or an outcome (Sharma 1997). This is augmented by the fact that service professionals often make decisions and work in environments characterized by substantial discretion (Lipsky 1980; Pedersen, Stritch, and Thuesen 2018). Members of a profession—service professionals—are therefore in a unique position to connect means to ends in an effort to deliver on core objectives of public organizations' missions. Consequently, we rely on service professionals to define what tasks are core and complementary to the organization and necessitate specialized, theoretical knowledge and professional norms for their successful execution. In our study, we use the distinction between core and complementary tasks since our central claim is that it is not the involvement of volunteers per se that trigger threat perceptions among service professionals, but rather the type of task volunteers solve in the organization. Before we develop this argument in more detail, we first construct a classification scheme to help outline different types of threat perceptions among service professionals. Professionals' Response to Volunteer Involvement: A Classification of Perceived Threat Although scholars have expressed concerns that volunteer involvement in production and delivery of services might be seen by professionals as threats to their professional standards, the privileged position of the profession, and professionals' own job (Alford and O'Flynn 2012; Nesbit et al. 2016), no conceptual framework has been advanced to help classify and differentiate different forms of threat perceptions. Looking outside of public administration research, insights from the sociology of professions and group conflict theory are useful for building such a classification scheme. Group conflict theory posits that people become more hostile toward an out-group if they perceive the group as a threat (McLaren 2003). Different types of threat can exist according to group conflict theory, but the concept of “realistic threat” is particularly relevant to our study. Realistic threats concern threats by an out-group to the very existence of an in-group, the power of the in-group, or the physical or material well-being of the in-group (Stephan et al. 1998). Threat perceptions arise due to competition over scarce resources such as jobs and power (McLaren 2003; Stephan et al. 1998) or due to concerns about the welfare of the in-group (Stephan et al. 1998). According to group conflict theory, perceptions of threat can occur at the individual level (e.g., an individual service professional) or at the group level (e.g., a profession). An example of a threat operating at the individual level is the risk of losing one's job, whereas a group-level threat could relate to concerns about increased unemployment rates (McLaren 2003). We adopt the distinction between perceived threats at the individual level and at the group level to guide our theoretical classification of different types of threat and their underlying motives. While the distinction between threats at an individual and group level is of theoretical importance, its empirical distinctiveness is specific to the context or the situation of study. This is, for example, the case if individuals do not perceive actions of an out-group as a threat to neither themselves nor to their group, or if they see the actions of an out-group as a threat to both themselves and their group. In these cases, the actions of an out-group will lead to identical behavioral implications among members of the profession such as resistance to the intrusion of the out-group into their domain. In addition to drawing on group conflict theory, we draw on two theoretical perspectives from the sociology of professions; the neo-Weberian perspective and the functionalist perspective. According to the neo-Weberian perspective, society is “an arena where competing groups struggle with each other and the state to gain power and status” (Roberts and Dietrich 1999, 987). Professions and their members are seen as inherently concerned with establishing and maintaining power and status (Johnson 1972; Parkin 1974). This is best ensured through “social closure” (Murphy 1988), a mechanism which—in this context—refers to members of a profession defending its privileged position and monopoly on providing certain services by excluding others from providing the same services. Social closure based on educational credentials may be a method to exclude individuals not belonging to the profession from providing certain services. The parallels to group conflict theory are straightforward. Volunteers as an out-group compete in the arena dominated by the service professionals by taking on tasks traditionally performed by members of the profession. For individual service professionals, this can cause concerns over loss of work tasks—and, ultimately, the risk of losing one's job. At the group level, members of a profession can see competition from volunteers as a cause of concern over potential loss of the profession's privileged position and monopoly on providing certain services from which the profession derives its power and status. In contrast, functionalists view service professionals as altruistic service providers concerned with “doing good for society” (Andersen and Pedersen 2012). Although service professionals, according to this perspective, also defend the monopoly of the profession on certain tasks, they do so for entirely different reasons. Service professionals, in this logic, seek to ensure monopoly on providing certain services out of a genuine concern for the quality of the services provided to beneficiaries (service users and clients). Only members of the profession possess the specialized, theoretical knowledge and have been socialized to the professional norms necessary for upholding high-quality services. In this light, competition from volunteers may cause concerns among individual members of the profession about the quality of services delivered to the service users. While service professionals in this perspective do not suffer pecuniary losses (e.g., material rewards), they incur psychological costs if they experience service levels fall short of their own expectations. This is, for example, illustrated in research on “public service motivation”, where individuals fueled by an altruistic desire to do good for other people and society through public service delivery have been shown to end up disillusioned, resigned, and burned out, if they experience a disconnect between their internalized desire for helping others and the opportunities their job offer to fulfill such desires (Giauque et al. 2012; Jensen, Andersen, and Holten 2019; van Loon, Vandenabeele, and Leisink 2015). Taken together, we can classify service professionals' perception of volunteers as a threat according to the types and levels outlined in figure 1. Threats may stem from egoistic concerns about losing one's own work tasks or even one's job. A similar concern operates at a group level if members of a profession are concerned about the profession losing its privileged position and monopoly on providing certain services to a competing out-group. These types of threats follow the motivational assumptions of the neo-Weberian perspective of individuals as self-interested actors. Threats can also derive from altruistic concerns about the quality of service delivered to the service users. This type of threat follows the motivational assumptions of the functionalist perspective of individuals as altruistic actors. Figure 1. Open in new tabDownload slide Classification of Perceived Threat. Figure 1. Open in new tabDownload slide Classification of Perceived Threat. Task Characteristic and Threat Perception Combining our classification of service professionals' perception of volunteers as threats and the distinction between core and complementary tasks in the two previous sections, we propose that the various types of threat perceptions are more likely to be triggered when volunteers solve core tasks compared to complementary tasks. Core tasks—in contrast to complementary ones—are at the heart of professions. Members of the same profession share understandings of concepts and causal relations specific to their field. This is grounded in specialized, theoretical knowledge acquired through a formal educational program (Freidson 2001). Core tasks necessitate the application of this knowledge base for their successful execution. In other words, solving core tasks are founded on and guided by a common base of theoretical knowledge and professional norms shared by members of the profession that are not readily accessible to outsiders. For instance, since elderly citizens with Alzheimer's or reduced chewing and swallowing ability may forget to chew and/or swallow their food properly, it is critical that people taking care of those elderly people have the theoretical knowledge and skills to prevent dangerous situations from occurring, or if unavoidable, to identify and resolve such emergencies. When volunteers solve tasks that require specialized, theoretical knowledge and professional norms, service professionals may be concerned for one or more reasons as outlined in our classification. Through the eyes of the functionalist perspective, service professionals are genuinely concerned about delivering high-quality service to end users (MacDonald 1995, 3). If volunteers with little or no insights into the specialized, theoretical knowledge and professional norms guiding behaviors relevant to certain domain solve core tasks, service professionals cannot guarantee that volunteers follow, for example, best practices and code of ethics. Rather, service professionals may be concerned that volunteers' work falls short of the expected level of service due to their lack of specialized, theoretical knowledge and familiarity with professional norms. In the case of complementary tasks, service professionals are less likely to view volunteers as a threat to the quality of services since the successful execution of the task is not directly linked to theoretical knowledge and professional norms upheld within the profession. For example, it does not require theoretical knowledge and professional norms acquired through a formal education for the successful execution of a complementary task such as assisting in entertaining elderly residents at nursing homes by reading books aloud or playing card games. In this case, we propose that service professionals are less likely to perceive volunteers as a threat to the quality of the services. This leads to our first theoretical proposition. Proposition 1. When volunteers solve core tasks—in contrast to complementary tasks—service professional will be more likely to perceive volunteers as a threat to the quality of services. A different reason for perceiving volunteers solving core tasks as a threat can be derived from perspectives emphasizing the egoistic preferences of service professionals. Through the eyes of the neo-Weberian perspective, members of a profession are socialized to occupational self-interest (Andersen and Pedersen 2012, 47), meaning that service professionals seek to establish and maintain the power and status of their group (Johnson 1972; Parkin 1974) by defending its privileged position and monopoly. In our case, one can see service professionals as the in-group protecting the profession's privileged position and monopoly on providing certain services against volunteers. Execution of core tasks rests on specialized, theoretical knowledge and familiarity with professional norms prescribing appropriate behaviors. If volunteers take over tasks identified as core by the service professionals, this directly questions the prerequisites for executing such tasks, and thereby directly challenging the profession's privileged position and monopoly on solving certain tasks from which the profession derives its power and distinguishes itself from other groups. Solving complementary tasks, on the other hand, are not tied as closely and directly to specialized, theoretical knowledge and professional norms. For instance, when volunteers solve a complementary task such as renovating a school playground it requires skills unrelated to teaching, and it is therefore unlikely that volunteers in such a context are perceived as a threat toward the teaching profession's privileged position and monopoly on providing educational services. Hence, when volunteers solve complementary tasks, we expect service professionals to be less likely to perceive volunteers as a threat to the position and monopoly of the profession. This leads to our second theoretical proposition. Proposition 2. When volunteers solve core tasks—in contrast to complementary tasks—service