Dual effects of economic dependence on platform workers’ work experiences? The role of autonomy, commitment and online review systems Open Access

Platform-enabled gig work is one form of fluid work, a broad term developed to describe work performed by people who acquire paid work assignments via an online labor platform (Kuhn and Maleki, 2017) without being in traditional employment relationships with an organization (Capgemini Research Institute, 2020). The presence of fluidity and flexibility in gig work and online labor platforms is counterbalanced by a theme of control in the research literature: which actors have control, how it is exercised, and what the effects are for both the platform and the worker (Cini, 2022; Kuhn and Maleki, 2017; Shapiro, 2018; Veen et al., 2020). While platforms claim authority over many important functions (e.g. worker membership, task allocation, and pricing, depending on the platform), they cede some control to the workers in terms of when they want to work and for how long, and to the clients by means of evaluating workers through online review systems (Meijerink et al., 2021; Vallas and Schor, 2020). As noted by Schor et al. (2020), notions of control and autonomy are inherently intertwined with elements of worker motivation and economic background. Although online labor platforms each offer a concrete set of rules, regulations, and algorithms (Duggan et al., 2020; Meijerink and Keegan, 2019), workers respond to them differently depending on their personal and professional situations (Kuhn and Maleki, 2017; Myhill et al., 2021).

Accordingly, this study examines control through the lens of economic dependence, investigating the relationship between levels of perceived dependence on income generated from platform work and work satisfaction of the platform workers. Here, platform workers are defined as individual workers who perform work via online labor platforms (e.g. Task Rabbit, Fiverr) that facilitate connections between supply and demand for freelance labor. The landscape of such online labor platforms is very broad, with platforms offering different features, controls, and opportunities for interaction between workers and clients (Kuhn and Maleki, 2017). In this paper we focus on platforms that facilitate work performed on location and planned in advance, excluding on-demand services such as delivery and ride hailing.

The dynamics related to control in fluid work affect the workers both professionally and personally, impacting aspects of their well-being. Without a traditional employment relationship, platform workers are supposed to enjoy considerable levels of autonomy (i.e. the freedom and discretion to make work-related decisions). Indeed, research shows that they can, in principle, decide themselves when, where and for whom they perform labor (Kuhn and Maleki, 2017; Shapiro, 2018; Wood et al., 2019). However, some scholars have noted that autonomy can be curtailed when workers have high levels of dependence on platform earnings, and that it is difficult to achieve work satisfaction under these circumstances (Goods et al., 2019; Schor et al., 2020). Workers who are highly dependent on platform earnings may experience limited autonomy and flexibility for several reasons. They may need to work long hours, potentially more than full-time, to generate sufficient income, leaving less room to decide when to work. They may be afraid to lose access to paid tasks if they do not comply with the strict and sometimes volatile rules set by a platform, making workers feel powerless to make decisions in their own favor (Goods et al., 2019; Schor et al., 2020). This situation becomes more severe when platforms operate online performance review systems that lock-in workers and require them to comply with idiosyncratic client wishes (Kuhn and Maleki, 2017; Wood et al., 2019; Veen et al., 2020). Dependence on the income derived from platform work may also have positive effects, however, when other job alternatives outside the platform economy (that involve standard employment relationships) are less attractive or simply unavailable to workers (Kalleberg and Dunn, 2016; Van Doorn et al., 2022). In such cases, platform firms provide workers with opportunities that may lead workers to have positive affect toward platform work that they depend on for their income (Kuhn and Maleki, 2017). It is against this backdrop that we examine the potential dual (i.e. negative and positive) effects of economic dependence on the work satisfaction of platform workers.

