Perceptions of COVID-19 related risks by platform-based couriers: An analysis of user comment threads on Reddit

1 Introduction

The destabilisation of formal employment and increase in entrepreneurial culture, including platform-based courier or ‘gig’ work, has given risk to new forms of risk and uncertainty for workers [1] (p. 3). ‘Gig’ work, promoted for its flexibility and autonomy [2], comes with limited to no labour protections, leading to financial uncertainty and related concerns for the health and welfare of workers [3]. As we discuss, platform-based work has increased considerably as a result of the pandemic, creating new and emerging risks for workers [3]. Job loss as the result of lockdowns and stay at home orders coupled with increased demand for ‘essential’ gig work (food and package delivery) as individuals were forced to stay at home, led to a growth in the gig workforce, particularly among already vulnerable populations [4]. The focus of the present paper is to understand, from the perspective of these workers, how COVID-19 has changed the risk landscape of platform-based courier work. Central to our discussion is the (mis)classification of platform-based workers as independent contractors as it relates to COVID-19 risk.

Platform-based workers connect with clientele through an app on their device. Clients ‘order’ products (a parcel, a ride, or a meal) via the app or online platform and the work is distributed to individuals registered with the platform host. Herein we focus on courier workers – that is, individuals who move people, food and packages from one location to another. Though this type of work offers flexibility and income in a period of significant job loss and income insecurity across developed nations due to COVID-19 [5], reports to date suggest that app-based courier workers (referred to often as ‘gig’ workers) have been adversely affected due to the nature of the work; namely, the absence of employer-provided prevention measures to mitigate risks of COVID-19 [6] and no or little protection through employer-provided or government-regulated health and safety standards [7]. At the same time, this form of work poses additional risks for exposure to the SARS-CoV-2 because these workers are highly mobile and in contact with many individuals [8].

Companies that govern platform-based couriers have introduced some limited COVID-19 related measures, though they differ by jurisdiction. For example, within Canada (the location of the present study), Uber Eats promised couriers sick pay for up to two weeks if they get COVID-19 [9]. However, workers have no control over the behaviour of clientele and may be exposed at the point of delivery to individuals who do not follow public health or other government guidance. Indeed, platform-based couriers have been identified as highly exposed, frontline, low-wage, non-unionised workers with almost no occupational health guidance or support in the context of COVID-19 [10]. Courier work and related SARS-CoV-2 exposures have already been documented (see for example, McGran [11]).

The nature of platform-based work in terms of government support schemes, coupled with the pressure to earn an income through courier work, means that platform-based couriers have a high risk of exposure, creating opportunities for contracting COVID-19. Research continues to be conducted to understand risk from the perspective of advocates for workers and the workers themselves. For example, survey data from workers across OECD countries suggest that only ∼1/3 of those surveyed said that their platform(s) had taken measures during the pandemic and only 44% of this group were satisfied with the measures taken. Further  30% indicated that it was hard to access the resources provided [7]. To date, there is limited data capturing the experiences of vulnerable platform-based workers – many of whom may be women, immigrants, and members of visible minority groups in low-income occupations [12], which may mean that they do not speak up or hesitate to share their experiences out of fear of repercussions related to job security or migrant status.

The present research draws on anonymous social media posts, publicly available, to overcome limitations related to accessing hard-to-reach populations and individuals who may hesitate to participate in interviews or focus groups. These data allowed us to hear the stories and experiences of workers not currently captured through survey data with pre-defined lines of questioning. Ultimately our aim was to examine discussions about platform-based courier work during the pandemic, to understand worker narratives related to virus exposures, risk and transmission to the public.

2 Materials and method

Social media posts, freely available through Reddit, were collected and analyzed using narrative analysis [13]. Reddit is used daily by millions around the world. Users can post (share content in the form of stories, links, images/videos), vote (to promote content up or down) and comment (provide discussion in reference to a post) in various communities organized around their specific interests [14]. Unlike social media sites, such as Facebook, that require users to provide their real names, Reddit differentiates itself as a site that values anonymity and pseudonymity [15]. Given our aim of accessing and documenting the voices of workers without hesitancy related to repercussion, Reddit provides a space for confidential sharing of information among platform-based workers to anonymously share experiences on the job.