professionals will be more likely to perceive volunteers as a threat to the privileged position and monopoly of their profession. While perspectives from the sociology of professions focus on professions as collective entities, implications of self-interest among professionals also operate at an individual level. Group conflict theory highlights the relevance of these multiple levels most explicitly, arguing that relevant threats can also arise as a matter of protecting individual interests. A logic similar to the one outlined by the neo-Weberian perspective thus also applies to individual service professionals' protection of their own self-interest. If volunteers solve core tasks, individual professionals no longer enjoy a privileged position in which their function in the organization is guaranteed by their specialized, theoretical knowledge and adherence to certain professional norms. Complementary tasks, on the other hand, are not directly linked to the standards and procedures guiding the work of members of the profession. Rather, complementary tasks may take away time from core tasks and be perceived as red tape or unnecessary by the service professionals. For these reasons, we expect service professionals to be more likely to perceive volunteers as a threat to their own work tasks and job security when volunteers solve core tasks in contrast to complementary ones. However, Brudney and Gazley (2002) and Kreutzer and Jäger (2011) argue that paid staff feel more threatened by volunteers when they can perform the same functions as the service professionals. At first sight, this may suggest that service professionals should be more concerned about losing work tasks or even their job when volunteers solve complementary tasks. However, it is important to keep in mind that core tasks are tasks that are part of the organization's core services, and therefore are services that are required to be produced and delivered, which is not necessarily the case for complementary tasks. Consequently, we expect service professionals to be more concerned about their own work tasks and job security when volunteers solve core tasks compared to complementary tasks. This leads to our third proposition. Proposition 3. When volunteers solve core tasks—in contrast to complementary tasks—service professionals will be more likely to perceive volunteers as a threat to their own work tasks and job security. Research Design and Data A number of endogeneity problems may arise when examining how and when tasks performed by volunteers in public organizations influence service professionals' attitudes toward volunteers. Public organizations may, for example, be more likely to involve volunteers in complementary rather than core tasks depending on prior experiences with volunteer involvement. If this, or other unobserved variables, shape both the kinds of tasks volunteers are assigned to and professionals' attitudes, our results may be biased (Antonakis et al. 2010). Professionals' attitudes toward volunteers may also influence the kind of tasks volunteers are assigned to introducing potential problems with reverse causality. In an ideal situation, we would randomly assign volunteers to either complementary or core tasks across nursing homes, and then measure threat perceptions among nursing home professionals. Since this is virtually impossible in practice, we test our hypotheses using a survey experimental vignette design in which respondents were randomly assigned to different versions of a single vignette treatment (e.g., a vignette describing volunteers solving a core task or a complementary task, respectively). If differences in threat perceptions can be identified across experimental groups, this difference can be attributed to the vignette treatments and thus be interpreted as a causal effect. Offering researchers combinations of flexibility, low costs, and strong internal validity, survey experiments are increasingly used, not only in related disciplines such as psychology and political science, but by public administration researchers to study the attitudes of citizens, service professionals, and leaders (e.g., Andersen 2017; James and Van Ryzin 2017; Jilke and Tummers 2018; Olsen 2017). Despite these appealing features, survey experiments also come with several challenges. Concerns have been raised over the ecological validity of such studies questioning the mundane realism of fictitious vignette descriptions and whether observed effects can be generalized to real-life settings (Barabas and Jerit 2010). Before we outline how we constructed our vignettes to ensure high mundane realism, we first introduce the empirical case of our study and describe the data collection process. Empirical Case and Study Population Our study is conducted among service professionals employed at public nursing homes in Denmark. Around 85% of nursing homes in Denmark are public organizations, owned, financed, and regulated by local municipalities (Hjelmer et al. 2016; Rostgaard et al. 2016). We chose this case for three reasons. First, nursing homes in Denmark draw extensively on volunteers to help service professionals with various service tasks (FOA 2016) such as social gathering (e.g., drinking coffee and talking), walking and driving with residents, and accompanying residents to the hospital or doctor. In a survey from 2016, 80% of health assistants employed at nursing homes reported that volunteers solve various tasks at their workplace (FOA 2016).1 This number exceeds those of other service areas in Denmark that involve volunteers such as preschools, elementary schools, and hospitals (FOA 2016). Nine out of 10 volunteers at nursing homes are 60 years old or older, and most have already retired from the labor market (Aarhus Municipality 2016). Second, nursing home professionals in Denmark have experienced an increase in the number of tasks solved by volunteers over the past years (FOA 2016). Third, and as we show later, volunteers assisting at Danish nursing homes both solve tasks perceived as “core” and tasks perceived as “complementary” by nursing home professionals. Taken together, developments in and type and degree of volunteer involvement at Danish nursing homes all indicate that various types of threat perceptions may be highly salient among nursing home professionals. While nursing homes in Denmark employ members of various occupational groups, including nurses, pedagogues, and therapists, we delimit our focus to health assistants. Health assistants constitute by far the largest share of nursing home staff in Denmark (Ministry of Health 2016) and fit the criteria for a profession (Andersen and Pedersen 2012; Freidson 2001). Through the completion of a theoretical and practical training program, health assistants obtain a shared base of specialized, theoretical knowledge necessary for successfully solving care-related tasks (such as assisting elderly residents with Alzheimer's) and are socialized to a shared set of professional norms prescribing appropriate behavior for interacting with and caring for elderly people. Health assistants in Denmark typically have either a “short education” consisting of 19 months of theoretical and practical training or a “long education” consisting of 39 months of theoretical and practical training (Ministry of Health 2016). They share a common educational background in the sense that individuals first complete the short education and then can choose to complete 20 months of additional training. While health assistants with short and long education thus differ with respect to the degree of specialized, theoretical knowledge, they do share a common knowledge and occupational domain. Both groups have received theoretical and practical training in executing tasks related to personal help and care, whereas only health assistants with a long education have received training in administering and giving patients medicine (Ministry of Health 2016). Focusing on health assistants is particularly useful for a first-order test of our propositions since this group of nursing professionals—in a comparative light—have less formal and specialized training than nurses, pedagogues, and therapists. This implies that health assistants' professional domain, including a monopoly on solving certain care-related tasks, may be more vulnerable to members of out-groups (such as volunteers) taking over such tasks. Furthermore, the intragroup variation in degree of professionalism among health assistants with a short and a long education allows us to test whether threat perceptions differ between the two subgroups, and whether they respond differently to the involvement of volunteers in solving core and complementary tasks. Data Collection The data for our study were collected among members of a web panel from a major labor union, FOA, which organizes members who work in various types of public sector services in Denmark, including health assistants working at nursing homes. Existing research on our topic is scarce and we were not able to identify a pool of existing studies to help inform expectations about anticipated effect sizes. Instead, we obtained information about the number of respondents that we could reasonably expect based on FOA's prior surveys among their web panel, and used this information to calculate the minimum detectable effect size. Power calculations conducted prior to the data collection showed that we would be able to detect average treatment effects of an effect size of around 0.22 of a standard deviation (SD; alpha level = 0.05; power level = 0.80, R2 = 0.05) with approximately 900 subjects and three groups. If we further analyze treatment effect across subgroups, we are able to identify even smaller effect sizes. In February 2018, an electronic survey was distributed to various members of FOA's web panel. The response rate for the survey was 39.4%. Only respondents who confirmed that (1) they were health assistants employed at nursing homes, and (2) volunteers were doing volunteer work at their workplace were included in the survey experiment, which ensures that the respondents are able to relate to the vignette scenario. This reduces the total number of health assistants in the survey experiment from 1,260 to 971 respondents (corresponding to 77.1% of the health assistants being employed at a nursing home in which volunteers perform different tasks, which corresponds to number reported in surveys by FOA (FOA 2016)). Of these respondents, 78 had no observations on neither of the outcome measures (see section on “Outcome Measures: Threat Perception”) or on covariates from FOA's web panel, yielding an analytical sample of 893 respondents.2 About 38.9% of the respondents had a short education (i.e., 19 months), while 61.1% had the longer education (i.e., 39 months). Given that the health assistants were part of a web panel where members sign up to participate, we cannot for sure know whether our sample is representative of the entire population of health assistants working at nursing homes in Denmark. However, the web panel includes members that vary in terms of geography, age, gender, work experience, and number of working hours.3 The Survey Experimental Vignettes As part of our web survey, subjects were randomly assigned to one of three different experimental conditions. One condition included a vignette that described volunteers solving a task identified as a core task by the health assistants themselves. A second condition included a vignette describing volunteers solving a task identified as complementary by health assistants. Finally, the third condition included only a baseline introductory text in order to function as a baseline group. The baseline group reports the mean threat level among health assistants in the absence of any priming of task characteristic. This is useful information both in the context of our study as well as a means for contextualizing the practical implications of our results. If we observe a difference in threat perceptions among service professionals as a function of the type of task volunteers solve this is interesting and important. The substantial implications of such findings for practice, however, are difficult to assess without insight into the level of threat perception among service professionals. For example, if volunteers are hardly ever seen as a threat to the quality of the services, a treatment effect might invoke fewer concerns among government and public managers, and vice versa. To improve the mundane realism of the vignettes, we set up two requirements for our vignettes. The first requirement is that the core and complementary task is defined and identified by the service professionals as such. Only in this case can we expect the vignette