This study makes two primary contributions to the literature. First, we strive to better understand inconsistencies in the literature regarding the effects of economic dependence on platform work and workers’ outcomes such as work satisfaction. While multiple studies have reported a negative relationship between platform workers’ economic dependence and worker outcomes (e.g. Lin et al., 2023; Jing et al., 2023; Schor et al., 2020), others have found non-significant (Lin, 2021; Wang et al., 2024) or even positive relationships between such variables (Wood et al., 2022). This suggests that economic dependence has a complex relationship with worker outcomes. In our current study, we seek to understand why these dual effects manifest by studying the mediating role of affective commitment and autonomy, and under what conditions the effect of economic dependence on work satisfaction tends to be more positive or negative by studying the moderating role of workers’ attributions of online review systems. In doing so, we contribute by highlighting the benefits of economic dependence, and how these depend on whether it affords worker autonomy and the use of online rating systems to challenge the prevailing assumption that economic dependence is solely detrimental to platform workers. Second, we build on the work of Schor et al. (2020) and examine the relationship between economic dependence and work satisfaction quantitatively in a large sample of over 1,000 platform workers across six different platforms in the Netherlands. This approach helps bring another perspective to the existing qualitative research on these relationships, while still focusing on the experiences and perspectives of the workers themselves, thus helping build toward a more complete picture on the dual relationship between economic dependence and work satisfaction among platform workers.

The remainder of the paper proceeds as follows. First, we examine the role of economic dependence in platform work and how it theoretically could lead to development of work satisfaction perceptions in two different ways, one positive and one negative. We then discuss the role of algorithmic control in the context of platform-operated review systems and how worker perceptions of the review system could affect the economic dependence-work satisfaction relationship. We develop a set of hypotheses and then describe the survey methodology used to collect data for testing our hypotheses. We discuss the implications of our findings for further development of the research literature and practical application.

In spite of their apparent freedom to operate as they choose, platform workers can be dependent in many ways on the online labor platforms and associated algorithms that facilitate matchmaking between clients and workers (Kuhn and Maleki, 2017; Schor et al., 2020; Wood et al., 2023). Dependence on platform earnings, or economic dependence, is a function of having access to job alternatives and economic insecurity (Greenhalgh and Rosenblatt, 1984). That is, platform workers who are highly dependent on platform earnings have few or no alternatives other than performing platform work for generating income and find that the earnings generated via platform work are important to them, for instance, because they constitute a significant portion of their total income. As outlined below, economic dependence on platform work can have implications for platform workers’ work satisfaction, that is, their positive feelings about platform work resulting from an evaluation of its characteristics. We expect that economic dependence and work satisfaction of platform workers can be both positively and negatively related. To better understand why this may be the case, we propose studying two mediator variables: autonomy (to explain the negative economic dependence-satisfaction relationship) and affective commitment (to explain the positive relationship).

Workers who are highly dependent on platform earnings face precarious working conditions that reduce the satisfaction derived from work (Goods et al., 2019; Schor et al., 2020). This negative relationship between dependence and work satisfaction can be understood by considering platform workers’ autonomy (Meijerink and Bondarouk, 2021; Wood et al., 2019). Without a traditional employment relationship, platform workers should have the freedom to choose when they work, for how long, and how many jobs they accept (Cropanzano et al., 2023; Kuhn and Maleki, 2017). However, their economic dependence on any given platform may result in lower levels of autonomy than anticipated if workers feel obligated to accept offers for undesirable tasks in order to build their reputation on a platform, work longer hours than they would prefer, or accept projects from unreliable clients to generate an income (Goods et al., 2019; Sutherland et al., 2020). This is consistent with resource dependence theory (Pfeffer and Salancik, 2003) which predicts that unmitigated resource dependencies (on the platform firm or clients) negatively affect autonomy as the dependent party needs to spend considerable time and energy to satisfy the demands of the external resource controlling party (Drees and Heugens, 2013). This dependence of platform workers on the platform firm and/or their clients is reinforced by an oversupply of workers on online labor platforms, clients that care little for workers’ (financial) interests, and algorithmic control (Kellogg et al., 2020; Smith et al., 2021). Accordingly, platform workers who are economically dependent on platform work are likely to experience lower levels of autonomy (Kuhn and Maleki, 2017). Ultimately, we predict that these low levels of autonomy drive down work satisfaction of platform workers. This is consistent with self-determination theory which argues that autonomy is a basic psychological need which, if not met, will reduce worker motivation and satisfaction (Gagné and Deci, 2005).

