Classroom 3

Understanding the nature of tasks in crowdwork platforms

Understanding the nature of tasks in crowdwork platforms Anoush Margaryan University of West London, UK Conference domain: of the three domains indicated, the paper is most closely aligned with Business Purpose and background: The paper analyses the nature of tasks in crowdwork platforms (Schmidt, 2017). Despite the rapid uptake of crowdwork, the nature of crowdwork tasks – their complexity and interdependence, the opportunities for discretion, creativity, and application of workers’ expertise they provide – is poorly understood. In particular, there is paucity of empirical research examining the nature of crowdwork tasks from the perspective of crowdworkers. This paper identifies how crowdworkers characterise the tasks they undertake on the platforms. Furthermore, we compare the perspectives of crowdworkers and ‘conventional’ employees to identify the similarities and differences in the perceived nature of tasks within these different work settings. Finally, we identify an indicative typology of crowdwork tasks. Prior research has shown that the design of work tasks is crucial for both individual and organisational outcomes: motivation, job satisfaction, well-being, professional development and productivity (Parker and Ohly, 2008). Also, previous research has highlighted the importance of social characteristics in work task design showing that interdependences built into tasks – collaboration, feedback from others and contact with beneficiaries of work – enhances workers’ motivation and performance (Morgeson and Campion, 2003). Yet a review found that many of the leading platforms lack basic task design features, for example adequate support for complex tasks or workflows, infrastructure tools to support collaboration and focus on workers’ conditions and motivational dimensions, among other limitations (Vakharia and Lease, 2013). Methodology: Using a validated scale (Margaryan et al, 2011) derived from an extant typology of knowledge work (Davenport, 2005), we surveyed 295 crowdworkers, including 260 (80%) from CrowdFlower and 35 (20%) from Upwork, to scope the tasks they undertake on the platforms. Using chi-square tests, we compared crowdworkers’ characterisations of tasks with similar survey data from 459 ‘conventional’ employees from a global company in the energy sector. We carried out a principal component analysis to identify a typology of crowdwork tasks emerging from the survey data. Findings: The most frequently selected crowdwork characteristics were ‘mostly routine’ (58%), ‘systematically repeatable’ (45%) and ‘highly reliant on my own individual experience’ (40%). Other key characterisations of crowdwork tasks included: ‘highly reliant on my personal expertise/judgement’ (37%); ‘highly reliant on formal standards’ (29%); ‘highly reliant on formal rules/procedures’ (24%); ‘improvisational and creative’ (21%); ‘lacking discretion’ (17%); ‘dependent on integration across functional/disciplinary boundaries’ (14%); and ‘dependent on collaboration’ (12%) Comparing crowdworkers’ and conventional workers’ characterisations of their work tasks we found that crowdwork is more likely to be characterised as being mostly routine (X 2 (1, N=256) = 128.16, p <.00001), repeatable (X 2 (1, N=235) = 45.91, p<.00001) and lacking in discretion (X 2 (1, N=86) = 12.99, p=.000313); but less likely to be perceived as being reliant on formal rules and standards (X 2 (1, N=277) = 33.47, p <.00001), less dependent on collaboration (X 2 (1, N=242) = 87.98, p<.00001) and interdisciplinary integration (X 2 (1, N=357) = 213.10, p<.00001), less improvisational and creative (X 2 (1, N=219) = 15.15, p=.000099), and less reliant on workers’ experience and expertise (X 2 (1, N=434) = 63.55, p<.00001). The PCA carried out on crowdworkers’ survey responses produced a four-factor solution that accounted for 53.99% of the variance. Due to low variance, the item ‘my work tasks are mostly routine’ was excluded. The following indicative typology of crowdwork tasks emerged: Cluster 1. High-agency crowdwork  Dependent on integration across functional/disciplinary boundaries  Improvisational/creative  Dependent on collaboration  Highly reliant on workers’ individual experience Cluster 2. Rule-based crowdwork  Highly reliant on formal rules/procedures  Highly reliant on formal standards Cluster 3. Low-agency crowdwork  No freedom to decide what should be done in any particular situation  Mostly systematically repeatable Cluster 4. Expert crowdwork  Highly reliant on workers’ deep expertise/personal judgment Although these findings corroborate the extant accounts characterising crowdwork as routine, systematically-repeatable and low-complexity, they also suggest that crowdwork is more nuanced incorporating elements of collaborative, high-agency and expert work. This typology should be further explored, refined and extended with a larger sample of crowdworkers. Originality/value: This research contributes new empirical data in an emergent and under- researched domain. Improved understanding of work tasks would help crowdwork platform providers and clients shape the design of tasks in ways that could be beneficial to all stakeholders improving crowd workplaces for current and future workers. References Davenport, T. (2005). Thinking for a living. Boston, MA: Harvard Business School Press. Margaryan, A., Milligan, C., & Littlejohn, A. (2011). Validation of Davenport’s classification structure of knowledge-intensive processes. Journal of Knowledge Management, 15(4), 568-581. Morgeson, F., & Campion, M. (2003). Work design. In Borman, W., et al (Eds.), Handbook of psychology, volume 12 (pp. 423-452). Hoboken, NJ: Wiley. Parker, S., & Ohly, S. (2008). Designing motivating jobs. In Kanfer, R., et al (Eds.), Work motivation (pp. 233-284). London: Routledge. Schmidt, F. (2017). Digital labour markets in the platform economy. Friedrich-Ebert Foundation, Germany. Vakharia, D, & Lease, M. (2015). An analysis of paid crowd work platforms.