treatments to influence service professionals' threat perceptions. We therefore conducted a pilot study in February–March 2017 to guide us on this issue. The pilot study included approximately 150 health assistants employed at nursing homes in four Danish municipalities. We asked the health assistants to categorize 14 different tasks as a core task for their profession or not. Ninety-one percent of the health assistants answered that “serving food and assist with eating” was a core task (see Thomsen and Jensen 2018), whereas health assistants were more split with respect to other tasks such as assisting elderly to the hospital or doctor and helping elderly with cleaning (Thomsen and Jensen 2018). We therefore decided that the core task vignette should describe a situation in which volunteers assist the elderly with eating. To ensure that any differences in threat perceptions were not caused by differences in the elderly's needs or abilities, the complementary task vignette also revolved around an eating situation. In this situation, however, the volunteers would be acting as a table host being responsible for chatting with the elderly and ensuring that they were having an enjoyable time. We choose this scenario as an example of a complementary task, since it, in contrast to assisting an elderly with consuming a meal, does not require specialized, theoretical knowledge and professional norms for its successful execution.4 The second requirement is that the core task vignette should include a task that volunteers perform in real life. In another survey conducted among health assistants in Aarhus Municipality, Denmark, we asked “Which of the following tasks do volunteers help with solving? [serving food and assist with eating].” Within the subgroup of respondents who had been in contact with volunteers within the last month, 14% confirmed that volunteers help with these tasks. In 63 cases—nursing homes—at least one employee confirmed that volunteers help with these particular tasks. The specific wording of the vignettes is outlined in figure 2. The bold text illustrates the difference between the vignette describing volunteers solving a core task and a complementary task, respectively. Figure 2. Open in new tabDownload slide Experimental Design: Complementary and Core Tasks in Nursing Home Contexts. Figure 2. Open in new tabDownload slide Experimental Design: Complementary and Core Tasks in Nursing Home Contexts. To further improve the mundane realism of the vignettes, we also conducted six semi-structured interviews with health assistants in December 2017. During the interviews, we received feedback on the content and wordings of the vignettes. First, we wanted to ensure that the description of the fictitious resident, Anna, resembles residents living at nursing homes. Second, we wanted to get additional insights on what tasks health assistants perceived as core and complementary. Several of the respondents expressed that since Anna and the other residents have problems with chewing and swallowing the food, it is important that the task is performed by service professionals with knowledge about how to feed elderly with chewing and swallowing problems, but also knowledge on how to respond in case Anna chokes on her food. The interviews thus supported that health assistants perceived assisting elderly residents with eating as a core task. Moreover, during the interviews many of the health assistants mentioned that they agreed that volunteers could solve tasks such as talking with elderly or acting as a table host. Finally, we received separate feedback on the vignettes from the labor union, FOA, to ensure that the text and words were clear and easy to read and understand for the respondents. As a result, we made some minor changes in the vignettes to increase the readability by shortening and simplifying sentences. Although we took a number of steps to improve the mundane realism of the vignettes there is at least one issue about the realism of the survey experiment that needs to be discussed. It was not specified in the vignettes whether service professionals were present in the two eating situations. However, we do not consider this to influence the service professionals' interpretation of the vignette and thereby the results, as it would be unrealistic to imagine an eating situation in which elderly residents gather for dinner at a nursing home in Denmark with no service professionals being present. The reason for this is that residents at Danish nursing homes are highly impaired and are not able to live in their own home. The residents are on average 84 years, 50% have one or more chronical diseases and two-thirds suffer from dementia (Ministry of Health 2016). An important precondition for the treatment effects to be interpreted as causal effects is that the randomization is successful in creating identical groups across the experimental conditions. In table 1, we present summary statistics across the three experimental groups. The table reveals very similar means, and the null hypothesis of no mean differences across the three experimental conditions cannot be rejected for any of the covariates. This bolsters our confidence that the randomization was successful in creating comparable groups at the outset. Table 1. Validation of Randomization: Balance Test of Means Across Experimental Groups Group 1: Baseline Group Group 2: Complementary Task Group 3: Core Task F-testa Gender (female “1”)b 0.97 0.97 0.94 n.s. Age (years)b 50.46 50.67 49.40 n.s. Education (health assistant, longer education “1”)b 0.60 0.62 0.61 n.s. Region (Zealand “1”)b 0.25 0.33 0.30 n.s. Employment hours (<37 hours/week “1”)c 0.53 0.57 0.52 n.s. Shift (day shift “1”)c 0.48 0.49 0.50 n.s. Work experience (<5 years “1”)c 0.32 0.38 0.39 n.s. Observations 286 283 324 Group 1: Baseline Group Group 2: Complementary Task Group 3: Core Task F-testa Gender (female “1”)b 0.97 0.97 0.94 n.s. Age (years)b 50.46 50.67 49.40 n.s. Education (health assistant, longer education “1”)b 0.60 0.62 0.61 n.s. Region (Zealand “1”)b 0.25 0.33 0.30 n.s. Employment hours (<37 hours/week “1”)c 0.53 0.57 0.52 n.s. Shift (day shift “1”)c 0.48 0.49 0.50 n.s. Work experience (<5 years “1”)c 0.32 0.38 0.39 n.s. Observations 286 283 324 Note: Means for each treatment condition reported alongside mean for baseline group. aF-test of treatment indicators being jointly zero. None of the p-values obtained in the F-test indicate statistically significant differences in means between groups by experimental condition (p < .10). bThese data were obtained through FOAs web panel and merged with the survey data after the data collection. cRespondents were asked about this information in the survey. A few respondents had either nonresponse or answered “don't know” on one of these items, and these are coded as “0.” Open in new tab Table 1. Validation of Randomization: Balance Test of Means Across Experimental Groups Group 1: Baseline Group Group 2: Complementary Task Group 3: Core Task F-testa Gender (female “1”)b 0.97 0.97 0.94 n.s. Age (years)b 50.46 50.67 49.40 n.s. Education (health assistant, longer education “1”)b 0.60 0.62 0.61 n.s. Region (Zealand “1”)b 0.25 0.33 0.30 n.s. Employment hours (<37 hours/week “1”)c 0.53 0.57 0.52 n.s. Shift (day shift “1”)c 0.48 0.49 0.50 n.s. Work experience (<5 years “1”)c 0.32 0.38 0.39 n.s. Observations 286 283 324 Group 1: Baseline Group Group 2: Complementary Task Group 3: Core Task F-testa Gender (female “1”)b 0.97 0.97 0.94 n.s. Age (years)b 50.46 50.67 49.40 n.s. Education (health assistant, longer education “1”)b 0.60 0.62 0.61 n.s. Region (Zealand “1”)b 0.25 0.33 0.30 n.s. Employment hours (<37 hours/week “1”)c 0.53 0.57 0.52 n.s. Shift (day shift “1”)c 0.48 0.49 0.50 n.s. Work experience (<5 years “1”)c 0.32 0.38 0.39 n.s. Observations 286 283 324 Note: Means for each treatment condition reported alongside mean for baseline group. aF-test of treatment indicators being jointly zero. None of the p-values obtained in the F-test indicate statistically significant differences in means between groups by experimental condition (p < .10). bThese data were obtained through FOAs web panel and merged with the survey data after the data collection. cRespondents were asked about this information in the survey. A few respondents had either nonresponse or answered “don't know” on one of these items, and these are coded as “0.” Open in new tab Outcome Measures: Threat Perception Following our theoretical classification of threat perception, we developed threat perception measures to capture (1) perceived threat that stems from altruistic concerns over the quality of service, and perceived threat that stems from egoistic concerns over (2) losing own work tasks or even job and (3) the privileged position and monopoly of one's profession. In a first iteration, we pretested our instrument in the February–March 2017 pilot survey. The pretest allowed us to obtain information on item response distribution. Furthermore, to obtain information on the interpretation of individual items, we asked health assistants during our December 2017 interviews to assess the clarity of each item. We also solicited suggestions for improvements to the readability and contextual realism among the interview participants. Based on the pilot survey and the qualitative interviews, we made some changes to the items, which we describe in detail below. All items measuring threat perception are measured using a 5-point Likert scale, ranging from “completely agree,” “agree,” “neither agree nor disagree,” “disagree,” to “completely disagree.” A “don't know” option was also available, and respondents who reported “don't know” or did not give a response to one or more items of an index measure were discarded from the analyses. Table A2 in the Appendix reports the results of an exploratory factor analysis for the four items designed to capture the perceived threat toward the quality of care, which is our first dependent variable. In our questionnaire, we systematically use the word “care” instead of “service,” since the interviews revealed that most respondents perceived the latter to be too broad and vague. The items are all phrased to capture different aspects of individuals' perception of the consequences of volunteer involvement for the quality of care delivered to the elderly residents. One of the four items is positively worded in contrast to the remaining three other items. It is not unusual that reversely coded items perform unsatisfactorily in factor analysis—as is the case here—but they still serve an important role in reducing risks of response set. Table A2 suggests a one-factor solution with three items and an eigenvalue of 2.01. Factor loadings are consistently high for the three items, ranging from 0.74 to 0.89. We exclude the reversely coded item due to its very low factor loading. The three-item instrument also demonstrates high internal consistency with a Cronbach's alpha score of 0.87, and we therefore create an unweighted index by simply adding the three items together and rescaling the index to range from 0 to 10, with 10 expressing the maximum perception of threat toward quality of care. Operationalizing our conceptualization of threat perception based on egoistic motives requires attention to both the individual and group (profession) level. The individual level includes threat to their own work tasks and job security. The group level includes threat to the profession's privileged position and monopoly on providing certain services. Despite our expectation of a two-dimensional solution with three items converging around threat perceptions operating at the individual level and threat perceptions operating at the group level, respectively, a factor analysis based on the February–March 2017 pilot survey revealed that it was not possible to make an empirical distinction between threats at the individual and group level. Based on the pilot survey and interviews, we removed three of the initial items, constructed one new item, and decided to include a total of four items. Two items were phrased to capture threat to their own work tasks and job security (“Volunteers are to an increasing extent taking over tasks that give me job satisfaction” and “The use of volunteers creates increased uncertainty about my own job situation”). The other two items followed the same sentiment but were operationalized at the group (profession) level in order to capture threat to the privileged position and monopoly of the profession (“I am concerned that an increasing use of volunteers will result in higher unemployment among my profession” and “I fear that volunteers in the future are going to take over work tasks from my profession”). We chose these two items for two