The above implies that economic dependence relates negatively to work satisfaction through its negative effect on autonomy (see Figure 1 which details our conceptual model). In some cases, substantial losses of autonomy in areas such as pricing and creative direction can result in workers leaving the platform altogether (McDonald et al., 2021b). However, when workers are dependent on platform earnings, they are less likely to have the choice to leave and will continue performing platform labor even when they experience negative outcomes, thereby deteriorating levels of satisfaction with platform work. This implies that economic dependence itself does not directly impact work satisfaction, but indirectly through platform workers’ experiences of the consequences this has for jobs.

Research shows that autonomy is a core dimension for evaluating jobs (Hackman and Oldham, 1975), particularly for workers who are attracted by online labor platforms’ promises to work where, when, and for whomever they want (Kuhn and Maleki, 2017). Psychological contract theory predicts that workers respond negatively to a firm’s failure to honor its promises (Rousseau, 1998; Sherman and Morley, 2020). In line with this, we expect negative responses such as low work satisfaction among platform workers if autonomy – as a core work dimension that is promised to them – is restrained, for instance, when they have little freedom due to high dependence on platform earnings. Taken together, this suggests that economic dependence on platform work is negatively related to work satisfaction due to the mediating role of autonomy. Accordingly, we hypothesize the following:

Although platform workers’ dependence on platform earnings may have negative consequences for work satisfaction due to low perceptions of autonomy, there are reasons to assert that economic dependence can play multiple roles in this relationship and also positively relate to work satisfaction through a different path (see Figure 1). As noted by Schor et al. (2020), workers’ lack of dependence, or independence, may come with negative effects of uncertainty, desperation, and risk to platform workers, especially those with lower skill levels. Moreover, for many workers who have limited access to “traditional” jobs due to migration status, language skills, or lived experience with a disability, dependence on platform work may be a better alternative to unemployment or other types of precarious work (Kalleberg and Dunn, 2016; Lee, 2023; Van Doorn et al., 2022). Although these platform workers are dependent on platform earnings, we predict that they appreciate this as a “less bad” alternative than precarious forms of informal work or jobs underpinned by a standard employment relationship that offer less flexibility and possibilities to turn down inconvenient tasks.

Accordingly, economic dependence may have positive effects for platform workers if the dependence reduces negative experiences and results in perceptions of being supported and cared for. Research has shown that this support builds affective commitment (Shore and Wayne, 1993), which, in a platform context, refers to workers’ emotional bond with a platform firm which manifests as the identification with, involvement in, and enjoyment of the relationship with the selected platform firm. In spite of the absence of a traditional employment relationship, affective commitment to platforms has been demonstrated (Mousa and Chaouali, 2023) and offers one potential path for understanding how workers relate with platforms (Bucher et al., 2024). For example, Panteli et al. (2020) describe findings in which MTurk workers expressed a range of bonds with the platform. Many workers indicated instrumental bonds with the platform, representing some form of dependence on the platform because they had few alternatives for other work, but also more personally meaningful kinds of bonds reflecting a positive affective commitment to the platform. Dependence on platform earnings may also serve to reduce role ambiguity, which has been found in two meta-analyses to be negatively related to affective commitment (Mathieu and Zajac, 1990; Meyer et al., 2002). When platform workers have too many alternatives for finding work, their daily tasks may become too ambiguous and effortful, consistent with the logic of decision fatigue (e.g. Vohs et al., 2008).

