Data from the Researcher Mental Health Observatory STAIRCASE Survey

2.5.1 Demographics

First, respondents were asked to provide the career level of their primary position, following the EURAXESS researcher career stage taxonomy that classifies researchers into four career stages. In addition, the question provided the option to answer in a free-text field. Respondents who indicated that they did not hold a research position were excluded from the survey at this point. A substantial number of respondents (1.5%) answered that they were researchers but did not have a PhD degree and had no intention of pursuing a PhD degree. To account for these respondents, an additional category “researcher without a PhD degree and not pursuing a PhD degree” was introduced and the corresponding free-text answers were mapped to this category.

Respondents were then asked to provide information about their gender, year of birth, nationality, country of their main place of work, research field- and subfield, civil status, information about whether their partner pursued paid work (if not single), number of children in the household, satisfaction with their income, and year of completion of their PhD degree (for career levels R2, R3, and R4).

Information about the content of each variable contained in the demographics section of the survey, the name of the variable in the data files, the percentage of missing responses, and a short summary of the processing steps applied to each variable in the Campus Use file (CUF) is provided in Table 2.

In the second section of the survey, we asked respondents about mental health outcomes (e.g., burnout, depression, stress, anxiety, and well-being), as well as work-engagement as a primary work-related outcome, and resilience as a personal resource and an important moderator. To measure each of these constructs we used standardized and validated psychometric scales. We preferred scales that were free to use for research purposes and for which validated translations existed in as many of the survey languages as possible. Below, we briefly describe the scales used, the number of items they contain, their English language sources and the sources of the translations. If no translation is explicitly referenced, translation was carried out by members of the ReMO consortium. Information about the number of items in each scale contained in the mental health outcomes section of the survey, the name of the associated variables in the data files, the percentage of missing responses, and the response scale – both the rating scale used in the measure and its numerical mapping following the mapping suggested by the developers of the scale – are provided in Table 3.

Burnout: Burnout was measured using the 12-item Burnout Assessment Tool (BAT) developed by Schaufeli et al. (2020), with answer options on a five-point scale ranging from “never” to “always”. The instrument contains four subscales relating to exhaustion (items 1, 2, and 3), mental distance (items 4, 5, and 6), cognitive impairment (items 7, 8, and 9), and emotional impairment (items 10, 11, and 12). The authors of the BAT also provide translations for all other survey languages except Hebrew.

2.5.2 Mental health outcomes

Depression, stress and anxiety: Depression, stress, and anxiety were measured using the 21-item Depression Anxiety Stress scale (DASS21) developed by Lovibond and Lovibond (1995), with answer options on a four-point scale ranging from “did not apply to me at all” to “applied to me very much or most of the time”. Depression is measured by averaging items 3, 5, 10, 13, 16, 17, and 21 of the scale. Anxiety is measured by averaging items 2, 4, 7, 9, 15, 19, and 20 of the scale. Stress is measured by averaging items 1, 6, 8, 11, 12, 14, and 18 of the scale. We used the Spanish translation of the DASS 21 by Daza et al. (2002), the French translation by Martin (2012), the Italian translation by Bottesi et al. (2015), the German translation by Nilges and Essau (2015), and the Hebrew translation by Puls (2016).

Well-being: Well-being was measured using the 5-item scale developed by the World Health Organization – The World Health Organization-Five well-being index (WHO-5) (Staehr, 1998), with answer options on a 6-point scale ranging from “all of the time” to “at no time”. The World Health Organization provides translations of the scale in all survey languages.

Work-engagement: Work-engagement was measured using a brief three-item scale (UWES) developed by Schaufeli et al. (2017), with answer options on a six-point scale ranging from “never” to “always”. The authors of the scale also provide translations for all other survey languages except Hebrew.

Resilience: Resilience was measured using the 6-item Brief Resilience Scale (BRS) developed by Smith et al. (2008) with answer options on a five-point scale ranging from “strongly disagree” to “strongly agree”. We used the Spanish translation by Rodríguez-Rey et al. (2016), the French translation by Jacobs and Horsch (2019), the Italian translation by Recipes for Well-being (2025), and the German translation by Chmitorz et al. (2018).

2.5.3 Working- and studying conditions

In the third section of the survey, we asked respondents about their working- and studying conditions using a number of items constructed for purpose. We explicitly included “studying” conditions alongside “working” conditions in the wording of the questions after receiving feedback during testing from doctoral researchers who perceive themselves as students and do not relate to questions about “working” conditions. We describe each question and reasoning behind including it in the survey below. Information about the content of each variable contained in the working- and studying conditions section of the survey, the name of the variable in the data files, the percentage of missing responses and a short summary of the processing steps applied to each variable in the Campus Use file (CUF) are provided in Table 4.