reasons. First, our interviews showed that questions asking the respondent directly about concerns over the monopoly of the profession were difficult to understand and would be interpreted in very different ways.5 Second, concerns over loss of a profession's monopoly can manifest in concerns over high unemployment rates or loss of work tasks among members of the profession if members of the in-group (profession) are no longer able to exclude members of the out-group (volunteers) from taking over important work tasks. Table A3 in the Appendix reports the results of an exploratory factor analysis for these four items. Similar to the February–March 2017 pilot survey, it indicates a one-factor solution with factor loadings ranging between 0.66 and 0.90 and a single eigenvalue greater than 1 [2.84]. The four items also show high internal consistency with a Cronbach's alpha score of 0.90. The single-factor solution suggests that health assistants—in our study—do not differentiate between threats operating at the individual level and at the group level. Rather, they perceive threats at the individual and group level in concert, and for analytical purposes, we therefore present results on both the individual level (threat to their own work tasks and job security) and group level (threat to the position and monopoly of the profession) as well as collapse the four items into a single scale. Thus, we constructed three additional indices that were rescaled to range from 0 to 10, with 10 expressing the maximum perception of threat. To simplify our terminology, we reference to the aggregate measure consisting of all four items as threat toward professionals' “job” in the remainder of the article. Figure 3 presents descriptive statistics for our four dependent variables. An important premise for the usefulness of the indices as dependent variables is that the distributions are not highly skewed (ceiling or floor effects). The histogram for the quality of care shows that 16% of the respondents report a score of 0, whereas the rest of the respondents are normally distributed across the range of the scale (mean = 4.47, SD = 3.13). If we compare across educational level, the distributions are similar, which suggests that the items are not sensitive to health assistants' educational level. The histogram for threat toward the respondents' own work tasks and job security shows that 26% of the respondents report a score of 0. The rest of the respondents are normally distributed across the range of the scale. In contrast, the histogram for threat toward the profession's privileged position and monopoly reveals that 38% of the respondents report a score of 0, while the rest of the respondents are almost equally distributed across the range of the scale. The relative high number of respondents reporting a score of 0 in these two latter histograms can be interpreted in two ways. One interpretation may be a floor effect, that is, the four items are not capable of differentiating between health assistants with a low and very low threat perception. A second interpretation may be that around one-fourth to one-third of the health assistants are actually not afraid of losing their work tasks or job or the profession's privileged position and monopoly as a consequence of volunteer involvement. Although the two distributions are right skewed, the substantial variation displayed by both is worth noting (mean = 3.66, SD = 3.03 for “own work tasks and job security”; mean = 3.13, SD = 3.21 for “position and monopoly of profession”). If we compare the distribution across educational level, the distributions are again similar. Figure 3. Open in new tabDownload slide Distribution of Dependent Variables. Histograms with Normal Density Curves. Notes: Variable “Job” aggregates variables “Own work tasks and job security” and “Profession’s position and monopoly.” All variables range from 0 to 10, with 10 denoting the highest level of threat perception. Figure 3. Open in new tabDownload slide Distribution of Dependent Variables. Histograms with Normal Density Curves. Notes: Variable “Job” aggregates variables “Own work tasks and job security” and “Profession’s position and monopoly.” All variables range from 0 to 10, with 10 denoting the highest level of threat perception. Results In this section, we first report the general level of perceived threat among health assistants who were not subject to any task vignette (baseline group) and compare it to both of our treatment groups, and second proceed to test our theoretical propositions stating that service professionals are more likely to perceive volunteers as a threat to (1) the quality of service, (2) the privileged position and monopoly of their profession, and (3) their own work tasks and job security, when volunteers solve core tasks in contrast to complementary tasks. Focusing first on health assistants in the baseline group, we see moderate levels of threat perceptions. On a scale from 0 to 10, health assistants in the baseline group, on average, report a mean threat to the “quality of care” of 3.99; a mean threat to the “position and monopoly of the profession” of 2.99; and a mean threat to their “own work tasks and job security” of 3.47 (all exhibiting considerable variation, SD = 3.12, SD = 3.29, and SD = 3.07, respectively). Table A4 in the Appendix reveals that health assistants receiving the “complementary task” vignette or the “core task” vignette express greater concern about the “quality of care” than the baseline group (reporting a mean of 4.44 and 4.93, respectively). Both mean differences are statistically significant (p < .05 and p < .001, respectively). In contrast, mean differences between the baseline group and either of the treatment group are indistinguishable from zero for both perceived threat to the “position and monopoly of the profession,” and to “own work tasks and job security.” Turning our attention to our theoretical propositions, table A5 in the Appendix reports estimates of ordinary least squares (OLS) regressions with an indicator variable for treatment, that is, whether the subject was presented with the “complementary task” vignette (reference category) or the “core task” vignette. This coefficient expresses the average treatment effect of being primed to think of volunteers performing a complementary task as opposed to a core task. We also reestimate models 1, 3, and 5 using a set of covariates to increase precision of the estimated treatment effect. Thus, the tables and figures presented in the rest of this section only include health assistants, who receive a “complementary” or a “core task” vignette. Given the clear direction of our theoretical propositions analyses are performed as one-tailed tests, but we also reference the corresponding results of two-tailed tests. To illustrate the estimated average treatment effect, figure 4 plots group means for each of our three outcomes by treatment group. Differences between bars can be interpreted as average treatment effects (unstandardized regressions coefficients from table A5, model 1, 3, and 5). Lines with spikes represent 95% confidence intervals. An average treatment effect is statistically significant at the 0.05 level, if the confidence interval for the treatment group (i.e., line with spike) does not overlap with other bars. This is clearly the case for the index measuring threat to quality of care, but not for the other outcome measures. Below, we describe the results in relation to each of our propositions. Figure 4. Open in new tabDownload slide Core Versus Complementary Task Prime on Professionals’ Threat Perception. Notes: Estimations based on OLS regressions (cf. A5, models 1, 3, and 5). Difference between bars depicts estimated average treatment effect. Lines depict 95% confidence interval for one-tailed tests of statistical significance. Absolute range of threat perception indexes: 0–10, with 10 denoting perception of volunteers as a threat to (1) quality of care, and (2) profession’s position and monopoly, and (3) own work tasks and job security, to the greatest extent possible. Figure 4. Open in new tabDownload slide Core Versus Complementary Task Prime on Professionals’ Threat Perception. Notes: Estimations based on OLS regressions (cf. A5, models 1, 3, and 5). Difference between bars depicts estimated average treatment effect. Lines depict 95% confidence interval for one-tailed tests of statistical significance. Absolute range of threat perception indexes: 0–10, with 10 denoting perception of volunteers as a threat to (1) quality of care, and (2) profession’s position and monopoly, and (3) own work tasks and job security, to the greatest extent possible. Testing our first proposition, the left-hand side of figure 4 reports the estimated average treatment effect of the “core task” versus “complementary task” vignette on perceived threat to “quality of care.” Figure 4 shows that health assistants receiving the “core task” vignette express greater concern about the quality of care compared with health assistants receiving the “complementary task” vignette. This difference (β = 0.49) is statistically significant at the 0.05 level for a one-sided test and at the 0.1 level for a two-sided test, and corresponds to an effect of 0.16 of a SD. These results are robust to the inclusion of covariates outlined in table 1 (β = 0.51; p = .27 for one-sided test; p = .53 for two-sided test).6 This suggests that simply being primed to think of volunteers performing a core task as opposed to a complementary one prompts concerns over the quality of care among health assistants. This finding offers support for our first proposition that service professionals perceive volunteers as a greater threat to the quality of services when volunteers solve core tasks compared to complementary tasks. Turning our attention to the second and third theoretical proposition, the right-hand side of figure 4 reports the estimated average treatment effect of the “core task” versus “complementary task” vignette on perceived threat to “position and monopoly of the profession” and “own work tasks and job security,” respectively. Figure 4 reveals that group means for the two experimental groups do not differ in any statistically significant way in models without and with covariates.7 Similar results emerge if we comprise the four items into the index, which we defined threat to “job” in the “Outcome Measures: Threat Perception” section (results not shown, but can be requested from corresponding author). In sum, we find no empirical support—at least in a Danish context—for the theoretical propositions that service professionals are more likely to perceive volunteers as a threat to the privileged position and monopoly of the profession or to service professionals' own work tasks or job security when volunteers solve core tasks compared to complementary tasks.8 For exploratory purposes, we test whether treatment effects differ among service professionals with different levels of contact with volunteers (which may be a proxy for experience with volunteers) and with a short and a long education, respectively. Figure 5 reports the effect of the “core task” versus “complementary task” vignette on threat to “quality of care” conditional on the level of contact with volunteers. We distinguish between “frequent contact” (at least once a week) and “infrequent contact” (two times a month or less). According to contact theory, lack of contact between competing groups may cause people to resort to negative stereotypes (Allport 1979) leading to different levels of threat perceptions. Such negative stereotypes can be associated with perceptions of volunteers as incompetent and unable to solve organizational tasks. Drawing on contact theory, we might therefore expect the average treatment effect on quality of care, as reported in figure 4, to be stronger among health assistants who do not frequently interact with volunteers. This group of service professionals, in contrast to their peers who interact on a frequent basis with volunteers, might resort to negative stereotypes of volunteers as generally incompetent or unable to solve organizational tasks—and this relates particularly to those (core) tasks that require specialized, theoretical knowledge, and familiarity with professional norms for their successful execution. Figure 5. Open in new tabDownload slide Heterogeneous Treatment Effect of Core Versus Complementary Task Prime on Perception of Threat to Quality of Care Across Professionals’ Contact with Volunteers. Notes: Estimations based on OLS regressions (full estimation results can be obtained from corresponding author). Difference between bars depicts estimated average treatment effect. Lines depict 95% confidence interval for