Consequently, we argue that workers who perceive they are dependent on gig work for their work opportunities and income will develop positive perceptions of affective commitment toward the platforms that provide them with these work opportunities. This is consistent with conservation of resource (COR) theory, which predicts that employees reinvest such types of perceived support by organizations by displaying higher levels of commitment (Boon and Kalshoven, 2014; Hobfoll, 2011). Affective commitment in turn has been shown to have a positive relationship with work satisfaction (e.g. Meyer et al., 2002). Accordingly, we predict that economic dependence itself does not directly impact work satisfaction but rather depends on whether workers see support coming from the platform which they reciprocate by committing themselves to the platform. Further, some workers who are currently estranged from traditional labor markets and therefore dependent on platform earnings may see platform work as a steppingstone to regular jobs (Bucher et al., 2024). This means that they may regard being dependent on platform work as desirable (at least for a while) and thus as a type of support coming from the platform firm when other job alternative prove less desirable in the interim (Lee, 2023). This type of platform-enabled support is what workers reciprocate by showing high(er) levels of affective commitment to the platform. On this basis, we argue that economic dependence positively relates to work satisfaction to the degree that it fosters affective commitment as a mediating state. Accordingly, we hypothesize the following:

Although we suggest that platform workers’ perceived economic dependence and work satisfaction are negatively related due to the mediating role of autonomy (Hypothesis 1), we suggest there are boundary conditions that may attenuate this negative relationship. Specifically, we predict that this depends on aspects of the performance management operated by online labor platforms. One of the most common features of online labor platforms is the review system that allows requesters to rate and comment on platform workers’ performance, with algorithms subsequently using these data for recommendations or decision making. Platform workers may see these review systems as a way to control their behavior and limit autonomy as the ratings in particular have implications for their access to future gigs, income, or accessibility to the online marketplace of the platform firm (Cropanzano et al., 2023; McDonald et al., 2021a; Wood et al., 2019). This view of online review systems has been the dominant assumption in the literature to date. Human resource management (HRM) research, however, shows that employees can have positive perceptions of performance appraisal systems to the extent that these help them improve and perform more effectively (Van De Voorde and Beijer, 2015; Nishii et al., 2008; Meijerink et al., 2021). In line with this, we examine platform workers’ perceptions of the extent to which online review systems afford such support to platform workers.

Existing research does show that online review systems can be supportive to platform workers as they help to build their skills, make them feel respected, or acquire desirable gigs (Cameron, 2020; Lehdonvirta et al., 2019; Meijerink and Bondarouk, 2021; Morales and Stecher, 2023). We expect that this helps to offset the negative relationship between economic dependence and autonomy. For instance, Lehdonvirta et al. (2019) show that positive online evaluations offer workers the freedom to charge higher prices for their services. Similarly, Sutherland et al. (2020, 464) found that platform workers with high-level reviews, also known as reputation scores, saw “clients bringing projects to them, rather than having to search and bid for them” which makes workers more autonomous to decide themselves what price to charge to clients or which clients to work for. Moreover, platform workers see online reviews as a source of feedback that allows them to build their competences and sell these to the highest bidder (Kost et al., 2018; Margaryan, 2019; Margaryan et al., 2022). Although economic dependence on platform work remains the same, in such cases, workers become less dependent on individual clients for making an income, thereby offering more autonomy to decide when, where and for whom to work. Taken together, if workers perceive that online review systems have a beneficial purpose and are intended to help them rather than control them, it is likely that this helps overcome negative effects of economic dependence, thereby driving up their feelings of autonomy. Accordingly, we propose the following moderator effect as shown in Figure 1:

Perceptions about online review systems could also impact the second half of this mediated relationship, that between autonomy and work satisfaction. As discussed above, we predict that worker perceptions of job autonomy will be positively related to work satisfaction. When workers have positive perceptions of the purpose of the online review system, the relationship between autonomy and satisfaction will be even stronger. This ties back to our earlier claim that platform workers’ autonomy may be technically high, but in practice perceived to be limited such as when online labor platforms create an oversupply of workers. In such cases, we expect that worker autonomy will limitedly translate into work satisfaction. This will be different in cases when platform workers perceive that online review systems help them improve and perform more effectively. Moreover, job design research has shown synergistic effects between autonomy and feedback in explaining work satisfaction (Hackman and Oldham, 1975; Parent-Roucheleau and Parker, 2022). Indeed, Cameron (2020) has shown how online ratings give platform workers a sense of appreciation coming from clients which adds to their satisfaction with platform work. Finally, by providing opportunities for feedback and learning (Kost et al., 2018; Margaryan, 2019; Margaryan et al., 2022), online review systems afford competence development to platform workers (Parent-Roucheleau and Parker, 2022). Research shows that competent workers are better able to derive value from HRM activities such as autonomy and performance feedback (Meijerink et al., 2016). Consequently, we expect that platform workers who have a positive perception of online review systems can make effective use of and enjoy the autonomy granted to them, thereby increasing levels of work satisfaction. We propose the following moderator effect:

We relied on survey data from a larger project on the work experiences of platform workers that perform “on-location” platform work in the Netherlands. We focused on this type of platform work to rule out the effect of cross-national differences in work experiences that are more characteristic of online platform work (e.g. via Fiverr or Upwork) conducted across multiple jurisdictions and (inter)national labor markets. The workers in our sample find work primarily via six platform firms (Charly Cares, Helpling, Roamler, Sjauf, Temper, and YoungOnes) that operate in industries such as hospitality, retail, logistics, domestic cleaning, and babysitting. The platforms agreed to facilitate access to the workers on their platform but did not set any conditions or restrictions for the survey content. The research was approved by the ethics board at University of Twente (#BCE 210123).

The platform workers were invited by email or other electronic messages sent by the platform via which they work. The message included a link to an online survey and informed the respondents that participation was voluntary. To reassure respondents about confidentiality, completed surveys were returned directly to the researchers. Participants who completed the 74-item (51 items to measure latent constructs and 23 items for demographics and control variables) survey could enter in a random drawing for a prize. The survey could be completed in Dutch or English. The original items were formulated in English and translated into Dutch. Following Brislin (1970) we conducted back-translation on all measures.

The survey was sent to 12,677 workers in the Netherlands who had completed at least one gig via one of the six platforms in the past 12 months. We received 1,114 responses. Twenty respondents were removed from the dataset because they reported they did not actually conduct gig work resulting in a final sample of 1,094, a response rate of 8.6%. The largest percentage of respondents worked via Roamler (46.9%), followed by Temper (21.9%), YoungOnes (10.4%), Helpling (10.1%), CharlyCares (9.5%) and Sjauf (1.2%). Of all respondents, 58.2% were female, with 84% having grown up in the Netherlands. Regarding highest level of education, 17.6% reported completing high school, 29.2% a vocational education, and 49.4% completed a bachelor or master’s degree. Most of the respondents (69%) used only one online labor platform to find work. The average age of the respondents was 32.4 years (SD = 11.2) and respondents reported having performed an average of 130 gigs through the platforms.

Whenever possible, we relied on existing measures that we adjusted to the platform context. As some of the respondents find work on multiple platforms, they were asked to first report the platform from which they generate most of their income and respond to the subsequent items with this platform in mind. Items were used with a five-point rating scale ranging from Strongly Disagree to Strongly Agree unless otherwise noted.

Dependence on platform earnings was measured using a 7-item scale assessing respondents’ perceptions of available job alternatives and economic insecurity. This scale was based on measures developed by Van Dam (2005) and Clark (2005) and had a coefficient alpha of 0.88. The items were coded to reflect perceptions of low alternatives and high insecurity, thus economic dependence on the platform work they were performing. Example items are “I can easily get another job outside the platform economy if I want to” (reverse coded) and “To maintain the standard of living I desire for myself, I must keep working via a platform.”

The autonomy measure was developed specifically for this study. Items were written to address four primary ways in which platform workers may reflect autonomy in their work; when to perform work, which jobs to carry out, for how many hours they work, and how they carry out their work (Schor et al., 2020). The four-item scale had a coefficient alpha of 0.78.