2.1 Data collection

Data were collected by author BR, a female graduate student with experience in data collection using social media. Data were pulled from six different months to reflect risks facing platform-based courier workers at the height of the first three waves of COVID-19 in Canada, where the third wave also saw the onset of concerns over the Delta variant. For periods of higher risk, data were collected in March and April 2020 (wave 1), November and December 2020 (wave 2), and March 2021 (wave 3 and increased discussion regarding variants of concern). Data were also collected in June 2020 to reflect periods of lower risk given the decline in cases throughout Canada, as a point of comparison (e.g., were people speaking to risk in the same way despite change in actual risk?). This design allowed us to identify COVID-19 risks to workers and the population (by way of transmission) as described by workers, during critical events in the first year of the pandemic.

We collected posts from subreddits; communities where users can post about and discuss a specific subject. The Reddit search bar was used to identify subreddits for each platform. To meet our aim, it was important to identify all types of platform-based courier worker voices (ride-hail, food delivery, parcel delivery) and we therefore focused on individuals with direct experience with courier work. Author BR started by searching for subreddits specific to platform-based companies, as identified in our scan of courier companies operating as of 2020. Due to the volume of data present across all 13 identified subreddits after searching for relevant posts using preliminary search terms (e.g., >40,000), the decision was made to collect data from one subreddit per type of work (food delivery, ride-hail, and parcel delivery). The chosen subreddits were: reddit.com/r/uberdrivers, reddit.com/r/skipthedishes, and reddit.com/r/amazonflexdrivers. Skip The Dishes was selected because it is a Canadian company and thus gave us greater insight into the risks faced by Canadians specifically, the site of the current study. Uber and Amazon Flex were chosen as the largest platform-based courier firms for ride-hail and parcel delivery.

Six initial search terms identified by our team and in consultation with a library scientist, were queried within each subreddit: ‘covid’, ‘covid-19’, ‘corona’, ‘coronavirus’, ‘pandemic’ and ‘variants’. We then reviewed the first 50 posts for each subreddit for search terms to be used across the whole dataset. The final nine search terms included: ‘covid’, ‘covid-19’, ‘corona’, ‘coronavirus’, ‘pandemic’, ‘mask’, ‘PPE’, ‘vaccine’ and ‘variants’.

Each search term was entered in the search bar of subreddit and the top three posts from the specified months were manually cut and pasted into an Excel file for analysis. By selecting the top three posts, data volume became manageable (n = 2,866) for manual scraping and human (rather than machine learning) analysis, which was critical given our use of narrative analysis. If a given month did not have three posts, this was noted and any posts that were present were scraped. All data (posts/comments) were copied and pasted. See Fig. 1 for a summary of steps taken to yield the final dataset for analysis.

OPEN IN VIEWER

Fig. 1

Steps taken to yield final dataset for analysis.

Time and date submitted, number of comments and username of poster (to identify recurring posters) was collected for each subreddit post. For each comment or threaded comment (comment that follows a comment that follows an original post), time and date submitted, and username of the commenter was collected. At the time of our search, we did not come across any cases where the comments or posts were deleted; however, there were instances where the usernames were no longer available and could not be recorded. We did not use a software to identify usernames as deletion incidences were rare and usernames did not feature in our analysis, except to identify frequent posters. We also excluded content external to written content posts or threads – e.g., external links, games and videos.

2.2 Study context

We document COVID-19 cases in Canada, as per the Public Health Agency of Canada [16], given the location of the research team, though the pandemic waves were similar across North America where, based on our review of the data, the majority of Uber drivers and Amazon Flex users are located (Skip The Dishes is only in Canada). Given the inability to limit our Reddit search to Canadian posters only, we capture the voices of couriers located across North America, and likely some internationally, despite our primary location of interest being Canada whereby the research team resides. These data nonetheless provide insight into how COVID-19 has changed the risk landscape for couriers working in Canada.

The first wave in Canada took place between March 2020 and May 2020, with the peak occurring on May 3rd with 2,760 cases of COVID-19 reported. The summer months (June-August) were characterized by decreasing case counts, where the lowest case counts were on both June 26th and July 7th, with 172 cases [16]. The second wave began in October, with case counts beginning to fall slightly in February, though they remained close to double that of the first wave. The peak of the second wave took place on January 2nd, with 10,209 daily cases. The last period, March 2021, was the earliest month of the third wave and emergence of concern over SARS-CoV-2 variants. The peak of this wave occurred on April 15th, with 9,565 cases reported. See Fig. 2 for a frequency of posts in our selected time periods, by subreddit, as aligned with our four study periods of interest.

