Hyperactive work? How platform work differs from other working arrangements in the quality of work

During the past decade, technological advances transformed how industries operate. The gig economy is no exception. While alternative work arrangements have been on the rise (Katz and Krueger, 2019), “platform work” is a recent, rapidly growing addition (Kässi et al., 2021). Digital labor platforms define this working arrangement (Watson et al., 2021). They connect contractors with individuals who perform “gigs” or short-term labor agreements (Cropanzano et al., 2023). While digital labor platforms have grown in popularity, so has their notoriety due to their precarious working conditions (Heeks et al., 2021; Melián-González and Bulchand-Gidumal, 2021; Vallas and Schor, 2020; Wood et al., 2019). Others highlight benefits such as flexibility (Schor et al., 2020). Yet, we lack clear, empirical evidence regarding the extent to which these experiences are distinctive or reflect more general features of contemporary work. Do individuals who perform platform work experience more autonomy and precarity than those in other work arrangements? Clarifying to what extent and how platform work differs from other working arrangements is crucial for researchers and policymakers. Hence, there is a pressing need to advance our understanding of platform work. In other words, we aim to answer to what extent platform work, in comparison to other working arrangements, has a unique pattern of relationships with job characteristics and well-being.

Platform work disrupts our current understanding of working arrangements (Duggan et al., 2020). This study examines whether platform work resembles or differs from other arrangements. To paint a comprehensive picture of platform work, we aim to complement the existing fine-grained strokes in the literature with broad ones. Research has predominantly focused on differences between different types of platform workers (Cropanzano et al., 2023). This methodological approach focuses on within-working arrangement differences, elucidating the experiences of platform workers and the important antecedents of those experiences. They may reveal why certain workers intend to leave the platform and others do not (Ihl, 2024; Wang et al., 2022). However, they cannot reveal whether platform work relates to better (or worse) quality of work than other work arrangements. While differences exist between platforms, such as location and in terms of content, they share common characteristics, including a tri-party labor structure consisting of a customer, intermediary and requester, absence of contractual relationships, short-lived assignments and piece-rate pay (Bérastégui, 2021) that justify studying between-work arrangement differences. Thus, we aim to address the lack of quantitative, empirical studies that compare platform work with other working arrangements in terms of work design and well-being.

Therefore, we draw on job demands-resources (JD-R) theory (Bakker et al., 2023) to propose that platform work is “hyperactive” (both straining and motivating) and relatively more characterized by technological characteristics than other working arrangements. We consider jobs to be hyperactive when job demands are excessive in combination with a high amount of job resources. This differs from other types of work, which might not have both at such levels, such as exhausting, engaging, boring and satisfying work (see Figure 1; cf. Bakker and Demerouti, 2007). To test these propositions, we conduct between-working arrangement and dominance analyses, which reveal the extent to which high JD-Rs and ambivalent well-being characterize platform work, and assess the relative impact of technological factors compared to other arrangements. This analysis helps determine whether traditional JD-R assumptions about well-being hold across different work forms, addressing Cropanzano et al. (2023) observation that: “While there have been several studies documenting the mounting stressors and accompanying emotional responses of gig workers, what remains less well understood is whether these responses are different than those of permanent workers” (p. 508).

The gig economy comprises any temporary job performed on an on-demand basis (e.g. live performance of music, babysitting). Platform work is a unique, relatively novel “gig.” The gig economy surged due to the recent addition of platform-mediated contracting (Kässi et al., 2021; Spreitzer et al., 2017): “labor contracted and compensated on a short-term basis” (Cropanzano et al., 2023, p. 494). Gigs such as transcribing, data entry, ride-hailing, translation and delivery are unified by the mediation of the platform. Platform-mediated gig work differs from traditional gig work due to its reliance on the digital labor platform. Namely, the platform coordinates human resources as an intermediary between requestors and workers (Pesole et al., 2018). They often use algorithms to coordinate human resources, such as matching “workers” with “requesters,” setting productivity targets and giving work assignments (Duggan et al., 2020; Watson et al., 2021). Additionally, platforms exert algorithmic control by monitoring users, enforcing rules, providing feedback and using “soft” power by nudging desired behaviors (Möhlmann et al., 2021; Yang et al., 2026). Examples of algorithmic control include using ratings to determine the number of work opportunities individuals receive, offering additional incentives when demand exceeds supply and imposing costs to inhibit undesired behaviors (Cameron, 2022; Rosenblat and Stark, 2016; Spreitzer et al., 2017).

Namely, we predict more JD-Rs and a higher relative importance of technological JD-Rs than in other working arrangements, characterize platform work. Namely, two perspectives organize platform work: scientific management (Taylor, 1911) and the gig working arrangement (Cropanzano et al., 2023). As scientific management prescribes, digital labor platforms collect data to optimize work processes, assign work to the most suitable candidates, subdivide tasks and responsibilities into smaller pieces and connect financial incentives to productivity (Duggan et al., 2020). This creates an asymmetric power dynamic between workers and other parties (Vallas and Schor, 2020). Workers have little to no influence on the rules and payment structures imposed by the platforms. Yet, platforms recognize workers as self-employed individuals who perform short-term contracts (Spreitzer et al., 2017). They can choose for whom they work and work independently (Cropanzano et al., 2023). While scientific management thwarts well-being (health impairment; Parker et al., 2017), self-employment and the ability to work according to one’s own volition have been linked to thriving (motivational; Stephan et al., 2020).