one-tailed tests of statistical significance. Absolute range of threat perception indexes: 0–10, with 10 denoting perception of volunteers as a threat to quality of care to the greatest extent possible. The indicator variable for contact frequency (“how often within the last month have you been in contact with volunteers at your workplace?”) is constructed by collapsing the following response categories; “everyday,” “3–4 times a week,” “1–2 times a week” = 1 “frequent contact”; “2 times a month,” “once a month,” “I haven’t been in contact with volunteers at my workplace within the last month” and “don’t know” = 0 “infrequent contact.” Figure 5. Open in new tabDownload slide Heterogeneous Treatment Effect of Core Versus Complementary Task Prime on Perception of Threat to Quality of Care Across Professionals’ Contact with Volunteers. Notes: Estimations based on OLS regressions (full estimation results can be obtained from corresponding author). Difference between bars depicts estimated average treatment effect. Lines depict 95% confidence interval for one-tailed tests of statistical significance. Absolute range of threat perception indexes: 0–10, with 10 denoting perception of volunteers as a threat to quality of care to the greatest extent possible. The indicator variable for contact frequency (“how often within the last month have you been in contact with volunteers at your workplace?”) is constructed by collapsing the following response categories; “everyday,” “3–4 times a week,” “1–2 times a week” = 1 “frequent contact”; “2 times a month,” “once a month,” “I haven’t been in contact with volunteers at my workplace within the last month” and “don’t know” = 0 “infrequent contact.” The left-hand side of figure 5 shows a significant treatment effect on perceived threat to quality of care for health assistants who report infrequent contact with volunteers. Thus, health assistants with infrequent contact with volunteers, who received the “core task” vignette in contrast to the “complementary task” vignette express much greater concern about the quality of care. This difference (β = 0.82) is statistically significant at the 0.05 level for both one-sided and two-sided tests, and corresponds to an effect size of 0.26 of a SD. For health assistants with frequent contact with volunteers there is no significant treatment effect. We do not find any evidence of heterogeneous treatment effects for our other outcome measures (results not shown, but can be obtained upon request to corresponding author). In sum, the results in figure 5 suggest that the average treatment effect on perceived threat to the quality of care displayed in figure 4 may—at least in part—be driven by health assistants with infrequent contact with volunteers. Figure 6 reports the effect of the “core task” versus “complementary task” vignette on threat to the “quality of care” conditional on health assistants' education. Theoretically, we might expect the average treatment effect, reported in figure 4, to be stronger among health assistants with a long education compared to a short education. Although health assistants with a short and a long education share a common educational background in the sense that both have completed the short education, health assistants with a long education have acquired more specialized, theoretical knowledge through the completion of an additional 20 months training. Consequently, the group of health assistants with a long education might hold firmer professional norms and more specialized, theoretical knowledge related to solving core service tasks compared to their peers with a shorter education, and thus be more skeptical of the quality of care that volunteers can bring to the production of (core) tasks that necessitate specialized, theoretical knowledge and familiarity with professional norms for their successful execution. Figure 6. Open in new tabDownload slide Heterogeneous Treatment Effect of Core Versus Complementary Task Prime on Perception of Threat to Quality of Care Across Professionals’ Education. Notes: Estimations based on OLS regressions (full estimation results can be obtained from corresponding author). Difference between bars depicts estimated average treatment effect. Lines depict 95% confidence interval for one-tailed tests of statistical significance. Absolute range of threat perception indexes: 0–10, with 10 denoting perception of volunteers as a threat to quality of care to the greatest extent possible. Figure 6. Open in new tabDownload slide Heterogeneous Treatment Effect of Core Versus Complementary Task Prime on Perception of Threat to Quality of Care Across Professionals’ Education. Notes: Estimations based on OLS regressions (full estimation results can be obtained from corresponding author). Difference between bars depicts estimated average treatment effect. Lines depict 95% confidence interval for one-tailed tests of statistical significance. Absolute range of threat perception indexes: 0–10, with 10 denoting perception of volunteers as a threat to quality of care to the greatest extent possible. The left hand-side of figure 6 shows a significant treatment effect on perceived threat to quality of care among health assistants with a long education. Thus, health assistants with a long education, who received the “core task” vignette in contrast to the “complementary task” vignette express greater concern about the quality of care (statistically significant at the 0.05 level for one-sided test and 0.10 level for two-sided test). There is no significant treatment effect when examining health assistants with a short education. We do not find any evidence of heterogeneous treatment effects on our other outcome measures (results not shown, but can be obtained upon request to corresponding author). Taken together, the results in figure 6 suggest that the average treatment effect on threat to the quality of care displayed in figure 4 may—at least to some extent—be driven by health assistants with a long education. In the next section, we discuss our main results in light of existing research, outline their implications for future research and practice, and revisit the main limitations of our design. Discussion and Conclusion This study takes a first step to understand how service professionals in public organizations respond to involvement of volunteers in public service production. Despite cautious warnings that service professionals may view volunteers as a threat to their professional standards, profession, or job (e.g., Alford and O'Flynn 2012; Nesbit et al. 2016), there is a lack of theoretical models and empirical evidence in public administration research to understand and assess the merits of these cautions. For example, there is a lack of knowledge about questions such as: what may cause service professionals to perceive volunteer involvement in service production as a threat to quality of services, the privileged position and monopoly of the profession, or their own work tasks or job security? And do service professionals view volunteers as a threat to only some of these aspects? These questions are critical to investigate since there is little reason to believe that service professionals will engage in proactive collaborative efforts with volunteers if they predominantly perceive volunteers, for example, as a threat to the service quality or their own work tasks and job security (Kreutzer and Jäger 2011; MacDuff 2011; Nesbit et al. 2016; Romanofsky 1973). In this study, we examine whether service professionals are more likely to perceive volunteers as a threat when volunteers solve a core task rather than a complementary task. Doing so, the study offers a number of contributions to the existing literature and research. The first contribution is theoretical in nature. Drawing on insights from group conflict theory and the sociology of professions, we first build a classification of threat perception. The classification is a useful heuristic for research as it allows scholars to disentangle perceptions of volunteers as threats to important yet different aspects of public service professionals' work. Volunteers may be perceived as a threat to the quality of the service, to the privileged position and monopoly of a profession, and to the work tasks or job security of individual service professionals. While subsets of the classification may be more or less relevant depending on the study context (e.g., service area or country), it provides a framework for scholars to explore and compare threat perceptions across studies. We combine the threat classification with a distinction between core and complementary tasks to propose that the type of task volunteers solve is a critical factor in shaping service professionals' perception of volunteers as threats. Specifically, we argue that service professionals are more likely to perceive volunteers as threats to (1) the quality of service, (2) the privileged position and monopoly of their profession, and (3) to professionals' own work tasks and job security when volunteers solve core tasks compared to complementary tasks. Consistent with our theoretical proposition, we find that health assistants are more likely to perceive volunteers as a threat to the quality of care, when health assistants are primed to think of volunteers as performing a task identified as core—in contrast to one identified as complementary—by the health assistants themselves. In contrast, and deviating from our expectation, we do not find any evidence in our data to support that health assistants perceive volunteers as a threat to the profession's privileged position and monopoly or their own work tasks and job security when they are primed to think of volunteers as performing a core task in contrast to a complementary one. We also present an explorative analysis investigating whether threat perceptions differ among service professionals with different levels of contact with volunteers, which may be a proxy for experience with volunteers. Our analysis suggests that our average treatment effect on threat toward the quality of care was driven by health assistants with infrequent contact, aligning well with assumptions outlined by contact theory (Allport 1979). Moreover, we tested whether threat perceptions differ across health assistants with a short and a long education. Our analysis reveals that the average treatment effect on threat to the quality of care—to some extent—was also driven by health assistants with a long education. The results from our study offer important contributions to our understanding of the factors that trigger threat perceptions among service professionals and help opening the black box for how service professionals come to resist volunteers in public organizations. But what might explain why we find service professionals to be concerned with quality of service, but not the profession's position and monopoly or their own work tasks and job security, when volunteers solve core tasks compared to complementary tasks in their organization? A possible explanation could be that many Danish nursing homes—due to low unemployment rates—experience difficulties in recruiting employees with a health assistant educational background (Local Government Denmark 2017). This structural condition favors high job security among this group of professionals and makes it unlikely that health assistants perceive outside factors, in general, and volunteer involvement in service production, in particular, as a threat to their own work tasks, job security, or the privileged position and monopoly of the profession. The second contribution of the article is that it responds to calls for advancing our theoretical understanding of professionalism in the public sector (Perry 2017). In his 2017 John Gaus Award Lecture, James L. Perry asks the question: “What if we took professionalism seriously?” and follows up with the assertion that “… public administration would be well served to pursue professionalism as a core theme because it is a central construct for understanding and acting in public administrative contexts” (Perry 2017, 93). Perry's call is substantiated by the fact that much of government is staffed with service professionals (Mosher 1968). While we are sympathetic to the sentiment, taking professionalism seriously requires us to zoom in on pieces of the puzzle and study these using sound conceptual frameworks and rigorous empirical methods. We are taking a first step toward building a theoretical explanation for how and when volunteers come to be seen as a threat to the quality of service, the privileged position and monopoly of the profession, and service professionals' own work tasks and job security. Doing so, we illustrate the multidisciplinary nature of