Theoretical background
There are no doubts about relevant criteria that high-stake selection tools need to meet. Most essential elements to determine the quality of a measurement are widely seen as the reliability and validity of a given measurement (Michell, 1999). However recent development in this area has focused and included User Experience characteristics (Garrett, 2011) and identified those as key acceptance criteria for successful administrations.

New Perspectives/ Contributions
This presentation will share light on the rise of user experience in the recruitment space for assessment tools, specifically applicant acceptance. Previously, this has mostly been targeted from an bipolar (psychometric properties vs UX) perspective, however there are design thinking approaches that can be applied to incorporate both. The presentation will help the audience to get a general understanding of where and how user experience techniques are used already, how they interact with reliability and validity and finally outlooks of where the industry might develop towards.

 
Research/ Practical Implications
The interaction between Candidate Experience and validity has started to become more prominent in recent studies, however much more insights can potentially be gained in this area. This presentation is aimed to empower both researchers and practitioners to not see validity and Candidate Experience as opponents but characteristics that can improve each other.
Originality/ Value
The presentation is setting the scene for the upcoming sessions, providing an historic overview of Candidate Experience and its current place in the recruitment industry as well as tangible guidelines how to incorporate Candidate Experience elements in the development of assessments. To our knowledge, this has not yet done before in the detail of this symposium.

References
Michell, J. (1999). Measurement in psychology: A critical history of a methodological concept. Cambridge University Press.
Garrett, J. (2011). Elements of User Experience: User-Centered Design for the Web. New Riders Press, Berkley.

Research goals and why the work was worth doing
The goal of this research is to investigate to what extent selection practices in small, medium-sized and large organizations in Germany are in line with the quality requirements of the dedicated norm for personnel selection, DIN 33430, to find reasons for non-adherence to the standard, and to get an overview of instruments currently in use.

Theoretical background
In the area of selection and assessment, there has been a huge gap between scientifically proven standards on the one hand and procedures and instruments used in practice (Rynes et al., 2002), and it seems that the gap has been widening during the last two decades because HR professionals find it difficult to identify selection myths (Fisher et al., 2021). For example, HR professionals often decide which tests and questionnaires to use in a selection process based on the test provider’s references or experience or based on self-tests rather than on quality criteria such as objectivity, reliability, and validity (Kanning, 2019). Amongst the reasons are lack of familiarity, resource constraints, and disbelief in the usefulness of the practices (Terpstra & Rozell, 1997).

Design/Methodology/Approach/Intervention
HR professionals in charge of selection processes in small, medium-sized and large organizations of various industries is invited to complete an online survey. The link is distributed in a snowballing system through the authors and through the business media LinkedIn and Xing. Participants are asked questions regarding a typical selection process they conducted during the last year. Questions are, for example, whether and how requirement analysis and evaluation were conducted, which types of instruments were included in the selection process, based on which criteria tests and questionnaires were selected (if applicable), which standards were used when designing and implementing interviews and assessment centres and reviewing application documents, whether and to what extent those involved in the selection process were trained, which decision criteria were used, and how they communicated with candidates. The process of gathering data just started. Average completion time is 8 minutes. There is no compensation for participating in the survey, but participants can enter a lottery where they can win one of several books on personnel selection.

Results obtained or expected
The data collection process just started and will, because of summer break, run until the end of September. Data will be evaluated in October. The results will shed light into the above-mentioned questions. In cases where organizations do not comply with selection standards, there will be reasons for this: is it because they are not aware of the standard, because they do not have the resources for complying with it, or because they do not see the necessity of complying with it. Moreover, it will provide an overview of selection instruments used (various online and paper-pencil tests and questionnaires, various types of partly technology-mediated interviews, assessment centres). As a follow-up, a qualitative study diving deeper into the reasons is planned.

Limitations
The study is conducted in Germany only, therefore it will not be possible to make inferences for other countries. Moreover, participants are acquired through the authors’ direct contacts and business media, therefore it will be difficult to assume representativity for all recruiting organizations in Germany. Finally, although participants are asked to describe a past recruitment process, socially desirable responding cannot be excluded.

Research/Practical Implications
The study provides a current overview of selection instruments and practices in Germany. Since it asks for the reasons for non-adherence to certain standards, it will be possible to address the underlying issues in a more targeted manner, something that the newly founded non-for-profit organization “recruitingrebels” where scientists and HR professionals get into dialogue is intending to do.

Originality/Value
The study not only provides a recent overview and compares it to past studies, but it also gives reasons for non-adherence to standards.