The review system attributions measure was based on items by Nishii et al. (2008) to assess how workers could attribute the purpose of different aspects of the online review feature to support platform workers (Nishii et al.’s category of Enhancing Quality and Employee HR Attribution). The five-item measure had a coefficient alpha of 0.87. An example item is: “The customer reviews are very useful for me to get hired by clients/customers via the platform”.

Affective commitment was measured using a five-item version of the scale developed by Allen and Meyer (1990), substituting “platform” for “organization” in each item. Coefficient alpha was somewhat low, but acceptable, at 0.65. A sample item of this measure is “The platform has a great deal of personal importance to me.”

Work satisfaction was assessed with a single item: “I am satisfied with the work I do through/via the platform.” Single item measures of satisfaction have been found to be highly correlated with broader measures of satisfaction and to demonstrate adequate levels of reliability (Nagy, 2002; Wanous et al., 1997).

As control variables, we included tenure, operationalized as the amount of time respondents had been working via the platform, and the number of gigs completed. We also included country of origin (coded as Netherlands vs. any other country) because people who immigrated to another country may be more dependent on platform work due to having fewer job market opportunities (van Doorn et al., 2022). We included level of education (a six-point scale where six equals a master’s degree or higher) as a control variable because people with a higher education may have skills that make them more attractive as a gig worker. Finally, we also included a set of five dummy variables representing the primary platform respondents used to find work. Roamler was set as the reference variable. We used mean replacement on all variables to account for missing data. Means, standard deviations, and intercorrelations of the variables are presented in Table 1.

We conducted a confirmatory factor analysis using Mplus version 8.4 to check the structure of the measures. We first tested a four-factor model representing economic dependence, autonomy, quality attributions, and affective commitment. This model did not fit the data well (see Table 2). We then tested a five-factor model with economic dependence broken down into two subcomponents (alternatives and economic dependence) and a second order factor model with alternatives and economic dependence loading onto the second order factor of dependence, neither of which represented a significant improvement in the model fit. We then modeled economic dependence using a bifactor model, which provided acceptable fit to the data (chi square = 1010.68 [df = 169], p < 0.01; BIC = 57,231.86; RMSEA = 0.067, CFI = 0.92, SRMR = 0.054).

We tested our hypotheses using structural equation modeling in MPlus version 8.4. We first tested the mediation relationships for both mediators simultaneously using the maximum likelihood estimator and 10,000 bootstrapping samples. Economic dependence was a significant predictor of both affective commitment (b = 0.164, p < 0.01) and autonomy (b = −0.184, p < 0.01). Affective commitment (b = 0.340, p < 0.01) and autonomy (b = 0.294, p < 0.01) were both then significant predictors of work satisfaction. The hypothesized mediation effect was supported for both paths. The standardized path through autonomy was negative (indirect effect = −0.054, 95% CI [-0.071, −0.037]) and the path through affective commitment was positive (indirect effect = 0.056, 95% CI [0.038, 0.074]). Total indirect effects were nearly zero (0.028, 95% CI [-0.024, 0.027]) as the commitment and the autonomy paths had opposite effects. The direct effect from economic dependence to satisfaction was non-significant (b = 0.027, p = 0.115). Thus, H1 and H2 were supported.

We tested the moderated mediation hypotheses using the maximum likelihood estimator with 10,000 bootstrapping samples and centering the predictors before creating the interaction terms. For testing the moderation of the relationship between economic dependence and autonomy, the interaction term is significant (b = 0.198, p < 0.01, 95% CI [0.127, 0.265]). The negative effect of economic dependence on autonomy remains negative but becomes less negative as attributions about the helpfulness of the online review systems increase (see Figure 2). This supports H3. In contrast, H4 regarding the moderation of the relationship between autonomy and satisfaction was not supported. The interaction term is not significant and zero appears in the 95% confidence interval (b = −0.015, p = 0.756, 95% CI [-0.096, 0.058]).