In the present study, we investigate how platform work differs from other forms of employment, including permanent (long-term), temporary (fixed-term), on-demand (flexible) and self-employment (freelancer or business owner). Permanent employment lacks a fixed end date and involves agreed-upon weekly hours (Statistics Netherlands, 2015). Contrastingly, a fixed duration defines temporary employment, serving as a short-term workforce supplement (Cappelli and Keller, 2013). On-demand employment involves flexible labor, including on-call, zero-hour contracts and temporary agency work, scheduled as needed. While such afford organizations to scale their workforce according to demand, they offer limited predictability and security. In both permanent and temporary contracts, organizations exert directive control. They determine the outcomes work should produce and often how individuals should work (Cappelli and Keller, 2013). Conversely, self-employed individuals negotiate their own terms with organizations, specifying the desired results without direct control. In line with the extended categories of work arrangements developed by Spreitzer et al. (2017), we argue that platform work constitutes a unique arrangement due to its digital, short-term and fragmented nature (Cropanzano et al., 2023; Pesole et al., 2018) and tri-party labor structure, algorithmic management and piece-rate pay (Bérastégui, 2021).

The organization of platform work, guided by scientific management and gig work principles, yields an ambivalent experience. Anxiety, depression and distress (Glavin and Schieman, 2022), but also thriving, happiness and engagement, characterize platform work (Ashford et al., 2018; Cameron, 2022; Petriglieri et al., 2019). A recent European policy report described platform work as flexible and autonomous but simultaneously arduous (Pesole et al., 2018). To understand these findings, we draw on JD-R theory (Bakker et al., 2023). JD-R theory integrates job stress and motivational perspectives to explain how job design relates to well-being. Job design is understood in terms of job demands and job resources. Job demands refer to aspects of work requiring physiological and psychological effort, which result in strain (Demerouti et al., 2001). In contrast, job resources refer to aspects of work that aid goal achievement and foster personal growth. While job demands instigate the health-impairment process, resulting in exhaustion, job resources activate the motivational pathway, promoting vigor, dedication and absorption (work engagement). JD-R theory theorizes these processes as independent.

Jobs with few job demands and adequate job resources motivate employees but do not energize them, whereas jobs with numerous job demands but limited job resources exhaust employees (Bakker and Oerlemans, 2011). Therefore, Bakker et al. (2023) argued that optimal jobs balance regular job demands with adequate job resources to foster work engagement. We extend this reasoning by arguing that excessive job demands combined with job resources result in hyperactive work. Unlike engaging work, hyperactive work entails job resources that are prevalent but insufficient to cope with demands. Accordingly, such jobs stimulate health-impairment and motivational pathways (Figure 1). The figure illustrates different work experiences according to the activation of the motivational (horizontal) and health-impairment processes (vertical; cf. Bakker and Demerouti, 2007). The taxonomy illustrates the implications of the balance (and imbalance) of job demands and job resources. That is, we argue that work becomes hyperactive when excessive job demands and job resources characterize work. The presence of hyperactive work may explain the ambivalent findings regarding platform work and well-being. Our study focuses on four common characteristics in the JD-R and platform work literature (i.e. job insecurity, workload, job autonomy, task feedback) and three technological characteristics (i.e. technological threat, technological opportunity and technology use).

Workload encompasses the quantitative and demanding aspects of work (Van Veldhoven et al., 2015). The labor intensity of platform work may exceed that of other working arrangements for several reasons. In contrast to permanent, temporary and on-demand working arrangements, platform work requires finding adequately compensated tasks, administration and navigating between tasks, often unpaid (Cameron, 2022; Cropanzano et al., 2023; Spreitzer et al., 2017). Consequently, there is evidence that platform workers are pushed to work longer hours at a high pace with tight deadlines to secure similar pay to other working arrangements (Pesole et al., 2018; Schor et al., 2020). Moreover, platform workers spend additional time working to buffer against drops in labor demand (Ashford et al., 2018). While self-employment shares similarities with platform work, platform work is unique in its short-term and fragmented nature (Cropanzano et al., 2023). This creates constant pressure to continue working (Melián-González and Bulchand-Gidumal, 2021).

In contrast, self-employed individuals often work on longer projects, enabling a steadier pace of work and building social capital. Finally, platform work differs from other work arrangements in the pressure platforms exert to continue working or accept assignments. Because platforms aim to maximize labor supply to cater to requests for labor, they may send “economic nudges” to entice workers to increase their efforts or hours (Rosenblat and Stark, 2016) or punish individuals who decline or receive low ratings (Cameron, 2022). The latter may even result in termination. Unsurprisingly, platform workers often describe their arrangement as intense and involving long hours (Pesole et al., 2018). It is important to note that most of these findings only provide preliminary evidence for our hypotheses since they did not conduct any quantitative, comparative analyses between platform work and other arrangements. Thus, while a higher workload is plausible, the relatively high workload compared to other working arrangements is assumed but not demonstrated.