References
Fisher, P. A., Risavy, S. D., Robie, C., König, C. J., Christiansen, N. D., Tett, R. P., & Simonet, D. V. (2021). Selection Myths. A Conceptual Replication of HR Professionals’ Beliefs About Effective Human Resource Practices in the US and Canada. Personnel Psychology, 20(2), 51-60. https://doi.org/10.1027/1866-5888/a000263.
Kanning, U. P. (2019). Was das Marketing verspricht –Wie Testverfahren vermarktet werden. Personalmagazin, 3, 68-71.
Rynes, S.L., Colbert, A.E., & Brown, K.G. (2002). HR professionals’ beliefs about effective human resource practices: Correspondence between research and practice. Human Resource Management, 41, 149–174. https://doi.org/10.1002/hrm.10029
Terpstra, D.E., & Rozell, E.J. (1997). Why some potentially effective staffing practices are seldom used. Public Personnel Management, 26, 483–495. https://doi.org/10.1177/009102609702600405

Research goals and why the work was worth doing
The goal of this research is to investigate the criteria on which HR professionals base decisions on the use of selection instruments. In particular, the impact of information on validity and applicants’ perception on the intention to use a selection instrument is tested.

Theoretical background
Psychometric theory stresses the importance of validity for selection instruments (Nunally, 1978). Utility models that calculate the monetary benefit of highly valid selection instruments (Holling, 1998) support this idea. However, valid selection instruments can only achieve monetary benefit, if selected applicants accept a job offer. Research into applicants’ reactions (Gilliland, 1993; Konradt, Warszta & Ellwart, 2013) has linked applicants’ perceptions of selection instruments to outcomes such as pursuit intentions. Research also found large variance in applicants’ perceptions of different selection instruments (Hausknecht, Day & Thomas, 2004). Nowadays, many economies experience problems in terms of skills shortage that have shifted market power on the labor market towards applicants. Therefore, the applicants’ perceptions may also play an important role in deciding which selection instruments to use.

 
Design/Methodology/Approach/Intervention
A sample of n=102 HR professionals from various organizations is surveyed using a 2x2 policy-capturing design. Participants read the descriptions of four selection instruments. These descriptions each contained experimentally varied information on the validity (low vs. high) and applicants’ perceptions (low vs. high) of these instruments. Participants rated whether they would use the instrument in a selection procedure.

Results obtained or expected
According to the hypotheses, hierarchical linear modelling shows that both variables, validity and applicants’ perceptions, impact on HR professionals’ usage decisions. However, validity shows a stronger impact than applicants’ perceptions. Finally, the instrument that was described as highly valid and high on applicants’ reactions received strongest usage intention ratings.

Limitations
Results were obtained from a scenario-based design. Therefore, external validity should be established through replicating the results in field studies.

Research/Practical Implications
This study shows that validity is the dominant criterion for HR professionals. However, in times of skills shortage, applicants’ reactions also play an important role in usage decisions of selection instruments. Therefore, assessment developers should ensure and provide information on both, psychometric soundness as well as favorable applicants’ reactions.

Originality/Value
This is – to our knowledge – the first study that examines HR decisions on selection instruments.

References
Gilliland, S. W. (1993). The perceived fairness of selection systems: An organizational jus-tice perspective. Academy of Management Review, 18, 694–734.
Hausknecht, J. P., Day, D. V., & Thomas, S. C. (2004). Applicant reactions to selection procedures: An updated model and meta-analysis. Personnel Psychology, 57, 639–683.
Holling, H. (1998). Utility Analysis Of Personnel Selection An Overview And Empirical Study Based On Objective Performance Measures, Methods of Psychological Research Online, 3, 1, 5 – 24.
Konradt, U., Warszta, T., & Ellwart, T. (2013). Fairness perceptions in web-based selection: Impact on applicants' pursuit intentions, recommendation intentions, and intentions to reapply. International Journal of Selection and Assessment, 21, 155-169.
Nunnally, J.C. (1978) Psychometric theory. 2nd Edition, McGraw-Hill, New York.

Research goals and why the work was worth doing
The aim of this study was to investigate if certain characteristics of psychometric assessments such as length and type have an impact on candidate’s fairness perceptions as well as perception of organizational attractiveness.

