A Factorial Survey on Sickness Presenteeism

1 Introduction

While there has been a vast amount of research on the predictors and consequences of absenteeism behavior, the opposite – sickness presenteeism – is still in need of further research. Sickness presenteeism, defined as working in the state of ill-health (Ruhle et al. 2019) has been found to be associated with negative consequences for both individuals and organizations. While some recent studies also point to positive aspects of sickness presenteeism, for example by avoiding a pile-up of work, reduced feelings of guilt or the maintenance of work-related self-efficacy (Karanika-Murray and Biron 2020; M. Wang et al. 2022; Lohaus et al. 2020; Biron et al. 2022), there is overwhelming evidence that working despite illness can have negative health consequences for employees (e.g. Bergström et al. 2009; Skagen and Collins 2016; Taloyan et al. 2012; Kivimäki et al. 2005; Demerouti et al. 2009) and may also entail negative economic consequences for organizations due to ego-depletion, productivity loss, contagion of diseases and increased sickness absence rates (for an overview, see Johns 2010; Rivkin et al. 2022; Ruhle et al. 2019). Nevertheless, previous studies demonstrate that sickness presenteeism is a common behavior among employees (Lohaus and Habermann 2019). However, the decision process between sickness absence and sickness presence of individuals faced with health complaints remains understudied, mostly because it is difficult to grasp empirically. With the rise of flexible work arrangements, this complex decision-making process has been extended by another option for many employees, i.e. working from home in case of illness, also called virtual sickness presenteeism (Eurofound 2020) or workahomeism (Brosi and Gerpott 2022). Virtual presenteeism is also discussed as an emerging form of “hidden sickness absence” where individuals avoid to call in sick to bypass expected negative impressions associated with sickness absence (Fiorini 2024; Ruhle and Schmoll 2021).

Previous studies often rely on cross-sectional data and convenience samples, which limits the generalizability of the results and precludes causal inference. Moreover, many studies are based on small samples drawn from specific occupational groups, such as hospital staff (Rivkin et al. 2022). Survey-based approaches typically use self-reported, retrospective data with long recall periods (e.g. up to 12 months) to capture sickness presenteeism making them prone to recall bias and thereby undermining the reliability of the results. In addition, relying solely on prevalence rates is insufficient to understand decision-making processes, as sickness presenteeism is mostly positively correlated with sickness absence, while both are strongly negatively associated with health status (Gerich 2015). Scholars have also emphasized that decisions concerning sickness absence or presence is not static but situation dependent. Individuals are assumed to seek for an “equilibrium” between performance, reputation, relational needs and health demands (Ruhle et al. 2019; Karanika-Murray and Biron 2020). To the contrary, the experimental design of our factorial survey addresses many of these limitations. It not only allows to simulate sickness-related decision-making across standardized scenarios, but also enables us to isolate causal factors that are often confounded in observational studies (Auspurg and Hinz 2015).

In light of previous research, this paper introduces a multifactorial survey (vignette design), to study the decision-making process and to identify different (individual and contextual/organizational) determinants of (virtual) sickness presenteeism. Specifically, respondents were shown different hypothetical scenarios where they wake up in the morning feeling unwell and then need to decide whether they would go to work, call-in sick, or work from home (if their job offers this possibility). Implementing this multifactorial survey as a follow-up survey of the BIBB/BAuA Employment Survey 2024 – a cross-sectional telephone survey representative of the German working population (Gensicke et al. 2024) – ensures access to a large and diverse sample of employees with many relevant background characteristics.

The selection of determinants (i.e. vignette dimensions) cover aspects that have been shown to be important factors in previous studies (for overview, see Miraglia and Johns 2016) and that can be credibly conveyed in the hypothetical scenarios, i.e. are easy to imagine for all respondents. More specifically, the dimensions cover variation in the severity of sickness and symptoms, contagiousness, attendance pressure, team cohesion, colleague’s workload, as well as the weekday and the weather. An information treatment was also considered in the main experiment, where a random sample of participants was informed about the negative health-related consequences of sickness presenteeism.

We also conducted further sub-experiments for specific groups of employees. First, we conducted a sub-experiment aimed at respondents working in projects. In this subsample, we analyze the role of increasing use of performance management practices – such as goal setting and tight deadlines – on presenteeism decisions (Miraglia et al. 2025; Rivkin et al. 2022). Second, we aim to shed light on period pain presenteeism among females, i.e. working when experiencing period pain. In all vignette scenarios we asked respondents about their associated intention to go to work, call in sick or work from home (if possible).

Additional survey-based indicators were used for some vignette situations to collect data on possible adjustments that participants may consider during their workday (such as working less hours or postpone more complex tasks) when they opt for sickness presence or virtual presenteeism. As the mere information regarding participants’ decision behavior gathered through the vignette design is not informative with respect to the underlying motives of these decisions, we included survey questions following the vignettes that ask for the relevance of different motive dimensions behind respondents’ choices – such as avoidance or approach motives (Y. Wang et al. 2018; L. Lu et al. 2013; Ma et al. 2018; Van Waeyenberg 2023).