the “professionalism” concept. Insights from different disciplines are required to understand service professionals' attitudes and behaviors, and we bridge insights from coproduction theory, the sociology of professions, and group conflict theory. The third contribution of the article is that it offers a first attempt at operationalizing threat perception among public service professionals. While we acknowledge the limitations and the distinctive context in which we created these measures, it is our hope that scholars will continue our efforts and that they can serve as inspiration for future inquiries into service professionals' reactions to volunteer involvement in service production. Any empirical assessment of service professionals' threat perception requires some validated approach or measures to capture such perceptions, and we believe a multi-item multidimensional measurement instrument that can capture the different types of threat outlined in our theoretical classification is an important aspiration for future scholarship in this venue. Limitations and Generalizability In light of our empirical findings, it is important to discuss the limitations of our design and the generalizability of our results. As a research design, we use a survey experimental vignette design describing volunteers solving a core or a complementary task. Although we made several attempts to improve the mundane realism of the vignettes, we cannot know for sure that health assistants respond exactly the same way in real-life situations. As often raised by scholars, one of the limitations of survey experiments is low ecological validity—that is, the findings are often difficult to generalize to real-life settings. However, based on our qualitative interviews and surveys by FOA (FOA 2016) we would not except our results to differ substantially from real-life situations, since many health assistants in these former studies, for example, express concerns about lack of coherence in services to elderly residents and volunteers taking over their work tasks. Another important issue to discuss is the generalizability of the results across service area, sector, and country context. In a Danish context, volunteers are more prevalent in nursing homes compared to other major welfare areas such as preschools, elementary schools, and hospitals (FOA 2016), and initially we would therefore expect to find effects of a smaller magnitude if the study was to be replicated in other service areas in Denmark. However, such effects also depend on how specialized services provided by service professionals are (Chum et al. 2013)—and thus how vulnerable service professionals are to volunteers taking over tasks perceived as core by the professionals. Service areas such as preschools in Denmark also employ professional staff with low levels of professionalism (e.g., pedagogical assistants) and we cannot therefore rule out that we would find similar results or even an effect on threat to their own work tasks or job security if we replicated the study at preschools in Denmark. In contrast, for services dominated by occupational groups characterized by higher levels of professionalism (e.g., nurses and doctors at hospitals), it is more difficult to see volunteers substituting for professional staff and performing core tasks, and we would generally be careful to expect similar findings in such areas. A common feature among many welfare services in Denmark is that organizations delivering such services are predominantly publicly owned, financed, and regulated. Does that mean that our propositions and results only apply to countries with extensive welfare systems or would we expect similar results if the services are provided in nonprofit or for-profit sectors? Public services such as elderly care, health care, and education can be provided by public, nonprofit, or for-profit organizations alike. For instance, in a study of physiotherapists working in private (for-profit) and public clinics in Denmark, Andersen and Jakobsen (2011) found little evidence that ownership matters for behaviors guided by professional norms. While the empirical effect on threat perceptions is sensitive to context, we do expect the theoretical propositions outlined in this article to apply to contexts in which professional providers of public services are employed by nonprofit or for-profit organizations. A third and final issue that is important to discuss is whether we would expect to find similar results in other country contexts? Countries differ with respect to traditions, norms, and scope of volunteer involvement in the production of public services. Although volunteers assist with producing elderly care at most nursing homes in Denmark (FOA 2016), volunteers mainly complement rather than substitute health assistants. This characterization finds empirical support in surveys asking health assistants to assess the kind of tasks that volunteers perform at their workplace (FOA 2016). Is this true across countries? A number of studies from North America have examined whether volunteer labor can be viewed as either substitutes or complements to paid work (Chum et al. 2013; Handy, Mook, and Quarter 2008; Simmons and Emanuele 2010; Stine 2008). One study from the United States shows that nonprofit organizations that use volunteer labor view them as substitutes to lower paid labor (Simmons and Emanuele 2010). Studies from Canada reveal that volunteers and paid labor at hospitals are substitutes to each other for certain tasks (Chum et al. 2013; Handy, Mook, and Quarter 2008). In contrast, another study from the United States on public libraries finds that the two inputs are complementary (Stine 2008). Thus, it is difficult to draw a clear conclusion regarding whether the Danish context is a least-likely or a most-likely case in a cross-country light. However, in service areas or country contexts in which volunteer labor mainly is viewed as substitutes to paid work, we cannot rule out that service professionals would be more inclined to view volunteers as a threat to the profession's position and monopoly or their own work tasks or job security than suggested by our study. Implications for Future Research and Practice The caveats of this study offer prospects for future research. In addition to exploring how service professionals behave and interact with volunteers when volunteers solve core tasks, scholars can benefit from assessing the relationship between threat perception and other important attitudes such as job satisfaction and work motivation. These have been linked to job performance across contexts (e.g., Cerasoli, Nicklin, and Ford 2014; Judge et al. 2001), and a deeper exploration of these various outcome measures could lead to a fuller understanding of the potential adverse consequences of volunteer involvement in service production for public organizations and public service provision. Other outcome measures that may be relevant to include in future research are service professionals' behavior and interactions with volunteers in public organizations. Following the line of reasoning in this article, one could expect service professionals to exert different types of negative behaviors such as obstructing volunteers' work, failing to coordinate proactively with volunteers, and ignoring volunteers' requests for help or guidance (Kreutzer and Jäger 2011; MacDuff 2011; Nesbit et al. 2016; Romanofsky 1973). Another fruitful venue for future research would be to examine additional moderators of the findings presented in this article. Do volunteers' qualifications, for instance, matter? One could expect service professionals to be less concerned with the quality of service if volunteers have an educational background as a health care worker (e.g., health assistant, nurse, or therapist) or have received training in solving tasks effectively and diligently. At the same time, this scenario might exacerbate perceptions of volunteers as threats to the privileged position of the profession and service professionals' own jobs—at least in contexts where volunteer involvement in service production is best described as substitutive and unemployment rates are high. This points to the importance of examining these relationships in other contexts as well. Threat perceptions may vary by service area, type of interaction with volunteers, and societal traditions and norms about volunteer involvement. We therefore urge researchers to not only replicate the study presented in this article, but to extend it across service areas and countries. Finally, our findings also have important implications for practice. Volunteers can provide important inputs to service production, but successful efforts require an active collaboration between volunteers and service professionals in public organizations. Our findings indicate that managers and policy makers should not assume that service professionals always view volunteers as a resource. Rather, volunteers can come to be seen as a threat to the quality of service; that is, the very thing that they were intended to help improve in the first place. We show that the type of task solved by volunteers is a crucial factor for managers to consider in relation to this paradox. Threat perceptions are triggered in particular when volunteers solve core tasks, and the main recommendation from this article is therefore to carefully consider task allocation between volunteers and service professionals. In particular, managers should be cautious to assign tasks to volunteers that service professionals perceive as core to their professional identity and profession. Acknowledgement An earlier version of this article was presented at the IRSPM2017, IRSPM2018, and EGPA2017 Conferences. We thank the participants at these conferences for their helpful comments. We also thank three anonymous reviewers, Martin Bækgaard, Morten Hjortskov, and Anne Mette Kjeldsen for helpful comments on previous versions of this article. Finally, we thank FOA for collaborating on the data collection. Footnotes 1 According to the survey by FOA (2016) 84%, 67%, 31%, and 26% of the health assistants report that volunteers assist with social gathering (e.g., drinking coffee and talking), walking and driving with residents, vigil and accompanying residents to the hospital or doctor, respectively. 2 Seven hundred and fifty-six respondents answered all seven items measuring threat perception (see section: “Outcome Measures: Threat Perception”); 89 respondents answered the four items measuring threat toward the profession's position and monopoly and their own work tasks and job security, but have either nonresponse or answered “don't know” on a least one item measuring threat toward the quality of care; 49 respondents answered all items measuring threat toward the quality of care, but have either nonresponse or answered “don't know” on a least one of the four items measuring threat toward the profession's position and monopoly and their own work tasks and job security. For one respondent information from FOAs web panel on age, region, and gender was not available. As a result, the analytical sample size is 893. 3 We are not able to compare our sample with the entire population of health assistants working at nursing homes in Denmark on, for example, age, gender, and education, since no official statistics exist on the background characteristics of health assistants employed at nursing homes. However, when we compare our analytical sample with the population of health assistants working at nursing homes who are a member of FOA (see table A1 in the Appendix), our sample is representative on gender and age, whereas the sample is not entirely representative on region (underrepresentation of members from Zealand). 4 Assisting elderly with Alzheimer or reduced chewing and swallowing ability with eating, for example, requires knowledge about cost habits and best practices and standards for how to feed people with disabilities. 5 Examples of questions that were removed: “Involving volunteers challenges my profession's professional standards” and “Involving volunteers breaks my profession's monopoly to solve certain work tasks.” 6 Most covariates are statistically insignificant with the exception of the following variables: age (older respondents are more concerned), region (Zealand residents are more concerned), and day shift (respondents working day shift are less concerned). In an exploratory analysis, we also examined whether contact with volunteers (proxy for experience) and workload is related to threat toward the quality of care. Both variables were statistically significant (respondents with frequently contact [at least once a week] and low levels of workload are less concerned). These results are not shown, but can be obtained upon request to corresponding author. 