Theoretical background
Next to being valid, psychometric assessments’ impact on candidate perceptions becomes increasingly important. While having been a research topic for a couple of decades, companies using psychometric assessments for their personnel selection only started to acknowledge recently that bad candidate perceptions indeed affect job-acceptance decisions, commitment, and even performance as proven in various studies (Ambrose & Cropanzano, 2003; Hausknecht et al., 2004; Konradt et al., 2013). While knowing that there is hardly any selection method that is more valid than psychometric assessments across all job levels and types (Hunter & Hunter, 1984; Schmidt et al., 1981), we also know that they are among the least accepted methods by candidates (Hausknecht et al., 2004).
A couple of factors are proven to generally impact candidates’ perceptions in a personnel selection process. Carpenter (2013) showed that candidates prefer short and timely processes over long ones. Since most cognitive assessments are designed to put candidates under pressure using time constraints, whereas personality inventories are usually not timed, it can be assumed that personality inventories are viewed more favorably. Hauk et al.’s (2018) research concludes that older generations usually have more difficulties with web-based technology, which could lead to a higher likelihood that older generations perceive web-based assessments as less fair than paper-and-pencil tests following Gilliland’s (1993) model.

Design/Methodology/Approach/Intervention
A sample of n=116 (102 after cleaning) voluntary participants were surveyed online using a 3x2 policy-capturing design. Participants read different scenarios of a selection process that were presented to them in random order. Each description included two experimentally varied variables: length of the assessment (10 minutes vs. 60 minutes) and the type of assessment (cognitive online assessment vs. personality only assessment vs. paper-and-pencil assessment). Participants indicated their fairness perception of the selection method, and organizational attractiveness of the selecting company.

Results obtained or expected
As hypothesized, online assessments are viewed as fairer (t(610) = -3.60, p < .01) and the selecting company as more attractive (t(610) = -4.72, p <. 01). However, contradicting our previous assumptions, candidates rated longer assessments to be fairer (t(610) = -8.93, p < .01) and the selecting company as more attractive when they used longer assessments (t(610) = -4.81, p .05 for fairness and r = .04, p > .05 for organizational attractiveness).

 
Limitations
External validity is a standard concern of policy-capturing designs, as it is for this study. Additionally, only a few participants were over 50 years of age, potentially impacting the results for the age-related hypothesis additionally to a potential method bias (participants who are willing to complete an online survey might be more in favor of technology in general).

Research/Practical Implications
Generally, online assessments are perceived more positively by potential candidates, implying that selecting companies should favor those over paper-and-pencil assessments when thinking about including psychometric assessments in their personnel selection process. While candidates prefer short selection processes in general, they still favor longer assessment (batteries) over short ones. Selecting companies should therefore watch out to not select assessments that take too little time, or combine several assessments with a short test-time.
According to our results, companies do not need to worry about more negative perceptions in older generations. However, this needs to be verified in the field especially as more senior people tend to apply for more senior roles, where psychometric assessments become less typical.

Originality/Value
Previous research focused more on candidates’ perceptions of psychometric assessments in general but has not taken the potential impact of specific characteristics like length and type into account. This research provides a first insight on what exactly impacts these perceptions of psychometric assessments in selection settings.

References
Ambrose, M. L., & Cropanzano, R. (2003). A Longitudinal Analysis of Organizational Fairness: An Examination of Reactions to Tenure and Pro-motion Decisions. Journal of Applied Psychology, 88(2), 266–275.

Carpenter, L. (2013). Improving the candidate experience. Strategic HR Review, 12(4), 203-208.

Gilliland, S. W. (1993). The Perceived Fairness of Selection Systems: An Organizational Justice Perspective. Academy of Management Review, 18(4), 694–734.

Hauk, N., Hüffmeier, J., & Krumm, S. (2018). Ready to be a Silver Surfer? A Meta-analysis on the Relationship Between Chronological Age and Technology Acceptance. Computers in Human Behavior, 84, 304-319.

Hausknecht, J. P., Day, D. V., & Thomas, S. C. (2004). Applicant Reactions to Selection Procedures: An Updated Model and Meta-Analysis. Personnel Psychology, 57, 639–683.

Hunter, J. E., & Hunter, R. F. (1984). Validity and utility of alternative predictors of job performance. Psychological Bulletin, 96(1), 72–98.

Konradt, U., Warszta, T., & Ellwart, T. (2013). Fairness Perceptions in Webbased Selection: Impact on applicants pursuit intentions, recommendation intentions, and intentions to reapply. International Journal of Selection and Assessment, 21(2), 155–169.

Schmidt, F. L., Hunter, J. E., & Caplan, J. R. (1981). Validity generalization results for two job groups in the petroleum industry. Journal of Applied Psychology, 66(3), 262–274.