Main and secondary place of work: First, we asked respondents about their main place of work. Recording this information is crucial, since according to the Individual, Group, Leader, Organisation, and the Outside Context (IGLOO) model (Nielsen et al., 2018) it is an important source of variance for mental health outcomes. For participants from European countries, we provided a drop-down menu with a closed list of research institutions for each country (Research Organization Registry, 2023). For respondents from non-European countries we only provided a free-text field where participants could enter the name of their institution. This choice was made due to practical reasons: providing a drop-down menu with research institutions for each country of the world would have meant including an excessive number of institutions. During testing we found out that this repeatedly caused the LimeSurvey questionnaire to become unresponsive or crash. Reducing the number of institutions included in the drop-down menus resolved the problem. Due to our geographical focus on European countries, we decided to include only closed lists of institutions there, and provide respondents from other parts of the world with a free-text field to name their institution. We also asked respondents if they had other workplaces in addition to their main place of work, after receiving feedback during survey testing that this is common for many researchers in European countries.

Contract situation and overwork: Temporary working contracts and the associated precarious working conditions in academia have been discussed as a primary reason for deteriorating mental health in researchers (e.g. Chia et al., 2024). Therefore, we assessed the contract situation of respondents by asking them about the duration of their current working contract, whether it is a tenure-track position (e.g., has the option of becoming a permanent position after fulfilling some specified criteria), or is tenured. Similarly, long working hours have been associated with unfavourable mental health outcomes (e.g., Sang et al., 2015). Therefore we asked participants to provide information about the number of working hours specified in their working contract, and the number of hours they actually work per week.

Time spent on different tasks and professional development: To gain a better understanding of the characteristics of the work performed by researchers and their relation to mental health outcomes, we asked respondents to provide information about the percentage of working time they regularly spend on research, teaching, applying for research funding, mentoring, administrative tasks, and other tasks. The different percentages provided for each task were required to sum to 100%. In addition, we asked respondents to provide information about the number of days they spent on professional development (e.g., attending conferences, courses, workshops, and training) in the last 12 months.

Research outputs: While we recognize the limited informative value of such an indicator, the number of publications is still oftentimes used by managers of research institutions to assess the performance of researchers in practice, and by research funders to assess the performance of institutions. We therefore considered it an important secondary outcome that could potentially capture the relationship between mental health outcomes, working conditions, and research outputs. However, in different research fields there are very different types of publications that are considered of value. We therefore asked respondents to provide information about the number of research outputs they published over the last 12 months in five different categories: peer reviewed research papers, peer reviewed conference articles, monographs, patents and “other”. Furthermore, different research fields also have vastly different rates of publishing. We therefore caution against the use of the reported research outputs for comparisons between different disciplines.

Intentions of leaving academia and recommending academia: Intentions of leaving academia or the willingness to recommend academia as a career are often considered important outcomes for human resource managers in institutions. We therefore assessed these two dimensions by asking respondents how often they seriously considered leaving academia in the last 12 months, and whether they would generally recommend a career in research and higher education to early-career researchers today. Both questions allowed for answers on a five-point scale ranging from “hardly ever/almost never” to “extremely often/always” for intentions of leaving academia, and from “fully disagree” to “fully agree” for the willingness to recommend working in academia.

2.5.4 Working and studying environment

Next to the more qualitative indicators characterising the working and studying conditions described in the section above, we relied on a number of validated scales to assess the working and studying environment. Again, we preferred scales that were free to use for research purposes and for which validated translations existed in as many of the survey languages as possible. Below, we briefly describe the scales used, the number of items they contain, their English language sources, and the sources of the translations to the other survey languages. Similar to the scales described in the “Mental health outcomes” section, if no translation is explicitly referenced, translation was carried out and validated by members of the ReMO consortium (see also section “Quality Control” below). Information about the number of items in each scale contained in the working and studying environment section, the name of the associated variables in the data files, the percentage of missing responses, and the response scale – both the scale used in the questionnaire and its numerical mapping following the mapping suggested by the developers of the scale– are provided in Table 5.