7 When we apply threat to “profession's position and monopoly” as dependent variable most covariates are statistically insignificant with the exception of age (older respondents are more concerned) and region (Zealand residents are more concerned). In an exploratory analysis, we also examined whether contact with volunteers (proxy for experience) and workload is related to threat to the profession's position and monopoly. Only workload was statistically significant (respondents with low levels of workload are less concerned). These results are not shown, but can be obtained upon request to corresponding author. When we apply threat to “own work tasks and job security” as dependent variable most covariates are statistically insignificant with the exception of age (older respondents are more concerned) and day shift (respondents working day shift are less concerned). In an exploratory analysis, we also examined whether contact with volunteers (proxy for experience) and workload is related to threat to the profession's work task and job security. Both variables were statistically significant (respondents with low levels of workload are less concerned and respondents with frequently contact [at least once a week] are more concerned). These results are not shown, but can be obtained upon request to corresponding author. 8 As a robustness check, we also examine treatment effects on threat perception based on (1) risk of losing work tasks (an index compromising items 1 and 4 in table A3) and (2) risk of losing one's job altogether (an index compromising items 2 and 3 in table A3) as dependent variables. We do not find any significant treatment effect on these two indices. References Aarhus Municipality . 2016 . Frivillighedsundersøgelsen 2016 . Aarhus Kommune, Denmark : Sundhedsstaben – Sundhed og Omsorg . Google Preview WorldCat COPAC Alford , John , and Janine O'Flynn . 2012 . Rethinking public service delivery. Managing with external providers . Houndmills, UK : Palgrave Macmillan . Google Preview WorldCat COPAC Allport , Gordon W . 1979 . The nature of prejudice . Cambridge, MA : Perseus Books . Google Preview WorldCat COPAC Andersen , Simon C . 2017 . From passive to active representation – Experimental evidence on the role of normative values in shaping white and minority bureaucrats' policy attitudes . Journal of Public Administration Research and Theory 27 : 400 – 14 . Google Scholar Crossref Search ADS WorldCat Andersen , Lotte B. , and Mads L. Jakobsen . 2011 . Does ownership matter for the provision of professionalized services? Hip operations at publicly and privately owned clinics in Denmark . Public Administration 89 : 956 – 74 . Google Scholar Crossref Search ADS PubMed WorldCat Andersen , Lotte B. , and Lene H. Pedersen . 2012 . Public service motivation and professionalism . International Journal of Public Administration 35 : 46 – 57 . Google Scholar Crossref Search ADS WorldCat Antonakis , John , Samuel Bendahan , Philippe Jacquart , and Rafael Lalive . 2010 . On making causal claims: A review and recommendations . The Leadership Quarterly 21 : 1086 – 120 . Google Scholar Crossref Search ADS WorldCat Barabas , Jason , and Jennifer Jerit . 2010 . Are survey experiments externally valid? American Political Science Review 104 : 226 – 42 . Google Scholar Crossref Search ADS WorldCat Brandsen , Taco , and Marlies Honingh . 2016 . Distinguishing different types of coproduction: A conceptual analysis based on the classical definitions . Public Administration Review 76 : 427 – 35 . Google Scholar Crossref Search ADS WorldCat Brudney , Jeffrey L . 1993 . Volunteer involvement in the delivery of public services: Advantages and disadvantages . Public Productivity & Management Review 16 : 283 – 97 . Google Scholar Crossref Search ADS WorldCat Brudney , Jeffrey L. , and Beth Gazley . 2002 . Testing the conventional wisdom regarding volunteer programs: A longitudinal analysis of the Service Corps of Retired Executives and U.S. Small Business Administration . Nonprofit and Voluntary Sector Quarterly 31 : 525 – 48 . Google Scholar Crossref Search ADS WorldCat Cerasoli , Christopher P. , Jessica M. Nicklin , and Michael T. Ford . 2014 . Intrinsic motivation and extrinsic incentives jointly predict performance: A 40-year meta-analysis . Psychological Bulletin 140 : 980 – 1008 . Google Scholar Crossref Search ADS PubMed WorldCat Chum , Antony , Laurie Mook , Femida Handy , Daniel Schugurensky , and Jack Quarter . 2013 . Degree and direction of paid employee/volunteer interchange in nonprofit organizations . Nonprofit Management & Leadership 23 : 409 – 26 . Google Scholar Crossref Search ADS WorldCat FOA . 2016 . Frivilligt arbejde på FOA arbejdspladser . Copenhagen, Denmark : FOA Kampagne og Analyse . Google Preview WorldCat COPAC Freidson , Eliot . 2001 . Professionalism. The third logic . Chicago, IL : Univ. of Chicago Press . Google Preview WorldCat COPAC Gazley , Beth , and Jeffrey Brudney . 2005 . Volunteer involvement in local government after September 11: The continuing question of capacity . Public Administration Review 65 : 131 – 42 . Google Scholar Crossref Search ADS WorldCat Giauque , David , Adrian Ritz , Frédéric Varone , and Simon Anderfuhren-Biget . 2012 . Resigned but satisfied: The negative impact of public service motivation and red tape on work satisfaction . Public Administration 90 : 175 – 83 . Google Scholar Crossref Search ADS WorldCat Handy , Femida , Laurie Mook , and Jack Quarter . 2008 . The interchangeability of paid staff and volunteers in nonprofit organizations . Nonprofit and Voluntary Sector Quarterly 37 : 76 – 92 . Google Scholar Crossref Search ADS WorldCat Hjelmer , Ulf , Yosef Bhatti , Tina Rostgaard , Ole H. Petersen , Karsten V. Vrangbæk , Patricia T. Larsen , and Lily M. M. Jacobsen . 2016 . Kvalitet på offentlige og private plejecentre i Danmark . Copenhagen, Denmark : KORA . Google Preview WorldCat COPAC James , Oliver , and Gregg G. Van Ryzin . 2017 . Motivated reasoning about public performance: An experimental study of how citizens judge the Affordable Care Act . Journal of Public Administration Research and Theory 27 : 197 – 209 . Google Scholar Crossref Search ADS WorldCat Jensen , Ulrich T. , Lotte B. Andersen , and Ann-Louise Holten . 2019 . Explaining a dark side: Public service motivation, presenteeism, and absenteeism . Review of Public Personnel Administration 39 : 487 – 510 . Google Scholar Crossref Search ADS WorldCat Jilke , Sebastian , and Lars Tummers . 2018 . Which clients are deserving of help? A theoretical model and experimental test . Journal of Public Administration Research and Theory 28 : 226 – 38 . Google Scholar Crossref Search ADS WorldCat Johnson , Terence J . 1972 . Professions and power . London & Basingstoke : The Macmillan Press Ltd . Google Preview WorldCat COPAC Judge , Timothy A. , Carl J. Thoresen , Joyce E. Bono , and Gregory K. Patton . 2001 . The job satisfaction-job performance relationship: A qualitative and quantitative review . Psychological Bulletin 127 : 376 – 407 . Google Scholar Crossref Search ADS PubMed WorldCat Kreutzer , Karin , and Urs Jäger . 2011 . Volunteering versus managerialism: Conflict over organizational identity in voluntary associations . Nonprofit and Voluntary Sector Quarterly 40 : 634 – 61 . Google Scholar Crossref Search ADS WorldCat Lipsky , Michael . 1980 . Street-level bureaucracy: Dilemmas of the individual in public services . New York : Russell Sage Foundation Publications . Google Preview WorldCat COPAC Local Government Denmark . 2017 . Arbejdsudbud og rekruttering i kommunerne i 2017 . Copenhagen, Denmark : Kommunernes Landsforening . Google Preview WorldCat COPAC van Loon , Nina M. , Wouter Vandenabeele , and Peter Leisink . 2015 . On the bright and dark side of public service motivation: The relationship between PSM and employee wellbeing . Public Money & Management 35 : 349 – 56 . Google Scholar Crossref Search ADS WorldCat MacDonald , Keith M . 1995 . The sociology of the professions . London : Sage Publications . Google Preview WorldCat COPAC MacDuff , Nancy . 2011 . Volunteer and staff relations. In The volunteer management handbook , ed. Tracy D. Connors , 255 – 71 . New York : John Wiley . Google Preview WorldCat COPAC McLaren , Lauren M . 2003 . Anti-immigrant prejudice in Europe: Contact, threat perception, and preferences for the exclusion of migrants . Social Forces 81 : 909 – 36 . Google Scholar Crossref Search ADS WorldCat Ministry of Health . 2016 . National undersøgelse af forholdene på plejecentre . Copenhagen, Denmark : Sundheds- og Ældreministeriet . Google Preview WorldCat COPAC Mosher , Frederick C . 1968 . Democracy and the public service . New York : Oxford Univ. Press . Google Preview WorldCat COPAC Murphy , Raymond . 1988 . Social closure — The theory of monopolization and exclusion . Oxford, UK : Clarendon . Google Preview WorldCat COPAC Nesbit , Rebecca , Robert K. Christensen , and Jeffrey L. Brudney . 2018 . The limits and possibilities of volunteering: A framework for explaining the scope of volunteer involvement in public and nonprofit organizations . Public Administration Review 78 : 502 – 13 . Google Scholar Crossref Search ADS WorldCat Nesbit , Rebecca , Heather Rimes , Robert K. Christensen , and Jeffrey L. Brudney . 2016 . Inadvertent volunteer managers: Exploring perceptions of volunteer managers' and volunteers' roles in the public workplace . Review of Public Personnel Administration 36 : 164 – 87 . Google Scholar Crossref Search ADS WorldCat Olsen , Asmus L . 2017 . Compared to what? How social and historical reference points affect citizens' performance evaluations . Journal of Public Administration Research and Theory 27 : 562 – 80 . Google Scholar Crossref Search ADS WorldCat Parkin , Frank . 1974 . Strategies of social closure in class formation. In The social analysis of class structure , ed. Frank Parkin , 1 – 19 . London : Tavistock Publications . Google Preview WorldCat COPAC Parks , Roger B. , Paula C. Baker , Larry Kiser , Ronald Oakerson , Elinor Ostrom , Vincent Ostrom , Stephen L. Percy , Martha B. Vandivort , Gordon P. Whitaker , and Rick Wilson . 1981 . Consumers as coproducers of public services: Some economic and institutional considerations . Policy Studies Journal 9 : 1001 – 11 . Google Scholar Crossref Search ADS WorldCat Pedersen , Mogens J. , Justin M. Stritch , and Frederik Thuesen . 2018 . Punishment on the frontlines of public service delivery: Client ethnicity and caseworker sanctioning decisions in a Scandinavian welfare state . Journal of Public Administration Research and Theory 28 : 339 – 54 . Google Scholar Crossref Search ADS WorldCat Perry , James L . 2017 . The 2017 John Gaus Award Lecture: What if we took professionalism seriously? PS: Political Science & Politics 51 : 93 – 102 . Google Scholar Crossref Search ADS WorldCat Rimes , Heather , Rebecca Nesbit , Robert K. Christensen , and Jeffrey L. Brudney . 2017 . Exploring the dynamics of volunteer and staff interactions . Nonprofit and Management Leadership 28 : 195 – 213 . Google Scholar Crossref Search ADS WorldCat Roberts , Jennifer , and Michael Dietrich . 1999 . Conceptualizing professionalism: Why economics needs sociology . The American Journal of Economics and Sociology 58 : 977 – 98 . Google Scholar Crossref Search ADS WorldCat Rogelberg , Steven G. , Joseph A. Allen , James M. Conway , Adrian Goh , Lamarra Currie , and Betsy McFarland . 2010 . Employee experiences with volunteers, assessment, description, antecedents, and outcomes . Nonprofit Management & Leadership 20 : 423 – 43 . Google Scholar Crossref Search ADS WorldCat Romanofsky , Peter . 1973 . Professionals versus volunteers: A case study of adoption workers in the 1920's . Nonprofit and Voluntary Sector Quarterly 2 : 95 – 101 . WorldCat Rostgaard , Tine , Ole H. Petersen , Ulf Hjelmar , and Yosef Bhatti . 2016 . Vi har sammenlignet offentlige og private plejecentre – her er de små forskelle . https://www.denoffentlige.dk/vi-har-sammenlignet-offentlige-og-private-plejecentre-her-er-de-smaa-forskelle (accessed September 26, 2019) . Google Preview WorldCat COPAC Sharma , Anurag . 1997 . Professional as agent: Knowledge asymmetry in agency exchange . The Academy of Management Review 22 : 758 – 98 . Google Scholar Crossref Search ADS WorldCat Simmons , Walter O. , and Rosemarie Emanuale . 2010 . Are volunteers substitute for paid labor in nonprofit organizations? Journal of Economics and Business 62 : 65 – 77 . Google Scholar Crossref Search ADS WorldCat Stephan , Walter G. , Oscar Ybarra , Carmen M. Martinez , Joseph Schwarzwald , and Michal Tur-Kaspa . 