Job insecurity, another important job demand for platform workers, involves uncertainty regarding job continuity (Greenhalgh and Rosenblatt, 1984). Contracts underlying permanent and temporary employment are longer than platform work, which implies contracts for single, short-term gigs (Cropanzano et al., 2023). The lack of a (fixed) contract underlying the labor agreement means that when individuals cannot work, or platforms cannot offer work (no requesters), individuals cannot make a living. Thus, there is a lack of fringe benefits and social security (Berg, 2016; Wood et al., 2019). Compared with self-employed and on-demand workers, platform workers’ length of employment and prospects still lag behind. On-demand and self-employed workers may have reasonable expectations regarding continuing their work (e.g. seasonal work, hospitality) or plan future employment using their network. Finally, platform work is further precarious due to its reliance on technology. Technological issues, broken electronics and algorithmic restrictions may limit access to the digital labor platform and, therefore, work (Watson et al., 2021). Platform work contrasts with other employment arrangements characterized by rights, institutional protections and a formal or psychological contract beyond a single project. Indeed, platform workers often lack a “clear, available, and relevant career path” (Ashford et al., 2018, p. 27) and express concerns about their finances and stability (Cropanzano et al., 2023). Thus, the continuity of platform work is particularly uncertain. Taken together, we predict:

Job autonomy comprises the freedom, independence and discretion in work-related decisions (Hackman and Oldham, 1975). The “open employment relationship” defining platform work entails significant flexibility and freedom (Cropanzano et al., 2023; Spreitzer et al., 2017), which is a core motivation for entering the platform economy (Pesole et al., 2018). Digital labor platforms enable users to work when and wherever they want. Workers can schedule work hours, work locations and what type of work they want to accept (Ashford et al., 2018; Spreitzer et al., 2017; Vallas and Schor, 2020). Indeed, despite algorithmic control, platform workers report freedom and discretion (Wood et al., 2019). This freedom contrasts with other employment arrangements, which involve formal or psychological contracts that imply explicit or implicit obligations (Watson et al., 2021). Permanently or temporarily contracted individuals work a specified number of hours a week. Yet, even on-demand and self-employed workers have less discretion over their time than platform workers. For instance, a hospitality establishment may assume on-demand workers’ availability during holidays without explicit agreement. The establishment may expect workers to work on holidays and may refrain from scheduling them in the future if they refuse to work on holidays. Similarly, self-employed individuals may establish relationships with clients, implying project deadlines and long-term service continuity. Finally, platform work implies the absence of formal contracts. Thus, the platform work arrangement implies that workers remain more anonymous and less dependent on a single client, providing greater freedom.

Feedback refers to direct and clear information about job activities and performance (Morgeson and Humphrey, 2006). In particular, the organization of platform work provides individuals with ample access to feedback (Deng et al., 2016). Because various stakeholders constantly produce data, algorithmic management enables platforms to continuously present feedback to users in easy-to-understand metrics (Parent-Rocheleau and Parker, 2022; Rosenblat and Stark, 2016). Indeed, performance feedback is integral to platform work (Durward et al., 2020). Some platforms even offer workers virtualization of their work history and access to performance evaluations (Deng et al., 2016). To illustrate, Cameron (2022) found that “ride-hailing” drivers use mobile applications for immediate and long-term feedback on their customer service. Permanent and temporary employment arrangements may include performance reviews to discuss progress and formal feedback. Similarly, on-demand and self-employed individuals may discuss how projects were conducted. Yet, we argue platform work implies unique short-term opportunities for feedback that go beyond what other working arrangements offer. Hence, we predict:

Technologies define platform work. Therefore, we argue that JD-Rs related to technology are more prevalent among platform workers and are relatively more important for their well-being than in other working arrangements. We focus on new technology threats, new technology opportunities and new technology use. New technology threats are worries about future technological changes making current job activities obsolete (Brougham and Haar, 2018). Thus, they refer to concerns or fear appraisals that technological advancements undermine future work opportunities. Conversely, new technology opportunities reflect an optimistic view that technological advances will facilitate goal achievement. In other words, they may appraise new technologies as a job resource. Platform workers may hold both stronger negative and positive evaluations regarding new technologies than workers with other work arrangements. To illustrate, platform work often consists of simple tasks that artificial intelligence is predicted to replace first (Jabagi et al., 2019). Nonetheless, technological advances that enable platform work may spur new job opportunities (Gray and Suri, 2019). Hence, platform workers may believe new technologies will benefit them. For instance, technological advances may enable workers to perform more complex and challenging activities remotely through virtual reality (McVeigh-Schultz and Isbister, 2022). Finally, new technology use concerns interacting with technologies such as platforms, robots and algorithms to achieve work goals. Due to the nature of platform work, we predict that they use new technologies more often than individuals with other working arrangements and that these technologies are relatively more important for their well-being. We expect them to be more prevalent (i.e. higher mean-score) and be more important for well-being (i.e. larger effect size):