Research Goals and Theoretical Background
According to Cerasoli et al. (2018), informal workplace learning (IWL) consists of "non-curricular behaviors and activities pursued in service of knowledge and skill acquisition that take place outside formally-designated learning contexts. Such activities are predominantly self-directed, intentional, and field-based." (p. 204). Typical informal learning behaviors include applying one's own solutions to work problems, reflecting on work processes, sharing experiences with colleagues, and seeking feedback on one's own work results (Decius et al., 2019).
Because of the high relevance of IWL, research recently called for a theoretically sound scale to operationalize this kind of learning (Cerasoli et al., 2018; Jeong et al., 2018). Decius et al. (2019), presented a newly developed scale for measuring IWL, building on the Dynamic Model of Informal Learning by Tannenbaum et al. (2010), and extending this four-factor model to an eight-factor model (Octagon Model of Informal Workplace Learning). Their scale contains eight factors with three items each and allows a differentiated assessment of IWL.
With reference to test economic considerations, Decius et al. (2019) also presented a short version consisting of eight items that covers each factor of the Octagon Model with one item. The scale was developed for the target group of blue-collar workers. For the important occupational group of white-collar workers, who account for most of all employees in western countries (van Horn & Schaffner, 2003), there is still no empirical validation available.
In this study, we thus examine whether the IWL short scale by Decius et al. (2019) for blue-collar workers is valid and reliable for use among white-collar workers. We focus on criterion validity and construct validity, considering factorial and discriminant validity.

Methods and Design
We collected two samples: 747 (45.3% male, 54.7% female) and 3,134 (49.3% male, 50.7% female) employees from white-collar professions in Germany on voluntary basis. We compared two competing models: Model A consists of the eight items that load onto the general factor IWL. Model B contains—as an additional level—the four factors Experience/Action, Feedback, Reflection and Intent to Learn from the frameworks of Tannenbaum et al. (2010) and Decius et al. (2019). Two items from the eight-item short scale are assigned to each of the four factors in Model B.
Using Sample 1, we tested criterion validity, applying a CFA: We modelled the relationships of job demands, job autonomy (decision latitude), knowledge/skill acquisition, age, and IWL. Using Sample 2, we applied CFA multi-group comparisons to test construct validity, i.e., measurement invariance among male and female white-collar workers, and among different school types and educational backgrounds.

Results
Model A did not provide a satisfactory fit and failed to meet the minimum requirements (e.g., CFI = .874 in sample 1 / .847 in sample 2). Model B proved to be superior and yielded a good fit (e.g., CFI = .923 / .955). The reliability of the IWL scale (Sample 1: ɑ = .76, ω = .78; Sample 2: ɑ = .85; ω = .86) is rather satisfactory—especially if one considers the scale’s heterogeneity in terms of content. CFA results also show convergent and discriminant validity.
Using sample 1, a CFA revealed the following latent correlations of IWL with other constructs (in brackets are the meta-analytical corrected population correlation from Cerasoli et al., 2018, as comparison values): job autonomy .51 (.30); job demands .24 (.13), knowledge/skill acquisition .63 (.41); age -.11 (-.07). All relationships are significant (p < .001; except for age, p = .003). Using sample 2, the results provide evidence for measurement invariance across gender and educational level, and partially across school types. Limitations, Implications and Value Our results show that the IWL short scale is a reliable, valid and economic instrument for white-collar workers. To our knowledge, our scale is currently the only short scale that operationalizes IWL as a multidimensional construct based on a theoretically sound model, including behavioral, cognitive, and motivational IWL components. The study’s main contribution is therefore to provide a broadly applicable scale to overcome the fragmentation of IWL research, and to promote IWL in organizations, as part of a comprehensive personnel development strategy (Bell et al., 2017; Kraiger & Ford, 2021). Our validation results additionally strengthen the nomological network of IWL (Cerasoli et al., 2018). Limitations refer to lack of investigation of predictive validity (further studies could address this using longitudinal designs) and potential common method variance in case of validation constructs. Furthermore, the study was conducted with German employees; a validation of the scale’s English version is still pending. References are available from the authors.

Introduction
The covid-19 pandemic shows how changes in the environment and society put organizations under great pressure. Abruptly organizations had to reorganize their work processes. Complex challenges such as the covid-19 crisis ask for teamwork and some teams are more effective in dealing with these challenges and related innovations than others. That was not different in higher education. Being able to work on innovations, as a teacher team, requires team innovative work behaviour (IWB).