The results of this study suggest that economic dependence on platform work can have both positive and negative effects on platform workers, with these effects partially driven by worker perceptions about algorithmic management through performance reviews. As suggested by Schor et al. (2020), economic dependence was associated with reduced perceptions of autonomy, contributing to lower levels of satisfaction. However, we also find that economic dependence can create a positive effect for workers as it was related to affective commitment toward the platform, which was positively linked with satisfaction. In this sample of over 1,000 platform workers in the Netherlands, these competing effects nearly canceled each other out.

Our results shed further light on the relationship between platform workers’ economic dependence on platform work and their subsequent work experiences. While previous studies suggest that economic dependence is negatively associated with work satisfaction (Kuhn and Maleki, 2017; Schor et al., 2020), we find that it also improves workers’ experiences. Which alternative (i.e. the high-road or low-road) outcome manifests in a given situation depends on whether platform workers experience economic dependence as curtailing the degree of autonomy at work or whether it fosters a sense of affective commitment to the platform firm. This implies that a negative effect of economic dependence on work satisfaction is more likely attributable to characteristics of the work (i.e. autonomy) while positive effects can be attributed to how workers relate to the platform (i.e. affective commitment). Accordingly, we call for future studies to examine under which conditions economic dependence and work satisfaction are more positively or negatively related, and how these differences depend on work and platform characteristics. For example, Mousa and Chaouali (2023) demonstrated that use of job crafting techniques among crowdsourced gig workers positively affected affective commitment through perceptions of meaningful work. Options like job crafting may increase the autonomy and fluidity of these work contexts. From a platform perspective, it could also be useful to explicitly consider differences between work tasks that are “on demand” (e.g. delivery or ride hailing) compared to those that are planned in advance (e.g. domestic cleaning, babysitting and pet care). The ability to plan work tasks in advance may lead to more positive worker outcomes by increasing autonomy or may be related to other perceptions such as increased work-life balance.

Our results also have implications for research into the effect of online review systems on labor platforms. While previous studies have generally reported on the downsides of platform-based performance appraisal through algorithmic control, we find that online review systems can benefit workers. In fact, workers do see value in online review systems when they help them get hired by clients or generate higher levels of income via the platform (e.g. Morales and Stecher, 2023). In such cases, the negative effect of economic dependence on job autonomy and thus work satisfaction weakens when employees perceive that online review systems support them. This is surprising since online review systems, especially those that focus on ratings, are said to make workers more dependent on platform work as they create lock-in effects (Wood et al., 2019). This implies that the effects of online review systems are dual in nature, which raises the call for future studies to examine under what conditions online review systems bring about desirable outcomes compared to undesirable outcomes.

From a practical perspective, this study has implications for how online labor platforms design and communicate about their software features. We found that when platform workers had positive attributions about the review systems, the negative relationship between dependence and autonomy was reduced. Platforms could design their systems in a way to provide fields for raters to provide more specific feedback for improvement rather than simply giving a rating on a five-point scale or offering a thumbs-up or thumbs-down signal as an evaluation. Algorithms could be used in a supportive way, providing recommendations for skill development. Further, the platforms should clearly communicate with gig workers about the usefulness of the review system for improving service quality. Communicating about the intention of an HR system may facilitate development of more positive attributions (Nishii et al., 2008). Platforms could also design performance improvement tools into the platforms, allowing workers to evaluate trends in their ratings across time or visualize them in other ways (e.g. by type of project or by client) that may help them better use and understand the data.

Given the positive impact of affective commitment on satisfaction, platforms may also take other steps to improve affective commitment among gig workers, pursuing a commitment-oriented strategy (Meijerink and Keegan, 2019) rather than a lock-in strategy for managing relationships with gig workers. Lock-in strategies seek to retain workers on the platform by making it difficult for them to leave because they would lose value, such as having to re-build their platform reputation and customer base. Commitment-oriented strategies, in contrast, use more positive inducements to add value for gig workers and build their voluntary commitment to the platform. Offering services such as client conflict resolution or facilitating worker voice (Gegenhuber et al., 2021) are other practical steps platforms could take to enhance commitment.