We propose that platform work simultaneously activates the health-impairment and motivational processes more than other work arrangements. While this may sound counterintuitive, it is important to note that the processes operate independently (Bakker et al., 2023). Thus, although the two processes tend to correlate negatively, they can co-occur (Mäkikangas et al., 2012). For instance, using latent profile analysis, Moeller et al. (2018) showed that individuals reported low burnout and high work engagement when they experienced low to moderate job demands combined with high job resources. However, work characterized by high job demands and high job resources was associated with concurrent high burnout and work engagement. Similarly, Wu et al. (2023) found that workload, coupled with high emotional support and low control, predicted greater burnout and lower work engagement. We argue a similar pattern arises more for platform work due to its hyperactive nature than for other work arrangements. Platforms may stimulate workers to overextend themselves or work irregular hours more than other working arrangements (Duggan et al., 2020; Rosenblat and Stark, 2016; Watson et al., 2021). At the same time, platform workers may also feel determined and engaged because they can choose when and how long they work (Pesole et al., 2018). The insights derived from the application regarding productivity targets and rankings may foster feelings of vigor, dedication and absorption (Duggan et al., 2020). In support of this reasoning, gig workers have been described as experiencing a world of “heightened emotions” characterized by a greater variety of emotions, more extreme emotions and more frequent oscillations between emotions (Ashford et al., 2018, p. 28). Therefore, the co-occurence of exhaustion and work engagement might be a distinctive characteristic platform work. Hence, we expect:

We make several contributions. First, we reveal how platform work differs from other working arrangements. Existing research provides important insights into platform workers’ experience (e.g. Ashford et al., 2018; Cameron, 2022; Schor et al., 2020). Yet, little is known about whether these experiences differ from other working arrangements. Our between-working arrangements approach builds upon and complements previous findings by enabling us to discern whether platform work differs from other work arrangements in JD-Rs and well-being. Thus, we answer calls for comparisons between traditional employment and platform work (Ashford et al., 2018; Cropanzano et al., 2023). Second, we reconcile ambivalent findings by contextualizing the literature with JD-R theory and proposing platform work as “hyperactive.” Our results elucidate why research has reported favorable in addition to unfavorable characteristics of platform work such as “flexible, but also intense” (Pesole et al., 2018, p. 5) and stark tensions in anxiety and fulfillment (Petriglieri et al., 2019). Finally, we investigate the relative importance of technological JD-Rs. These findings emphasize the need for novel instruments to measure the job conditions of platform workers (Caza et al., 2022) and answer calls for investigating more specific JD-Rs in the platform economy (Hakanen et al., 2019).

An independent panel company recruited participants in The Netherlands in 2021 to ensure a heterogeneous sample of various worker types. The panel company primarily sources participants through their panels, co-branded panels and opt-in databases to collect representative samples. In addition, they recruit through multiple channels such as mobile app panels, social media influencers, billboards and social networks. The company collects data on over 100 variables to create profiles. Based on these profiles, the company estimated that 80% of their panel members qualified for the study. Within the segment of self-employed panelists, 25% qualified as platform workers. Individuals eligible for the study received an invitation through email or could sign up through the panel portal. The sampling frame consisted of working adults in The Netherlands who were active in one primary working arrangement (i.e. permanent, temporary, on-demand, self-employed or platform work). We deliberately restricted the sample to individuals with a single working arrangement to ensure clear attribution of job demands, job resources and well-being outcomes to one form of work. Although this decision limits generalizability to multiple job holders, it strengthens internal validity by avoiding confounding experiences across arrangements. The study was registered and approved by the ethical review board of Erasmus University Rotterdam (number: 20-11 Oerlemans).

Two independent samples were collected one month apart. Participants first received information about the general purpose of the study, informed consent and eligibility criteria (i.e. working and above 18 years of age). To assess the robustness of our results, we tested our hypotheses in two independent samples by comparing platform work with permanent, part-time, on-demand and self-employed working arrangements. The second sample served as a replication. We only included individuals with a single employment arrangement to compare platform work with other working arrangements. Namely, multiple job holders have unique experiences beyond the scope of the current article (Campion et al., 2020; Ilsøe et al., 2021). Our analyses provide a unique insight into platform work by focusing on single jobholders.

In total, 984 individuals participated (Sample 1: N = 626, Sample 2: N = 358). Sample 1 comprised 348 men (55.6%) and 278 women (44.4%). On average, the participants were 44.4 years old (SD = 13.5) and worked 32.3 h a week (SD = 11.4). Most were cohabiting with a partner (n = 416; 66.5%) and received higher vocational (n = 211; 33.7%), post-secondary (n = 161; 25.7%), university (n = 121; 19.4%) or high school education (n = 114; 18.5%). In terms of working arrangements, participants were permanently employed (n = 390; 62.3%), temporarily employed (n = 45; 7.2%), self-employed (n = 120; 19.2%), on-demand (n = 34; 5.4%) or worked on a platform (n = 37; 5.9%). The platform workers further specified that they performed platform work on average for 19.1 h a week (SD = 14.1) and worked mostly through Workspot, Freelance or Uber (51.4%). Most of them indicated that digital platforms were important for obtaining paid work (51.4%), followed by somewhat important (21.6%) and very important (18.9%). We also included one non-compulsory, open-ended question regarding what work participants did. The majority of the sample performed hospitality, such as a shift at an event, and food (delivery) services through platforms (34.3%) or did logistics and provided transport, such as ride-hailing (22.9%). In addition, they also used platforms to freelance small tasks such as ICT, cleaning and teaching language (25.7%).