IWB can be defined as: “a multi-stage iterative process in which employee behaviour targets the exploration of opportunities, idea generation, idea promotion, idea realization and the sustainable implementation of these ideas, processes, products or procedures within a role, a group or an organization, whereby the ideas are (relatively) new and intended to benefit the relevant unit of adoption” (Messmann and Mulder, 2012; Lambriex et al. 2020).

Traditionally four phases of IWB are distinguished, opportunity exploration, idea generation, idea promotion, idea realization. Lambriex et al. (2020) found that idea realization could be divided in two sub-dimensions criterion-based implementation and learning-based communication and added a fifth phase: idea sustainability, consisting of two sub-dimensions, internal embedding (ensuring that the innovation is securely anchored in the organization) and external dissemination, i.e. the distribution and broader application of the newly developed ideas including networking.

The different phases of team IWB were mostly studied through survey-based, cross-sectional studies (Wildmann & Mulder, 2018), focusing on the first four phases. Given the cross-sectional design, these studies did not consider the premise that IWB phases are iterative and fluid. Also, given the quantitative approach these studies did not unravel which types of innovative work behaviours are taking place in the various phases. Therefore, this study wants to study the five phases of IWB in a longitudinal, qualitative manner. We expect to get a more detailed insight into how IWB occurs in the various phases. It gives also more insight into the dynamic interplay between the various IWB phases and their sequencing. More specifically, the following two research questions are formulated:
1.How does IWB emerge within the different innovation phases?
2.How do the different IWB phases develop over time?

Method
A case study was conducted involving a teacher team working at a Dutch university that recently introduced a new educational model for all bachelor’s degree programs. This team had to design, develop and execute an educational module for Communication Sciences in the 3rd period of the 1st year. The team was followed for 8 months, during 19 team meetings. Each meeting took approximately 1,5 hours. All meetings were audiotaped, transcribed and analyzed. A codebook was used, based on the definitions of the five main dimensions of IWB and based on the items of the validated IWB questionnaire (Lambriex, et al., 2020). Next, meaningful segments were selected and coded in atlas TI (10% double-blind). The Cohen’s Kappa was .80.

Results
The phases of the IWB process follow each other closely and the process is quite linearly. In general there is a lot of attention for Idea Generation and Idea Realization and in addition there is little to no attention for Sustainability. In line with theory, specific innovative work behaviours were identified per phase. Co-occurrence of certain behavior is observed within both the Idea generation and the Idea realization phase. First, critical analysis or evaluation were observed followed by new ideas, presented in line with the analysis or evaluation. Two new behaviors were detected that did not occur in the original codebook i.e. implementation-oriented idea generation and promotion.

Limitations
This study focused on one case study, thus it is hard to generalize the results to teams in other settings. Furthermore, the audiotapes only focused on scheduled meetings. Consequently, informal contacts between team members were not included in the analysis. Future research could use socio-metric badges to include both formal and informal contexts in the analysis.

Implications
This study gives insight into which kind of behaviour is fruitful for innovation and in which phase and thus which behavior should be fostered to increase the success of educational innovations in teacher teams.

Originality/Value
Through a longitudinal qualitative approach this study unraveld the concept of IWB further, by studying both the relationship between the different phases as well as the kind of innovative behaviour that is taking place in every phase.

References
Lambriex-Schmitz, P., Van der Klink, M. R., Beausaert, S., Bijker, M., & Segers, M. (2020). Towards successful innovations in education: Development and validation of a multi-dimensional Innovative Work Behaviour Instrument. Vocations and Learning, 1-28.

Messmann, G., & Mulder, R. H. (2012). Development of a measurement instrument for innovative work behaviour as a dynamic and context-bound construct. Human Resource Development International, 15(1), 43-59.

Widmann, A., & Mulder, R. H. (2018). Team learning behaviours and innovative work behaviour in work teams. European Journal of Innovation Management.

Research goals
As teamwork in organizations increases, so does interdependence – within teams as well as across team boundaries. Exchanging knowledge between teams becomes more and more essential to solve complex work challenges. At the same time, diversity within work teams increases. Next to a range of positive consequences, work team diversity can also spark subgroup formation and with this, negative consequences such as conflict. We want to shed light on the link between within-team variables such as perceived subgroups and between-team knowledge sharing.