Predictability, role clarity, sense of community, emotional support, and instrumental support: To assess the working and studying environment in terms of its predictability, role clarity, the prevalent sense of community, as well as the emotional and instrumental support available to the respondent, we used scales contained in the Copenhagen Psychosocial Questionnaire (COPSOQ) developed by Burr et al. (2019). We used the Spanish translation by Moncada et al. (2021), and the French, Italian and German translations by Peter et al. (2022). Within COPSOQ, predictability is assessed with three items, role clarity with four items, and sense of community with three items. We adapted the COPSOQ questions for emotional and instrumental support to the academic working and studying environment, asking respondents about how often their direct supervisor, superior (e.g., department head, etc.), colleagues, administrative personnel, or partner, friends/family members or other people in their personal community are willing to listen to their problems at work (emotional support, five items) or provide help and support (instrumental support, five items). Each scale had answer options on a five-point scale ranging from “hardly ever/almost never” to “extremely often/always” for sense of community, instrumental support, and emotional support, and from “to a very large extent” to “to a very little extent” for predictability and role clarity.

Policy perceptions: We assessed the respondent’s perceptions of workplace policies for safety, health and well-being using the 7-item scale developed by Sorensen et al. (2018), with answer options on a five-point scale ranging from “to a very large extent” to “to a very little extent”.

Job insecurity: We assessed the respondent’s feeling of job insecurity using the 4-item scale developed by Vander Elst et al. (2014), who also provide a Spanish translation. The scale has answer options on a five-point scale, ranging from “fully disagree” to “fully agree”.

Job control: The feeling of control over the day-to-day work of respondents was assessed using the 3-item scale developed by Hackman and Oldham (1975), and the German translation by Ehrlich (2008). The scale has answer options on a five-point scale ranging from “fully disagree” to “fully agree”.

Interpersonal conflict: To assess the extent of interpersonal conflict present at the workplace, we used the 3-item scale developed by Færgestad (2017), with answer options on a five-point scale ranging from “fully disagree” to “fully agree”.

Work-family conflict: To assess how compatible work/studying is with care obligations of respondents, we used the 5-item scale developed by Netemeyer et al. (1996), translated to Spanish by Moncada et al. (2021), to Italian by Colombo and Ghislieri (2008), and to German by Nübling et al. (2005). The scale has answer options on a five-point scale ranging from “fully disagree” to “fully agree”.

Illegitimate tasks: We assessed the prevalence of tasks the respondents have to do at work that they feel are illegitimate, using the scale developed by Semmer et al. (2010) with a translation to Spanish by Valdivieso et al. (2021), and to German by Jakobshagen (2006). The scale has four items and answer options on a five-point scale ranging from “hardly ever/almost never” to “extremely often/always”.

Work-related stress: In addition to general stress which we assessed using DASS 21 (as described in the “Mental health outcomes” section), we also assessed specific work-related stress using a three-item scale developed by Edwards et al. (2008), with answer options on a five-point scale ranging from “hardly ever/almost never” to “extremely often/always”.

Job satisfaction: Finally, we also assessed overall job satisfaction asking respondents “All things considered, how satisfied are you with your job?”, providing a reply scale between 0 “very dissatisfied” to 10 “very satisfied”.

2.5.5 Perceived and experienced harassment

Given recent attention directed at instances of power abuse, bullying and harassment at academic institutions (see e.g., Roupas, 2025), we dedicated a relatively large part of the “Working and studying environment” section to assess the respondent’s perceptions of and experiences with harassment, as well as the perpetrators of the harassment and the perceived reason for the harassment. To this end, to assess perceived harassment we asked respondents “Have you noticed co-workers being subjected to degrading experiences or harassment at your place of work or study in the last 6 months?”. To assess experienced harassment, we asked respondents “Have you been subjected to degrading experiences or harassment at your place of work or study in the last 6 months?”.

For both perceived and experienced harassment, if a respondent answered “yes” to the question, we followed up with the question “Who was the perpetrator or the perceived/experienced degrading experiences or harassment?” which provided the non-exclusive response options “supervisor”, “coworker”, “I prefer not to say” and a free-text response field. Analysis of the free-text answers to this question yielded a high number of answers indicating students and administrative personnel as perpetrators in perceived and experienced harassment, warranting the creation of these categories which were coded from the respective free-text answers.

We also asked respondents to state the basis for the perceived/experienced harassment, providing the non-exclusive response options “gender”, “sexual orientation”, “religion”, “ethnicity”, “culture”, “hierarchy”, “age”, “disability”, “I don’t know”, “I prefer not to say” and a free-text response field.

Information about the content of each variable contained in the harassment subsection of the working environment section, the name of the variable in the data files, the percentage of missing responses and a short summary of the processing steps applied to each variable are provided in Table 6.