Sample 2 comprised 169 men (47.2%), 188 women (52.5%), and one individual did not identify as male or female (0.3%). The mean age of the participants was 43.26 (SD = 14.93), and the average weekly work hours were 32.30 (SD = 11.42). Most were cohabiting with a partner (n = 225; 62.8%) and received education such as higher vocational education (n = 118; 33.0%), post-secondary education (n = 97; 27.1%), high school education (n = 69; 19.3%) or university education (n = 62; 17.4%). Participants had a permanent (n = 177; 50.6%), temporary (n = 36; 10.1%), self-employed (n = 78; 21.8%), on-demand (n = 39; 10.9%), or platform worker (n = 28; 7.8%). The platform workers worked on average 20.7 h a week (SD = 7.6), and half of the sample worked through Workspot, Freelance or Uber (50.0%). Most platform workers stated that digital platforms were important for obtaining paid work (46.4%), followed by somewhat important (32.1%) and very important (14.3%). The open-ended question revealed that the majority provided hospitality services through platforms (59.2%), whereas others freelanced delivery services (11.1%), ICT, content designing, blogging (7.4%), care services and consulting (7.4%) and accountancy (3.7%).

We calculated the omega reliability for all measures (Table 1). Omega reliability (ω) estimates are generally more accurate than alpha reliability estimates because they do not assume normally distributed items or equal factor loadings for all items (Dunn et al., 2014). The reliability of all scales was excellent.

The job demands were evaluated on a five-point scale (1 = almost never, 5 = almost always). First, workload was measured with the three-item workload subscale of the Experience and Evaluation of Work instrument (Van Veldhoven et al., 2015), e.g. “Do you have to work very fast?” Second, we used the five-item job insecurity scale (Huang et al., 2012), e.g. “I am insecure about keeping my work for as long as I wish.”

Participants evaluated job resources on a five-point scale (1 = almost never, 5 = almost always). We measured job autonomy with the four-item scale of the Experience and Evaluation of Work instrument (Van Veldhoven et al., 2015), e.g. “Do you have freedom in performing your work?” Second, the three-item scale developed by Morgeson and Humphrey (2006) was adapted to measure feedback, e.g. “I receive sufficient information about the results of my work.”

We measured one technological job demand, one technological job resource and one ambivalent job characteristic related to using new technologies. First, we measured technological threats with six items based on the four-item scale developed by Brougham and Haar (2018), e.g. “worried that in the future, my current work will be performed by new technologies.” Second, we developed five items to measure new technology opportunities, e.g. “I am positive about the developments of new technologies in my industry” (1 = completely disagree, 5 = completely agree). Finally, similar to Park et al. (2011), participants read a brief description of new technologies, in our study, digital labor platforms, robots, algorithms and other new technologies, after which they rated their use of each technology (1 = not at all, 7 = all the time).

For well-being, we measured exhaustion and work engagement. Exhaustion was measured with five items developed by Schaufeli et al. (2002), including “I feel mentally exhausted from my work.” A sample statement is, “During work, I feel bursting with energy” (1 = never, 7 = always). We measured work engagement with the nine-item Utrecht Work Engagement Scale (Schaufeli et al., 2006).

We specified a measurement model to assess the construct validity of the instruments for each individual sample. We evaluated model fit with chi-square/df ratio (χ2/df), comparative fit index (CFI), Tucker−Lewis Index (TLI), root mean square error of approximation (RMSEA) and standardized root mean square residual (SRMR). Chi-square/df ratios lower than three, TLI and CFI values above 0.90 and RMSEA and SRMR values lower than 0.08 indicate proper fit (Marsh et al., 2004). Finally, we conducted Harman’s single-factor test to assess the severity of common method bias. A factor that accounts for more than 50% of the variance signifies that the common method variance is an issue (Podsakoff et al., 2003).

We tested H1–H3 and H5 by conducting analyses of variance. We calculate partial eta squared (η²) for the effect size. To assess the differences between groups, we conducted a priori simple contrasts. H4 was tested with dominance analyses for each group by including all job demands and job resources. A dominance analysis determines the relative importance of predictors by comparing predictors across all possible sub-models (Azen and Budescu, 2003). In the present study, we focus on general dominance, which refers to the average conditional contribution of a predictor. By conducting this analysis for each working arrangement, we can determine to what extent the relevance of the job demands and job resources is universal or relatively specific to the working arrangement. Finally, to test H5c, we conducted multinomial regression analyses to calculate odds ratios, which signify whether the probability of group membership increases (values exceeding 1) or decreases (values below 1). Platform workers were the reference group. Hence, values below 1 suggest that the platform workers are more likely to experience exhaustion and engagement concurrently. Similar to the analyses for H4, these analyses constitute a complementary comparative analysis that shows whether the co-occurrence of exhaustion and well-being is relatively unique to the platform compared to other working arrangements.

Table 1 presents the means, standard deviations and correlations of Sample 1 and Sample 2.