Theoretical background
According to social identity theory, team members derive part of their identity from their team membership. In doing so, team members are assumed to form a mental prototype of the team including norms for how work is done in this team. We suppose that perceived subgroups within a team could make this mental prototype more complex due to each subgroup possibly having its own team norms. A more complex team prototype could lead to less psychological safety within the team because team members must consider which team norms to follow in a specific situation (the subgroup norms vs. the overall team norms). Thus, we hypothesize that subgroup perceptions within a team are negatively related to psychological safety (H1). Based on social exchange theory, we assume that low psychological safety within a team could lead team members to expect the costs of sharing knowledge with other teams to be higher than the benefits of doing so. This might be because psychological safety in other teams could be low as well increasing the possibility of negative reactions to behaviors such as knowledge sharing. This could result in team members refraining from sharing knowledge with other teams to avoid the associated social exchange costs. Thus, we hypothesize that within-team psychological safety is positively related to between-team knowledge sharing (H2) and that psychological safety mediates the relationship between perceived subgroups and knowledge sharing (H3).

Methodology
We initiated three two-wave studies to test our hypotheses. Study 1 had a multilevel design and involved 85 employees nested in 29 teams as well as their leaders from one large industrial organization. The leaders evaluated knowledge sharing at t2. Knowledge sharing was operationalized as the transfer of knowledge in one direction, from members of the team into other teams. Study 2 and 3 targeted employees from different organizations to rule out possible organization-specific explanations in Study 1. In both studies, employees rated their team’s knowledge sharing. Study 2 further included organizational transactive memory as a control variable. Study 3 further included knowledge acquisition to examine bidirectional knowledge sharing (from members of the team into other teams and vice versa). The sample in Study 2 consists of 193 employees, data collection for Study 3 is still under way. Influences of the COVID-19 pandemic are controlled for in all studies.

Results
Employing path and structural equation models, we found negative associations between perceived subgroups and psychological safety within teams as well as positive associations between psychological safety and knowledge sharing between teams in Study 1 and 2. The mediation was only evident in Study 2. Including organizational transactive memory as a control variable did not diminish the relationships. Results for Study 3 will be available in July 2021. Thus far, this lends support for H1 and H2 and partial support for H3.

Limitations
Our studies are not without limitations. Our measures of knowledge sharing were subjective and might not reflect the amount or quality of knowledge that was objectively shared. We do not know what kind of knowledge was shared, thus leaving room for untapped variance. We encourage future research to examine knowledge sharing objectively and shed light on the shared content. Studies 2 and 3 were single level studies, thus only allowing us to assess one team member’s perspective at a time. Larger multilevel studies could help investigate the possible role of within-team variance in more detail.

Implications for Research and Practice
Our results indicate that within-team variables such as perceived subgroups and psychological safety make up an important and untapped category of predictors of between-team knowledge sharing. Human resource development practitioners wanting to improve knowledge sharing between teams in their organization could start by avoiding subgroup perceptions and promoting psychological safety within teams.

Originaility/Value
What happens within the team does not seem to stay within the team. To meet increasingly complex work challenges, it seems more important than ever to break down "silo thinking". This article sheds light on a crossroad of within- and between-team processes that is crucial for interdependent collaboration.

Research goals and theoretical background
Learning in the workplace is more important than ever for creating a skilled workforce and remaining competitive in a constantly changing organizational environment. There is now a vast literature that contributes greatly to our understanding of how workers learn and develop. Yet, there are multiple approaches when it comes to learning – from formal approaches such as training and coaching to newer approaches like informal learning. The literature is separated into silos where research concentrates on a single learning approach, rather than investigating them in conjunction. This is unfortunate, because these research streams focus more on what makes these learning approaches unique rather what they have in common.
We therefore aim to bridge those research silos and take stock of what is known so far with the help of a combined narrative review and secondary meta-analysis. More specifically, we identify which approaches to learning in the workplace have been investigated until to date, how they are defined, and which outcomes they can achieve. We categorize learning outcomes on Kirkpatrick’s and Kraiger’s evaluation criteria to synthesize what these learning approaches have in common and where they target different learning outcomes. Moreover, we quantitatively analyze their effectiveness to answer the question of which learning approaches are potentially more effective than others and for which outcomes. Importantly, we have a look at the factors that may explain their differential effectiveness, such as individualization, formality, and virtuality. This may contribute to the same discussions across the research silos and may help to draw more holistic conclusions on learning the workplace, its effectiveness, and how it can be supported.

Method
To identify literature to include in our secondary meta-analysis, we searched Web of Science Core Collection. We identified and checked 621 potential articles for inclusion. We incorporated primary meta-analyses that investigated at least one learning approach, in relation to the workplace, and were written in English. For instance, we excluded primary meta-analyses that solely relied on student samples or when learning was not the goal of the workplace intervention. In total, we included 152 primary meta-analyses. All studies were published in peer-reviewed journals, between 1989 and 2021.