2.5.6 Leadership and supervision

The Individual, Group, Leader, Organisation, and the Outside Context (IGLOO) model (Nielsen et al., 2018) identifies the relationship with the respective superiors in the workplace as an important determinant of work-related mental health outcomes. In the academic context, we differentiate between “leaders” and “supervisors” and explain this differentiation to participants in an explainer text provided at the start of the survey section. Here, “leader” refers to that person who directly influences how work in a team is organized. This could for example be a group leader or department head, depending on the researcher’s position. “Supervisor” refers to a person that has a say in how research or studies are conducted. We mainly expected doctoral researchers and potentially PostDocs to have supervisors, whereas we expected all researchers to have leaders – even though in the case of leading researchers (e.g., professors) there is oftentimes no superior that exerts close influence on how work is conducted. For this reason, we also provided a “does not apply” answer option to questions in this section.

We assessed the characteristics of the leader using two validated scales and the characteristics of the supervisor with one scale, described below. Information about the number of items in each scale contained in the leadership and supervision section of the survey, the name of the associated variables in the data files, the percentage of missing responses, and the response scale – both the response scale used in the questionnaire and its numerical mapping following the mapping suggested by the developers of the scale – are provided in Table 7.

Leader demands & leader lack of care: Almeida et al. (2022) provide a 6-item scale to assess the characteristics of demands a leader has on their team, and a 4-item scale assessing the (lack of) care of the leader. Both scales have answer options on a five-point scale ranging from “fully disagree” to “fully agree” and include an additional “does not apply” item for respondents that do not feel that they have a leader they report to in their working environment.

Supervisor integrity: We first ask respondents whether they currently have a person they consider their supervisor. If this is the case, we use the 4-item scale developed by Colquitt and Rodell (2011) to assess the respondent’s perception of the integrity of their supervisor. The scale provides answer options on a five-point scale, ranging from “fully disagree” to “fully agree”. Lastly, we also ask respondents to provide information about how often they talk with their supervisor about how well they carry out their work.

Findable

The data release on the DZHW data center is assigned a globally unique and persistent identifier, e.g. DOI. The data is described with rich metadata that is associated with (i) the data package, (ii) the survey, (iii) the survey instrument, and (iv) associated publications, facilitating integration of data from future survey waves in the same reporting framework.

Data package: The metadata associated with the data package describes the study series (STAIRCASE survey), the responsible institution (TIB – Leibniz Information Centre for Science and Technology University Library) and funding agency (European Research Council), names and (where available) ORCIDs of project contributors, the survey design (cross-sectional), the survey data type (quantitative data), information about the person responsible for data curation (Jana Lasser), the current data release version, links to additional resources (e.g., project website and survey website), and the allowed use cases. It also contains a brief description of the data contained in the data package (e.g. the number of responses, demographic and working condition related variables recorded, and scales used). It also includes additional documents related to the data package, specifically the informed consent form and participant information in all available translations, and the data protection plan.

Survey: The metadata associated with the survey includes the serial number of the data collection wave, describes the researched population, sampling procedure (non-probability sample: respondent-assisted sample) and geographic coverage, the data type (quantitative data), the survey method (standardised online survey), the field period (Sep 15, 2023–Aug 26, 2024), and the net sample size (4,296).

Survey instrument: The metadata associated with the survey instrument includes the title, the language and the technical implementation (CAWI) of the instrument.

Associated publications: The associated publications will be updated continually and link to any publications originating from the published data set.

Accessible

The metadata is retrievable via the data set’s DOI and via other metadata hubs like DataCite. The CUF can be downloaded for teaching purposes (free of charge) after an informal usage request using the order function on the DOI landing page (see more detailed description in Section 3.7 above).

Interoperable

The data is provided in a number of widely used data formats, including as CSV, Stata, and SPSS data files. The metadata includes qualified references to other metadata, e.g. associated publications and (potential) future survey waves. The metadata is registered at da|ra (https://www.da-ra.de/; a DataCite registration agency) and follows the DataCite metadata standard. Further, the metadata is mapped to the DublinCore, Schema.org and DDI 3.2 metadata standards.

Reusable

The metadata provided alongside the data describes the data with a plurality of accurate and relevant attributes that meet domain-relevant community standards (see description of metadata under “Findable” and “Interoperable” above), including a clear and accessible data usage license and detailed provenance information.

Codebook

The codebook of the data set is provided as supplementary information to the article. In the codebook, we provide a description of all variables contained in the STAIRCASE data set on researcher mental health. The codebook is structured into five sections, corresponding to the sections of the online questionnaire and the subsections of section 2.5 “Materials/Survey instruments”. Each section starts with an introductory text that was shown to participants taking the survey, followed by tables that summarise the information of each variable derived from questions that were asked in the respective section. For most variables, we provide the full wording of the question text in all languages. However, due to copyright we cannot provide the wording of some items of instruments that are not published under a creative commons license. For these items, we provide the reference to the source of the question texts. However, a non-public version of this codebook containing all questions is made accessible to researchers using the Scientific Use File in the secure data processing environment.