The measurement model consisted of nine latent factors: workload (three items), job insecurity (five items), job autonomy (three items), feedback (three items), exhaustion (five items), work engagement (nine items), new technology threats (six items), new technology opportunities (five items) and new technology use (four items). The hypothesized model adequately fitted the data in Sample 1 (χ² = 2,393.01, df = 909, CFI = 0.938, TLI = 0.932, RMSEA = 0.051, SRMR = 0.046) and Sample 2 (χ² = 1,886.96, df = 909, CFI = 0.931, TLI = 0.925, RMSEA = 0.055, SRMR = 0.048). All standardized factor loadings exceeded 0.60 in both samples (p’s < 0.01). Finally, Harman’s single-factor test indicated that common method variance (CMV) was not an issue in Sample 1 (21.49%) and Sample 2 (23.77%). We further tested the amount of CMV by estimating a common latent factor in each sample by constraining the loadings to be equal and squaring the common latent factor (i.e. CMV = λ²). The common method variance estimates in Sample 1 (CMV = 17.6%) and Sample 2 (CMV = 16.6%) also suggest common method bias is not of major concern in the current study. Hence, we proceeded to test the hypotheses.

To test our hypotheses, we assessed how platform work differed from permanent, temporary, on-demand and self-employed work in two independent samples (Tables 2 and 3). The differences were significant unless specified otherwise. H1 states that platform work relates to more (a) workload and (b) job insecurity than other employment arrangements. Indeed, working arrangements were related to workload [F(4, 621) = 14.79, p < 0.01, η² = 0.087]. The eta-squared (η²) quantifies the ratio of variance attributed to the predictor (between arrangements) in relation to the overall variance in the data. Planned contrasts revealed that workload is higher for platform work than for permanent (D = 0.58), temporary (D = 0.89), self-employed (D = 1.19) and on-demand working arrangements (D = 1.05). Likewise, working arrangements were associated with job insecurity in Sample 1 [F(4, 621) = 13.32, p < 0.01, η² = 0.079]. Contrasts indicated job insecurity was higher for platform work than permanent (D = 1.10), temporary (D = 0.64), self-employed (D = 0.89) and on-demand work (D = 0.74) in Sample 1. The results of Sample 2 confirmed these findings (Table 2). Hence, H1 is supported.

H2 suggests that platform work relates to more (a) autonomy and (b) feedback than working arrangements. The analysis of variance revealed a main effect of working arrangements on autonomy, F(4, 621) = 35.83, p < 0.01, η² = 0.188. Largely supporting H2a, platform work related to more job autonomy than permanent (D = 0.45), temporary (D = 0.79), and on-demand working arrangements (D = 1.41), but less job autonomy than self-employed workers in Sample 1 (D = −0.65). Sample 2 replicated these results except for the comparison with self-employed (Table 2). In partial support of H2b, there were significant differences between working arrangements in feedback [F(4, 621) = 12.93, p < 0.01, η² = 0.077]. However, platform work related to more feedback than permanent (D = 0.74), temporary (D = 0.73) and on-demand workers (D = 1.123), but they did not differ in feedback from self-employed workers (D = 0.16, p = 0.428). These results were replicated in Sample 2 (Table 2).

H3 states that platform work relates to more (a) new technology threats, (b) new technology opportunities and (c) new technology use than other working arrangements. There were substantial differences between working arrangements in new technology threats in Sample 1 [F(4, 621) = 21.60, p < 0.01, η² = 0.122]. Specifically, platform workers experienced more new technology threats than individuals with a permanent (D = 1.09), temporary (D = 0.92), self-employed (D = 1.56) and on-demand working arrangement (D = 1.34). The findings in Sample 2 also supported H3a. Considering new technological opportunities, workers differed in Sample 1 [F(4, 621) = 9.28, p < 0.01, η² = 0.56]. Platform workers reported more new technology opportunities than permanent (D = 0.34), self-employed (D = 0.81) and on-demand workers (D = 0.55), but not more than temporary workers (D = 0.25, p = 0.216). Sample 2 replicated these results but did show the hypothesized difference with temporary workers. Hence, the results mainly supported H3b. Finally, workers across different arrangements varied in their use of new technologies as expected. Specifically, platform workers reported more use of new technologies than permanent (D = 1.98), temporary (D = 1.64), self-employed (D = 2.37) and on-demand workers (D = 2.33) in both samples (Table 2). Hence, H3c was supported.

H4 specifies that technology-related JD-Rs are relatively more important predictors of (a) exhaustion and (b) work engagement for platform work than for workers with other working arrangements. Accordingly, we conducted dominance analyses, including the previously specified job demands and job resources, to compare the explained variance of the predictors for each group (Table 3). In Sample 1, technological JD-Rs were relatively more important predictors of exhaustion for platform workers (19.84%) than for permanent (12.96%), temporary (12.52%) and self-employed workers (14.12%), but not in comparison to on-demand workers (25.36%). These results were partially replicated in Sample 2, where technology-related JDRs were more important predictors of exhaustion for platform workers than all other working arrangements except temporary workers (Figure 2). Regarding work engagement, technological JD-Rs were much more important predictors for platform workers than for all other groups. Technological JD-Rs were relatively more important predictors of work engagement for platform workers (49.71%) than for permanent (9.41%), temporary (19.37%), self-employed (17.03%) and on-demand workers (5.48%). Sample 2 replicated these results (Figure 3). These findings mostly support H4.