Results
The results show that research on learning in the workplace investigates a plethora of learning approaches. Most studies analyzed training, followed by other approaches such as coaching and mentoring. Many studies did not prespecify the learning approach and investigated workplace interventions with learning components, such as burnout prevention interventions. The included meta-analyses varied greatly regarding the specificity of the learning approaches. Where some primary meta-analyses included a wide array of different studies within one learning approach (such as executive and life coaching combined within the coaching category), other meta-analyses focused on specific target groups (predominantly the health professions) or specific goals (for instance, social skills or error management training). We identified common questions across the learning approaches, such as the influence of virtuality (f2f, blended, or virtual) and individualization (e.g., individual or group setting). Regarding the effectiveness of the different learning approaches and the factors that may explain them, the final results of the quantitative analyses will be presented in January.

Limitations
First and foremost, the included studies are often correlational in nature and we thus cannot make causal assumptions. Second, the number of primary meta-analyses vary by learning approach with most studies referring to training. A more balanced number of studies would be helpful to represent the full range of learning approaches in the workplace and would be beneficial for conducting more fined-grained analyses in the future.

Implications and originality
We deliver an integrated perspective on more than 30 years of research on learning in the workplace and build bridges between the specific learning approaches, such as training, coaching, and informal learning. Despite the differences between the learning approaches, our study shows that they share commonalities and researchers strive to answer the same questions, namely if these learning approaches are effective, why, and how their effectiveness can be supported. We hope to inspire researchers to utilize what is known in neighboring learning approaches to answer shared discussions on questions such as virtuality, that will likely shape the future of learning in the workplace. Our findings eventually help organizations, HRD professionals, and employees navigating through these learning approaches and help them make informed decisions.

Background. Concerted attention among organizational scholars on the topic of job insecurity arguably began with the publication of Greenhalgh and Rosenblatt’s (1984) seminal article. While this was a “defining moment” among I/O and OHP researchers, interest in job insecurity and its antecedents and consequences has been apparent for many decades within other disciplinary literatures including public health, economics, and sociology. Over the past several decades, the literature on job insecurity has grown immensely, yet has also often appeared fragmented and occurring within disciplinary, measurement, and even geographic silos. Despite these apparent “trends,” to date there has been no empirical investigation into the content domains and trajectories of job insecurity research over time. Yet, understanding where our field has been and how it has developed historically can give us valuable insights into the strengths and weaknesses of our scientific approaches and provide a path forward for fruitful future research.
Method. The authors conducted a literature search on several academic databases in order to identify papers that may be considered relevant. This process resulted in 845 unique entries.
Results. The authors implemented a textual analysis using the Reinert method (Reinert, 1986). This process identified seven clusters, representing different classes of meaning, which as a whole represented 96.52% of the abstracts. Each cluster was interpreted on the basis of the words clustered by the software and a reading of the abstracts that featured such words. For ease of interpretation, the authors named each cluster with a descriptive label and analyzed its trajectory over time. The first software iteration divided the clusters in two groups: one referred to public health research and the other to everything else. The second iteration further divided the clusters in two more groups: one referred to macroeconomic themes and the other broadly referred to organizational psychology and organizational behavior. The third iteration split the OP/OB clusters in two subgroups: psychometrics and more content-specific clusters. The former encompasses the development and conceptualization of measurement scales to assess job insecurity, scholars’ concern with reliability and validity of such measures, and to some extent the job insecurity nomological network. This cluster has diminished over time, reflecting an initial interest in and need for developing job insecurity measures in the early 2000s. A further iteration split the OP/OB content area into three distinct lines of research, reflecting the so-called stressor perspective (Sverke et al., 2002), relying on Conservation of Resources theory (COR; Hobfoll, 1989). This cluster reflects research done on organizational outcomes, safety-related outcomes, and statistical techniques used to investigate such relationships. This cluster has evolved over time, showing an increased scholarly activity and interest in this approach to job insecurity, driven mainly by US scholars. Another cluster reflects scholars’ effort to develop a model that could explain job insecurity correlates. This cluster has been relatively stable over the past 20 years, noting a constant (although limited) interest in the theoretical development of job insecurity as a construct. The last cluster represents the so-called social exchange perspective (Ashford et al., 1989) and the transactional theory of stress (Lazarus & Folkman, 1984) in job insecurity research. This cluster featured research on individual outcomes and different conceptualizations of job insecurity, reflecting a stable and substantial interest in the social exchange perspective, mainly driven by European scholars.
Originality/Value. The current study’s textual analysis approach provides a unique synthesis of the primary research clusters directions associated with scholarly research into job insecurity over the past 20 years and across several distinct scientific disciplines. This is especially valuable in light of the future research directions this paper might stimulate.
Research Implications. First, this synthesis highlighted definitional issues in job insecurity research: that is, different lines of research conceptualize job insecurity differently, even within the same disciplinary framework. Future research should attempt to understand how these definitional differences have impacted research. Related to this, it would be interesting to investigate whether different definitions produced different measurement scales. Third, all of the research represented in the clusters relied on quantitative methods, indicating a lack of qualitative research in this area. This is in spite of several calls for scholars to use more qualitative methods. Fourth, we believe scholars should attempt to integrate different research and disciplinary lines, when possible. We suggest that a multilevel framework could be effective in incorporating subjective perceptions of job insecurity within the context of macroeconomic and other relevant contextual variables (e.g., culture, income).