OPEN IN VIEWER

Fig. 2

Graph of number of comments across each month for selected subreddits.

2.2.1 Analysis

Narrative analysis, as discussed by Riessman [13], was used though modified for the analysis of written text relevant to our aim. Narratives were developed by analyzing the content of each post and threaded comment and by drawing on our own prior understanding of what constitutes risks for exposure and transmission. Data were organized to identify common narratives both within and across platforms, and at periods of high (beginning of waves 1, 2 and 3) and low (June 2020) COVID-19 risk in North America. Given the inability to geographically locate users (as per the anonymity of Reddit), we did not attempt to interpret data speaking to company policy or regulation; rather, we provide descriptions of what workers share as their perceptions of company/regulatory role in risk exposure or mitigation. Illustrative examples of posts/comments related to each narrative were recorded and are presented in the results. The presented narratives are relevant to the research question; meaning that the decision to include a narrative is not based on how many posters report that theme but rather, if it has direct relevance to the study aim.

The analysis process involved author BR reviewing the posts a minimum of three times to gain familiarity with the data. Authors BR and SBM then conducted narrative analysis of all data pulled from Skip The Dishes. Narratives were compared between authors to discuss consistency in approach and emerging results. Author BR then completed narratives for the remaining two subreddits which were reviewed by SBM and in very few cases, made minor changes in discussion with BR to the analysis. The results, as presented, draw on common narratives across the data in relation to our research aim. These narratives were discussed by SBM and BR before being presented to the remaining authors.

In consultation with the Human Research Ethics Board at the University of Waterloo, it was determined that ethics approval was not required. The data collected are publicly available user comments, which are themselves posted on a publicly available website with uncontrolled access.

3 Results

In the following we present three key findings. The first is the changing nature of discussions of risk across the time periods of study (high vs. low risk). The second is risk as it relates to the nature of platform-based courier work. The third is the polarization of discourse around the nature and perception of risk between commenters. Note that in the case where we attribute a quote to a post, it is the original (first) post in a thread. ‘Comments’ are those that were left in response to the original post or comments preceding. Also note that quotes are excerpted verbatim and not corrected by the authors.

4 Discussion

The aim of this paper was to explore how platform-based couriers discuss risks associated with their work during periods of high (first wave, second wave, third wave/rise in concerns regarding variants) and low COVID-19 risk.

Our findings regarding dialogue in the first wave being dominated by tips and asking for advice about how to manage risk is unsurprising, particularly given how little was known regarding the virus and transmission by officials and the general public in March 2020 [17]. Dialogue after March/April 2020 shifted from support as a community of couriers to one of polarized debate as we moved through waves of the pandemic and the political nature of COVID-19 became more salient. Discussions, for example, regarding social responsibility vs. individual choice noted in forums also dominated the media, with anti-mask and anti-lockdown rallies in favour of the latter. Echoed on the subreddits was the rhetoric around the acceptance (or not) of public health measures that were prevalent across traditional and social media at the time [18].

While platform-based workers may be defined as independent contractors, or as engaged in “platform-based self-employment” [1], this classification is contentious with advocacy groups calling for reclassification [19]. Indeed, while courier work, being public-facing, is akin to many other jobs deemed essential throughout the pandemic (though the extent to which food/parcel delivery is essential has been questioned, particularly in relation to who these essential services benefit [20]), our data spoke to the unique nature of courier work on the basis of their classification as independent contractors; that is, they are not privy to the same worker benefits, such as paid sick leave and workplace safety measures, as employees. Data were consistent with literature documenting that COVID-19 has given rise to new forms of risk for platform-based workers. Indeed, the lack of labour protections (e.g. regulated provision of protective equipment) and compensation for time off when experiencing COVID-19 symptoms, have been noted as amplifying risk for workers [3]. Within our sample, some of the couriers reliant on courier income were forced to make a choice of whether to stay home (e.g., to protect themselves and their family, and for some in consideration of reducing public transmission) or to earn an income. Across the dataset, there was no admission that commenters had gone to work while ill, though there were statements that claimed individuals might not be able to take time off work if sick due to income loss. Rather, many spoke of their concerns that they might pose a risk to the population if working while asymptomatic, identifying the role they play in community protection.