1998 . Prejudice toward immigrants to Spain and Israel. An integrated threat theory analysis . Journal of Cross-Cultural Psychology 29 : 559 – 76 . Google Scholar Crossref Search ADS WorldCat Stine , William F . 2008 . An empirical analysis of the effect of volunteer labor on public library employment . Managerial and Decision Economics 29 : 525 – 38 . Google Scholar Crossref Search ADS WorldCat Thomsen , Mette K. , and Ulrich T. Jensen . 2018 . Fagprofessionelles klassifikation af kerneopgaver og komplementære opgaver . Politica 50 : 345 – 63 . WorldCat Wandersman , Abraham , and Jeanette Alderman . 1993 . Incentives, costs, and barriers for volunteers: A staff perspective on volunteers in one state . Review of Public Personnel Administration 13 : 67 – 76 . Google Scholar Crossref Search ADS WorldCat Appendix Table A1. Comparison of Analytical Sample and FOA Members Health Assistants (Analytical Sample) Health Assistants (FOA Members) Gender (female) 96% 95% Age  15–19 years 0% 0%  20–29 years 3% 8%  30–39 years 16% 16%  40–49 years 20% 23%  50–59 years 42% 36%  60–years 19% 16% Region (Zealand) 29% 41% Observations 893 27,375 Health Assistants (Analytical Sample) Health Assistants (FOA Members) Gender (female) 96% 95% Age  15–19 years 0% 0%  20–29 years 3% 8%  30–39 years 16% 16%  40–49 years 20% 23%  50–59 years 42% 36%  60–years 19% 16% Region (Zealand) 29% 41% Observations 893 27,375 Note: The information in the second column was kindly provided by FOA. Open in new tab Table A1. Comparison of Analytical Sample and FOA Members Health Assistants (Analytical Sample) Health Assistants (FOA Members) Gender (female) 96% 95% Age  15–19 years 0% 0%  20–29 years 3% 8%  30–39 years 16% 16%  40–49 years 20% 23%  50–59 years 42% 36%  60–years 19% 16% Region (Zealand) 29% 41% Observations 893 27,375 Health Assistants (Analytical Sample) Health Assistants (FOA Members) Gender (female) 96% 95% Age  15–19 years 0% 0%  20–29 years 3% 8%  30–39 years 16% 16%  40–49 years 20% 23%  50–59 years 42% 36%  60–years 19% 16% Region (Zealand) 29% 41% Observations 893 27,375 Note: The information in the second column was kindly provided by FOA. Open in new tab Table A2. Factor Analysis: Items for Threat Perception, “Quality of Care” (1) Factor 1 I fear that citizens in the future will receive worse care because their needs increasingly will be covered by volunteers 0.74 The use of volunteers makes it easier to provide good care to citizens (reversed) 0.07 I am concerned that the use of volunteers means that citizens do not receive the care they are entitled to 0.89 The use of volunteers makes it difficult to provide citizens with a coherent care 0.81 Observations 791 (1) Factor 1 I fear that citizens in the future will receive worse care because their needs increasingly will be covered by volunteers 0.74 The use of volunteers makes it easier to provide good care to citizens (reversed) 0.07 I am concerned that the use of volunteers means that citizens do not receive the care they are entitled to 0.89 The use of volunteers makes it difficult to provide citizens with a coherent care 0.81 Observations 791 Note: Principal factor analysis; no rotation. Eigenvalue = 2.01; only one factor with eigenvalue > 1. Cronbach's alpha = 0.87 for the three items with loading > 0.70. Open in new tab Table A2. Factor Analysis: Items for Threat Perception, “Quality of Care” (1) Factor 1 I fear that citizens in the future will receive worse care because their needs increasingly will be covered by volunteers 0.74 The use of volunteers makes it easier to provide good care to citizens (reversed) 0.07 I am concerned that the use of volunteers means that citizens do not receive the care they are entitled to 0.89 The use of volunteers makes it difficult to provide citizens with a coherent care 0.81 Observations 791 (1) Factor 1 I fear that citizens in the future will receive worse care because their needs increasingly will be covered by volunteers 0.74 The use of volunteers makes it easier to provide good care to citizens (reversed) 0.07 I am concerned that the use of volunteers means that citizens do not receive the care they are entitled to 0.89 The use of volunteers makes it difficult to provide citizens with a coherent care 0.81 Observations 791 Note: Principal factor analysis; no rotation. Eigenvalue = 2.01; only one factor with eigenvalue > 1. Cronbach's alpha = 0.87 for the three items with loading > 0.70. Open in new tab Table A3. Factor Analysis: Items for Threat Perception, “Job” (1) Factor 1 Volunteers are to an increasing extent taking over work tasks that give me job satisfaction 0.66 I am concerned that an increasing use of volunteers will result in higher unemployment among my profession 0.89 The use of volunteers creates increased uncertainty about my own job situation 0.90 I fear that volunteers in the future are going to take over work tasks from my profession 0.90 Observations 844 (1) Factor 1 Volunteers are to an increasing extent taking over work tasks that give me job satisfaction 0.66 I am concerned that an increasing use of volunteers will result in higher unemployment among my profession 0.89 The use of volunteers creates increased uncertainty about my own job situation 0.90 I fear that volunteers in the future are going to take over work tasks from my profession 0.90 Observations 844 Note: Principal factor analysis; no rotation. Eigenvalue = 2.84; only one factor with eigenvalue > 1. Cronbach's alpha = 0.90 for the four items. Open in new tab Table A3. Factor Analysis: Items for Threat Perception, “Job” (1) Factor 1 Volunteers are to an increasing extent taking over work tasks that give me job satisfaction 0.66 I am concerned that an increasing use of volunteers will result in higher unemployment among my profession 0.89 The use of volunteers creates increased uncertainty about my own job situation 0.90 I fear that volunteers in the future are going to take over work tasks from my profession 0.90 Observations 844 (1) Factor 1 Volunteers are to an increasing extent taking over work tasks that give me job satisfaction 0.66 I am concerned that an increasing use of volunteers will result in higher unemployment among my profession 0.89 The use of volunteers creates increased uncertainty about my own job situation 0.90 I fear that volunteers in the future are going to take over work tasks from my profession 0.90 Observations 844 Note: Principal factor analysis; no rotation. Eigenvalue = 2.84; only one factor with eigenvalue > 1. Cronbach's alpha = 0.90 for the four items. Open in new tab Table A4. Survey Experiment: Threat Perception as a Function of Task Type (Baseline Group and Treatment Groups). OLS Estimation DV: Quality of Care DV: Profession's Position and Monopoly DV: Own Work Tasks and Job Security (1) (2) (3) (4) (5) (6) Ref.: Baseline group  Complementary task vignette 0.458* (.047) 0.416+ (.062) 0.211 (.223) 0.170 (.267) 0.225 (.194) 0.194 (.227)  Core task vignette 0.948*** (.000) 0.896*** (.000) 0.198 (.230) 0.186 (.242) 0.314 (.107) 0.315 (.106) Constant 3.987*** (.000) 3.893*** (.000) 2.989*** (.000) 2.527** (.002) 3.474*** (.000) 2.620** (.001) Observations 804 804 844 844 844 844 Adjusted R2 0.013 0.039 −0.002 0.030 −0.000 0.014 Covariates NO YES NO YES NO YES DV: Quality of Care DV: Profession's Position and Monopoly DV: Own Work Tasks and Job Security (1) (2) (3) (4) (5) (6) Ref.: Baseline group  Complementary task vignette 0.458* (.047) 0.416+ (.062) 0.211 (.223) 0.170 (.267) 0.225 (.194) 0.194 (.227)  Core task vignette 0.948*** (.000) 0.896*** (.000) 0.198 (.230) 0.186 (.242) 0.314 (.107) 0.315 (.106) Constant 3.987*** (.000) 3.893*** (.000) 2.989*** (.000) 2.527** (.002) 3.474*** (.000) 2.620** (.001) Observations 804 804 844 844 844 844 Adjusted R2 0.013 0.039 −0.002 0.030 −0.000 0.014 Covariates NO YES NO YES NO YES Note: OLS regressions with p-values in parentheses. One-tailed test for statistical significance. Dependent variables are measured as additive measures ranging from 0 to 10. Covariates: variables listed in Table 1. ***p < .001, **p < .01, *p < .05, +p < .10. Open in new tab Table A4. Survey Experiment: Threat Perception as a Function of Task Type (Baseline Group and Treatment Groups). OLS Estimation DV: Quality of Care DV: Profession's Position and Monopoly DV: Own Work Tasks and Job Security (1) (2) (3) (4) (5) (6) Ref.: Baseline group  Complementary task vignette 0.458* (.047) 0.416+ (.062) 0.211 (.223) 0.170 (.267) 0.225 (.194) 0.194 (.227)  Core task vignette 0.948*** (.000) 0.896*** (.000) 0.198 (.230) 0.186 (.242) 0.314 (.107) 0.315 (.106) Constant 3.987*** (.000) 3.893*** (.000) 2.989*** (.000) 2.527** (.002) 3.474*** (.000) 2.620** (.001) Observations 804 804 844 844 844 844 Adjusted R2 0.013 0.039 −0.002 0.030 −0.000 0.014 Covariates NO YES NO YES NO YES DV: Quality of Care DV: Profession's Position and Monopoly DV: Own Work Tasks and Job Security (1) (2) (3) (4) (5) (6) Ref.: Baseline group  Complementary task vignette 0.458* (.047) 0.416+ (.062) 0.211 (.223) 0.170 (.267) 0.225 (.194) 0.194 (.227)  Core task vignette 0.948*** (.000) 0.896*** (.000) 0.198 (.230) 0.186 (.242) 0.314 (.107) 0.315 (.106) Constant 3.987*** (.000) 3.893*** (.000) 2.989*** (.000) 2.527** (.002) 3.474*** (.000) 2.620** (.001) Observations 804 804 844 844 844 844 Adjusted R2 0.013 0.039 −0.002 0.030 −0.000 0.014 Covariates NO YES NO YES NO YES Note: OLS regressions with p-values in parentheses. One-tailed test for statistical significance. Dependent variables are measured as additive measures ranging from 0 to 10. Covariates: variables listed in Table 1. ***p < .001, **p < .01, *p < .05, +p < .10. Open in new tab Table A5. Survey Experiment: Threat Perception as a Function of Task Type (Treatment Groups). OLS Estimation DV: Quality of Care DV: Profession's Position and Monopoly DV: Own Work Tasks and Job Security (1) (2) (3) (4) (5) (6) Ref.: Complementary task vignette  Core task vignette 0.491* (.033) 0.510* (.027) −0.013 (.481) 0.072 (.393) 0.089 (.363) 0.169 (.253) Constant 4.444*** (.000) 3.411** (.001) 3.200*** (.000) 1.219 (.125) 3.698*** (.000) 1.664* (.049) Observations 549 549 568 568 568 568 Adjusted R2 0.004 0.032 −0.002 0.027 −0.002 0.013 Covariates NO YES NO YES NO YES DV: Quality of Care DV: Profession's Position and Monopoly DV: Own Work Tasks and Job Security (1) (2) (3) (4) (5) (6) Ref.: Complementary task vignette  Core task vignette 0.491* (.033) 0.510* (.027) −0.013 (.481) 0.072 (.393) 0.089 (.363) 0.169 (.253) Constant 4.444*** (.000) 3.411** (.001) 3.200*** (.000) 1.219 (.125) 3.698*** (.000) 1.664* (.049) Observations 549 549 568 568 568 568 Adjusted R2 0.004 0.032 −0.002 0.027 −0.002 0.013 Covariates NO YES NO YES NO YES Note: OLS regressions with p-values in parentheses. One-tailed test for statistical significance. Dependent variables are measured as additive measures ranging from 0 to 10. Covariates: variables listed in Table 1. ***p < .001, **p < .01, *p < .05, +p < .10. Open in new tab Table A5. Survey Experiment: Threat Perception as a Function of Task Type (Treatment Groups). OLS Estimation DV: Quality of Care DV: Profession's Position and Monopoly DV: Own Work Tasks and Job Security (1) (2) (3) (4) (5) (6) Ref.: Complementary task vignette  Core task vignette 0.491* (.033) 0.510* (.027) −0.013 (.481) 0.072 (.393) 0.089 (.363) 0.169 (.253) Constant 4.444*** (.000) 3.411** (.001) 3.200*** (.000) 1.219 (.125) 3.698*** (.000) 1.664* (.049) Observations 549 549 568 568 568 568 Adjusted R2 0.004 0.032 −0.002 0.027 −0.002 0.013 Covariates NO YES NO YES NO YES DV: Quality of Care DV: Profession's Position and Monopoly DV: Own Work Tasks and Job Security (1) (2) (3) (4) (5) (6) Ref.: Complementary task vignette  Core task vignette 0.491* (.033) 0.510* (.027) −0.013 (.481) 0.072 (.393) 0.089 (.363) 0.169 (.253) Constant 4.444*** (.000) 3.411** (.001) 3.200*** (.000) 1.219 (.125) 3.698*** (.000) 1.664* (.049) Observations 549 549 568 568 568 568 Adjusted R2 0.004 0.032 −0.002 0.027 −0.002 0.013 Covariates NO YES NO YES NO YES Note: OLS regressions with p-values in parentheses. One-tailed test for statistical significance. Dependent variables are measured as additive measures ranging from 0 to 10. Covariates: variables listed in Table 1. ***p < .001, **p < .01, *p < .05, +p < .10. Open in new tab © The Author(s) 2019. Published by Oxford University Press on behalf of the Public Management Research Association. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)
TI - Service Professionals' Response to Volunteer Involvement in Service Production
JO - Journal of Public Administration Research and Theory
DO - 10.1093/jopart/muz028
DA - 2020-04-15
UR - https://www.deepdyve.com/lp/oxford-university-press/service-professionals-response-to-volunteer-involvement-in-service-5f0t2FrGNu
SP - 1
VL - Advance Article
IS - 
DP - DeepDyve
ER -