2 Design of the Factorial Survey

We conducted three factorial survey experiments to analyze the determinants of working while sick (sickness presenteeism). In each experiment, respondents were presented with hypothetical scenarios in which they wake up feeling unwell. These vignettes varied both symptoms described and contextual factors. Respondents were then asked whether, in the given scenario, they would go to work, stay home, or work from home (if their job offers this possibility). This experimental design enables a comprehensive analysis of the determinants of sickness presenteeism as well as the underlying mechanisms behind these decisions.

Figure 1 summarizes the design of the survey, which consisted of three distinct experiments: (1) a main experiment conducted with a broad sample of respondents, (2) an experiment aimed at analyzing the role of career ambitions, targeted at highly educated respondents primarily working in projects, and (3) an experiment focused on women between the age of 18 and 50 to examine the impact of menstrual discomfort. For each of the three experiments we designed different vignettes. In experiments (1) and (2), a subgroup of respondents randomly received additional information about the potential health and career consequences of their choices. Table A 1 in the online appendix displays the generic vignettes. Each respondent evaluated 10 vignettes regarding her/his work attendance decision.

Figure 1:

Survey design.

Building on the experimental setup, the vignettes were designed in a way that the characteristics (dimensions) are statistically independent from one another in order to facilitate causal analyses. To ensure orthogonality of the vignette dimensions, we drew a D-efficient vignette sample from the full factorials (vignette universe, see Auspurg and Hinz 2015). To ensure identification of all relevant interactions, the design was optimized to maximize the orthogonality of selected interactions.

3 Survey Conduction and Survey Population

The factorial survey (Knigge et al. 2025) was conducted as a follow-up survey of the BIBB/BAuA Employment Survey 2024^[1] (Gensicke et al. 2024). The BIBB/BAuA Employment Survey is a representative telephone survey covering approximately 20,000 individuals in the labor force, jointly conducted by the Federal Institute for Vocational Education and Training (BIBB) and the Federal Institute for Occupational Safety and Health (BAuA). The cross-sectional data are collected every six years and target individuals aged 15 and older who work at least 10 h per week on a regular basis. For our follow-up survey, we included only dependent employees aged 18 or older from the main survey who had agreed to participate in a follow-up.

The survey was administered as a computer-assisted web interview (CAWI) by a market and social research institute (IFAK Institut GmbH & Co. KG). To assign respondents to one of the three experimental groups, a screening process first determined eligibility based on gender, age, education, and whether the respondent worked in projects. Eligible respondents were then randomly assigned to one of the experiments with assignment probabilities based on target sample sizes of each experiment and adjusted during the field phase. Priority was given to the main experiment, with six times as many participants assigned to it compared to the other experiments. The final realized sample size amounts to 5,106 individuals. The interviews lasted 18.7 min, on average, the median duration was 12 min.

The survey received ethical approval from the BAuA ethics committee (approval given on 30.04.2024, number: 084_2024). No external funding was received.

Table 1 compares the basic socioeconomic characteristics of respondents in the factorial survey to those in the BIBB/BAuA-ETB 2024. Sampling weights were developed to account for the study design and to closely calibrate the study data to the BIBB/BAuA-ETB 2024.

4 Data Preparation

In order to enable analyses of the vignette evaluations the responses to the vignettes were matched to the setup data on vignette level, additionally, variables were named and labeled in an instructive manner.

5 Research Potential

The factorial survey on sickness presenteeism offers various research opportunities. The three different experiments allow (causal) analysis of the factors influencing (virtual) sickness presenteeism, such as symptoms, team cohesion, attendance pressure, career ambitions, or the role of information about the health consequences of presenteeism on decision-making. A unique feature of the data is the potential to examine presenteeism behavior among women with menstrual discomfort, addressing an under-researched yet relevant phenomenon. Apart from the vignettes, the implementation of standardized questions, regarding job characteristics (e.g. commuting, work from home days, performance-related pay, commitment towards the organization) and regarding the broader context of the decision further enriches the scope of study, allowing for in-depth research applications. The data thus also allow analyzing heterogeneous effects across sociodemographic groups (e.g. gender, qualification, occupation) and stratified analyses tailored to specific research questions. In addition, the linkage with data from the BIBB/BAuA Employment Survey 2024 allows to include individual work situations and contextual characteristics. Beyond the primary focus on studying sickness presenteeism (and absenteeism), the data also support the exploration of other outcomes within the BIBB/BAuA Employment Survey 2024, substantially expanding the analytical potential of these data.^[2]

6 Data Access and Documentation

For data documentation including the questionnaire see Knigge et al. (2025) (German only). The data can be requested for scientific purposes at the Federal Institute for Occupational Safety and Health (BAuA). The data are usually made available in Stata format (.dta). For data protection reasons, the data can only be used via a guest researcher workstation on the BAuA’s premise in Dortmund. To apply for data access, please send an e-mail to the Research Data Center of the Federal Institute for Occupational Safety and Health (Forschungsdaten@baua.bund.de).