We did not identify increased risks to the public from workers working while ill, as has been documented elsewhere [3]. However, we did identify cases of couriers having to draw on credit to take sick leave, consistent with reports that low-income workers often lack savings and are unable to go without a paycheque [21]. Although workers in this study may have been eligible for government provisions (e.g., Canadian workers would have received the Canadian Emergency Response Benefit [22]), as demonstrated in the data, the booming demand for courier services [23] and the revenue generated from platform-based work attracted individuals to this work, even in early days when risks were poorly understood and measures provided by courier companies uncertain. Of concern then is how long these workers can continue to take time off when symptomatic and how long can companies can continue to evade the role of ‘employer’ while engaging in employer-like practices when it serves the interest of the company [24].

Recent data suggest that as the financial, social, physical and psychological impact of the pandemic for platform-based workers became more pronounced, platform-based companies responded to pressure to increase health and safety measures for their workers. However, without formal employment standard, there are reported gaps in their provision that continue to put workers at risk [6]. Fairwork, a network of researchers examining the working conditions and practices of platform companies, provide data from 191 platforms in 43 countries across the world suggest that platform-based companies put workers at risk; those consistent with our data are presented below:

A gap between rhetoric and reality. There is data to suggest that some platform safety measures remain rhetoric and are not actioned. For example, enforcement of cashless payments.

The criticisms of platform-based work noted above are not new – rather, the documented prioritization of clients/stakeholder, downloading management and risk responsibility to workers, and surveillance for organisational gain have only become more pronounced as organizational responses to the pandemic are documented [6]. In keeping with the tradition of offloading the financial costs of platform-based work [25], platforms place the responsibility of health and safety on government and workers, both in providing protective measures (policies and personal protection), as well as recovery in the case workers becoming ill [6]. The non-unionised nature of their work, with almost no occupational health guidance or support in the context of COVID-19 [10], places the responsibility on workers to adhere, or not, to public health guideline to reduce transmission and in many cases to cover the costs of personal protection (masks, sanitizer). Due to the nature of their classification as independent contractors, platform companies do not have obligations to provide workers with protective equipment; this classification has thus been contradicted by the provision of masks, etc. by some platform companies [26].

Our data speaking to personal choice and individual freedoms in following public health measures (e.g., removing masks between riders in the case of UberDrivers) are concerning and pose risk to individuals travelling in vehicles. Unlike employees, couriers may choose to work while symptomatic or choose not to abide by public health measures (e.g., choosing not to wear a mask between drop-off points). While courier-based companies did introduce procedures to monitor compliance with measures – for example, penalties when customers report no mask use; selfies with masks prior to starting a shift; deactivation when symptoms are noted – monitoring and enforcement are not akin to a formal place of work due to the isolated nature of the work. Some platforms also claim that they have no obligation to provide protection for couriers, only providing guidelines for the introduction of contactless services [6]. However, this is only possible with food and package delivery. Indeed, OECD data suggest that only 25% of platforms surveyed (n = 64) reported providing personal protective equipment or hygiene products to workers (and some workers were dissatisfied with the quality of items provided). OECD data also suggest that only 23% of platform-based workers reported providing full or partial pay for sick or self-isolating workers, generally up to a maximum period of two weeks [7].

In terms of trends in the frequency of discussion, March and April 2020 had the highest number of posts, even across waves, for all subreddits. Notable however, was the continued dialogue on UberDrivers across all months, including the period of ‘low risk’. UberDrivers had many posters who self-identified as American in their posts. In June, there continued to be elevated case counts of COVID-19 across summer months [27] which may explain the continued dialogue within this subreddit. Another explanation might be that UberDrivers has the largest number of couriers across those in our study, with 3.9 million monthly active drivers [28]. The marked decline in discussion after the first wave more broadly might be explained by the normalization of COVID-19; that is, as officials and the public started to adjust to life in COVID-19, forums for support/tips became less desired but instead served as a platform for complaints and voicing of opinion. Polarization between couriers regarding the nature of their work – as independent contractors – is also notable. Users took two sides – one sympathetic to companies as they are not employers while the others advocating for employee conditions.

5 Conclusions

Our study is the first to document risks, from the perspectives of anonymous couriers who may be unwilling to share their honest opinions and thoughts through primary data collection where anonymity is not guaranteed. From a policy perspective, data highlight the increased risk to platform-based couriers as a consequence of their status as independent contractors.