H5 proposes that platform work relates to more (a) exhaustion, (b) work engagement, (c) and exhaustion and work engagement concurrently than other working arrangements. Indeed, there were substantial differences in exhaustion between workers in Sample 1. In support of H3a, platform workers experienced more exhaustion than permanent (D = 1.02), temporary (D = 1.31), self-employed (D = 1.58) and on-demand workers (D = 1.45). These findings were replicated in Sample 2 (Table 3). Hence, H5a was supported. Similarly, working arrangements related to differences in work engagement. Platform workers were more engaged with work than permanent (D = 0.49) and temporary workers (D = 0.53), but not self-employed (D = 0.03, p = 0.912) and on-demand workers (D = 0.38, p = 0.287). However, in Sample 2, platform work related to more engagement than all other working arrangements. Hence, H5b is only partially supported.

Finally, we conducted multinomial regression analyses to assess whether concurrently experiencing exhaustion and work engagement was indicative of platform work. Using platform workers as the reference category, odds ratio (OD) values below 1 indicate that the probability of platform work as a working arrangement increases. Indeed, the co-occurrence of high levels of exhaustion and work engagement increased the probability of platform work relative to all other working arrangements in Sample 1 (Table 4; Nagelkerke Pseudo R² = 0.11; medium effect size): permanent workers (OD = 0.81, 95% CI: 0.68, 0.97), self-employed (OD = 0.74, 95% CI: 0.61, 0.91), on-demand workers (OD = 0.68, 95% CI: 0.54, 0.87), but not temporary workers (OD = 0.81, p = 0.074, 95% CI: 0.61, 0.91). However, these odds did not replicate in Sample 2 (Table 4; Nagelkerke Pseudo R² = 0.23; large effect size). Hence, the results provide only partial evidence for H5c.

The number and importance of platform workers have vastly increased in recent years. However, knowledge regarding the implications of this work remains scarce. We know little about how platform work is unique or similar to other working arrangements (Cropanzano et al., 2023). The present article addressed this gap by adopting a between-arrangement approach. We predicted and found that platform work is characterized by a pattern of associations we recognized as “hyperactive.” Platform work was characterized by more job demands and job resources, and more ill-being and well-being. Additionally, the findings indicate that technological JD-Rs are more prevalent and relatively important for platform work than other working arrangements. Together, these findings address the objective of our research to clarify the differences between working arrangements and have several implications.

First, our research advances the literature on general platform work and technology-mediated work. We proposed that platform work is a unique working arrangement in terms of JD-Rs and well-being. Platform work is often studied in isolation from other working arrangements (e.g. Caza et al., 2022; Deng et al., 2016; Yang et al., 2026). Such research is essential for understanding what platform work entails and what experiences are typical. However, such research approaches cannot elucidate how platform work differs from permanent, temporary, on-demand and self-employment. Indeed, comparative studies that compare platform work with other work arrangements are scarce (Cropanzano et al., 2023). We addressed this gap by adopting a between-working arrangement perspective that enabled us to investigate how platform work is different (or similar) to other working arrangements in terms of job demands, job resources, exhaustion and work engagement (cf. Standaert et al., 2025). Although platform workers are often formally classified as independent contractors, their work is embedded in platform governance systems that can shape task access, performance evaluation and income volatility through algorithmic matching, rating systems and incentive structures (Duggan et al., 2020; Möhlmann et al., 2021; Rosenblat and Stark, 2016). This combination of legal independence with technologically mediated dependence offers a plausible explanation for why platform work can resemble self-employment in terms of resources such as autonomy while simultaneously exhibiting higher demands such as job insecurity and intensified effort (Kidron and Gedalya, 2026). While platform work has characteristics that resonate with on-demand and self-employment, the findings indicate that equating platform work to other working arrangements is problematic and, instead, should be recognized as unique.

Second, we contribute to the literature by reconciling conflicting findings regarding the experience of platform workers. Research has described platform work in terms of negative and positive experiences (Ashford et al., 2018; Petriglieri et al., 2019). We expand and complement previous research that has argued the ambivalency of platform work (Petriglieri et al., 2019). We argued that platform work constitutes “hyperactive” work that activates both the health-impairment and motivational processes (Bakker et al., 2023). Our findings highlight that positive and negative evaluations regarding platform work are not contradictory (Cameron, 2022; Pesole et al., 2018; Yang et al., 2026). Indeed, we found that high job demands and job resources, as well as high exhaustion and work engagement, characterize platform work (cf. Park et al., 2019). It is important to note that the current findings do not imply that hyperactive work does not exist for other working arrangements. Instead, the findings indicate that hyperactive work more consistently describes platform work than other working arrangements. The results emphasize the need for initiatives that minimize the activation of the health impairment pathway while maintaining the motivational process (Marathe et al., 2019).