The study does have some limitations. First, we acknowledge limitations related to our sampling choices. While we have a relatively large sample, we sampled from platforms in a single country (the Netherlands) that focused on in-person delivery of services. Both cultural values and laws that vary between countries could impact how people respond to online labor platforms. Because autonomy was such an important construct in our study, we could imagine that our results would generalize more readily to cultures characterized by similarly high levels of individualism. Regarding the legal context, platforms in some countries (e.g. the Netherlands) may grant more autonomy to workers to prevent legal efforts to reclassify workers as employees. In other countries, such as Taiwan (Lee, 2022) or the UK this may be less of an issue. Future research should examine these relationships among a broader set of platforms and workers. The data were collected from the platform workers at a single point in time, creating a risk of common method bias. In the future it would be interesting to examine these processes over time, looking at how the relationships unfold over multiple time periods. In addition, the data were collected in 2021 during the COVID-19 pandemic. Workers may have been more economically dependent on platform work during that time because a lack of alternatives. There was also higher competition between workers on the platforms at that time because demand for in-person services dropped due to lockdowns and other government restrictions.

We also acknowledge some limitations related to measurement. The affective commitment measure had surprisingly low reliability at 0.65. It is unclear if this is due to having translated the items from English into Dutch. It is also possible that affective commitment might take on a different meaning in the gig work context (e.g. Kuhn and Maleki, 2017), and thus the measure does not function in the same way with this population. We did review the survey items for face validity with a set of platform workers prior to collecting data and they indicated that the affective commitment measure was meaningful to them. In addition, our measure of economic dependence was somewhat different than how Schor et al. (2020) evaluated dependence, which could have contributed to the diverging results. Schor et al. identified three different categories of workers: dependent workers who are wholly or primarily dependent on the platform for their income, partially dependent workers who either have a part time job in addition to their gig work or work on multiple platforms, and supplemental workers who are not dependent on the platforms at all, engaging in gig work for fun or for some extra money. We asked workers to evaluate the extent to which they are dependent on platform work due to a lack of other job alternatives and the level of economic security provided by their platform work. The bifactor measurement model that we fit to our data suggests that in addition to these two dimensions of economic dependence, there is a general factor of economic dependence that could be more clearly articulated in future research.

In addition to addressing these limitations, future research could examine the role of affective commitment in the negative path through autonomy ^[1]. Noting the observed positive correlation between these two variables, it is possible that perceptions of autonomy lead to higher levels of affective commitment through a social exchange mechanism. Longitudinal research could help shed more light on this relationship. Future research could also more directly examine the role of job characteristics in the gig worker experience. We considered role ambiguity as a potential explanatory factor in the relationship between economic dependence and affective commitment, and future research should directly assess role ambiguity and model it specifically to understand its impact. Other job characteristics such as role conflict have also been found to impact affective commitment (Meyer et al., 2002) and could be studied in the gig work context. For example, higher levels of economic dependence may also reduce role conflict as the platform worker does not have to balance competing commitments, and this could result in higher levels of affective commitment to a particular platform.

In a large sample of Dutch platform workers, this study found that economic dependence can have both positive and negative effects on the worker experience in this segment of the fluid workforce. This stands in contrast to earlier work that found evidence of primarily negative impacts (Schor et al., 2020). Further, the negative path through autonomy was moderated by the extent to which workers made positive attributions about the extent to which online evaluation systems offered them support for service quality enhancement, suggesting that algorithmic management can also have positive effects. We contribute to better understanding of the complex interactions between platform workers and the intermediary online labor platforms.

We thank Jorn Diekmeijer for assistance with data collection and Jenna Van Fossen for assistance with data analysis. An earlier version of the paper was presented at the 8th International eHRM Conference (2022), Aarhus, Denmark.