Third, while new technologies define platform work, little is known about how they impact workers (Burhan, 2025). Therefore, Hakanen et al. (2019) recommended that “more attention should be paid to the specific job demands and job resources − and also to the lack of demands and resources − and their relationships with vitality at work of those working in platforms” (p. 13). We answer this call by investigating technology-related JD-Rs of platform workers and their relative importance for exhaustion and work engagement. Platform workers perceived more threats and opportunities related to technological developments and used new technologies more frequently than individuals with other working arrangements. Moreover, dominance analyses revealed they were relatively more important for the well-being of individuals who performed platform work than those with other working arrangements (permanent, temporary, on-demand, self-employed work). These findings extend prior work that has highlighted algorithmic management, digital monitoring and platform interfaces as defining features of platform work (Duggan et al., 2020; Möhlmann et al., 2021; Rosenblat and Stark, 2016) by showing that technology-related appraisals and behaviors are not only more prevalent but also more important for well-being outcomes. In doing so, our results support calls to move beyond generic job characteristics and develop platform-specific measures of job demands and resources that capture workers’ relationships with technology more precisely (Ashford et al., 2018; Caza et al., 2022). This aligns with the proposition of JD-R theory that the relevance of specific job resources and job demands may differ across occupations (Bakker et al., 2023). Our research extends this idea to specific working arrangements. The findings emphasize the need for new instruments to measure job demands and job resources that are relatively unique to the experience of platform work (cf. Ashford et al., 2018). For instance, we found that using new technologies was a relatively important predictor of platform workers’ work engagement. Finally, dominance analyses revealed an interesting, contrasting pattern. While individuals who performed platform work received more feedback than those with other work arrangements (i.e. they reported a higher mean score), for platform workers, feedback was relatively unimportant for burnout and work engagement.

The study has several strengths and limitations that merit consideration while interpreting the findings. First, we adopted a cross-sectional approach with self-reports. This implies we cannot make claims of causality and may raise concerns regarding self-report biases (Podsakoff et al., 2003). Yet, the Harman single-factor test revealed that common method variance was not an issue in both samples. Second, a strength of the present study is that we derive the findings from two samples of individuals with a single working arrangement. This approach enables us to discern the overall experiences associated with platform work compared to other working arrangements. Third, research is needed on how to manage well-being effectively. Future research may investigate how platform workers proactively alter the content of their work (job crafting; Tims et al., 2012) or change how they perform and approach the content of their work (playful work design; Bakker et al., 2020; Scharp et al., 2023). Crafting may enable workers to decrease their platform dependency (Cropanzano and Mitchell, 2005; Watson et al., 2021), whereas playful work design may help maintain engagement despite demands imposed by algorithmic management (Scharp et al., 2021).

Since the present study focused on between-arrangement differences, we did not assess within-arrangements theorized in the platform literature, such as on-location (vs online) and skilled (vs non-skilled) platform work (Duggan et al., 2020; Cropanzano et al., 2023; Spreitzer et al., 2017). Various studies indicate that remote (vs on-site) work (e.g. Charalampous et al., 2019) and job resources such as skill variety have important implications for well-being. Similarly, such work aspects may explain differences in well-being between platform workers. What might especially be interesting to investigate is the extent to which these variables relate to similar and different psychological processes as in other working arrangements, and what makes the platform work hyperactive. To investigate this, we recommend that future research compares platform work with other work arrangements that are similar in terms of being on-location (vs online) and skilled (vs non-skilled) work to investigate the extent to which the motivational (and strain) processes proposed by JD-R theory replicate (Bakker et al., 2023).

The present findings have several implications for practice. First, job insecurity was especially important for the well-being of platform workers (cf. Chin et al., 2024). Hence, developing and enforcing policies that limit the uncertain nature of platform work is imperative. For instance, some scholars have called for a new category of workers so that platform workers would qualify for certain benefits through risk-pooling or portable benefits (i.e. benefits attached to the worker instead of the employer; Webster, 2016). Second, our findings suggest that organizations, in addition to individuals, should not merely reflect on work engagement levels but also exhaustion levels. Only monitoring engagement may obscure that individuals are at risk to burn out. Additionally, government efforts may improve labor conditions by introducing new legislation that holds platforms more accountable (Georgiou, 2022). Platforms may help users use the application in new ways through user experience design and by stimulating them to join online communities. To illustrate, platforms such as Uber actively use gamification (introducing game elements) in the application to stimulate engagement (Cameron and Rahman, 2022). Additionally, supportive relations with peers in online communities, in addition to a supportive relation with the platform, reduce platform turnover intentions by promoting affective commitment (Ihl, 2024).

The present study advances knowledge regarding the nature of platform work. We built on the JD-R theory to contextualize the literature on algorithmic management. The findings highlight that the hyperactive nature of platform work is ambivalent. As Pines (1993) stated: “In order for one to burn out, one must first be on fire” (p. 40). The findings are therefore concerning. While workers endure precarious conditions, they also enjoy flexibility. They feel exhausted but are engaged as well. To ensure longevity, the findings call for platforms to develop strategies that enable platform workers to capitalize on their resources and limit their risk of burning out.

The authors want to thank the Municipality of Rotterdam, Wido Oerlemans, and GovLab010 for their support in the data collection.