Purpose: Over the past decades, considerable research attention has been devoted to the consequences of quantitative and - to a lesser extent - qualitative job insecurity. Along with the consistent findings on the detrimental effects of each dimension on employees’ health and wellbeing, a debate has arisen over how those two components are interrelated. The current paper presents a longitudinal assessment of the relationship between quantitative and qualitative job insecurity and their simultaneous effect on a range of outcomes.
Theoretical Background: The Job Insecurity Integrated Model (JIIM) (Chirumbolo et al., 2017) is a mediation model which suggests that quantitative job insecurity precedes and directly affects qualitative job insecurity, which then leads to job strains and employees’ withdrawal reactions. We contrast that model with an equally plausible reverse relationship, where qualitative job insecurity directly affects quantitative job insecurity. Based on Jahoda’s Deprivation theory and the Conservation of Resources theory, we construct a theoretical research model, in which we argue that quantitative and qualitative job
insecurity affect one another overtime forming a bidirectional cyclic relationship. We thus propose a two-way mediation model, which embodies the theorised reciprocal relationships between the dimensions. We test their direct (overtime associations) and indirect (mediated through the other dimension) effects on a range of outcomes, classified as: job strains (exhaustion, emotional and cognitive impairment), psychological coping reactions (work engagement, job satisfaction, turnover intention) and behavioural coping reactions (in-role/extra-role performance, counterproductive behaviour).
Methodology: We use a three-wave panel design (6 month-time lag) with data collected across one year, between September 2017 and September 2018. Data were collected by means of an online survey using a nonprobability sampling method. The final sample consists of 2003 Flemish employees. To address the research questions, we conducted cross-lagged panel modelling.
Results: The results showed that the dual-mediation model had the best fit to the data. However, whereas qualitative job insecurity predicts an increase in quantitative job insecurity and the outcome variables six months later, quantitative job insecurity did not affect qualitative job insecurity, nor the outcomes over time.
Research/Practical Implications: The study demonstrates the importance of qualitative job insecurity not only as a severe work stressor but also as an antecedent of quantitative job insecurity. Our results suggest that the impact of qualitative job insecurity on employees and organizations is more severe when compared to the impact of quantitative job insecurity. Furthermore, in contrast with the previous cross-sectional research, the longitudinal examination of the relationship between quantitative and qualitative job insecurity indicates that an increased level of qualitative job insecurity is associated with an increased level of quantitative job insecurity over time.
Future Research and Limitations: For the current analysis we used a variable-centre approach, which assumes that the associations between the variables to be the same for everyone in the sample. While it is a clear limitation for this study it also presents an opportunity for the future research. More specifically it is recommended to examine the time-specific relationship between quantitative and qualitative job insecurity, while simultaneously controlling for the individual differences in the initial levels and trajectories of change.
Originality/Value: Our study contributes to the literature in three ways. First, we examine the simultaneous longitudinal effects of quantitative and qualitative job insecurity on a wide range of outcomes, which adds to the understanding of the relative importance of each dimension of job insecurity. To our knowledge this is the first study to control for the effects of one dimension of job insecurity while estimating the effect of the other dimension. We thus obtain more robust estimations of the effects of each dimension of job insecurity. Secondly, we longitudinally re-examine the Chirumbolo et al. JII model and extend it with a theoretically grounded reverse relationship between quantitative and qualitative job insecurity. Finally, we address the limitations of previous cross-sectional research by implementing a three-wave longitudinal research design, which allows to control for the previous levels of the outcome variables and to examine the temporal order in the mediation processes.

Theoretical background: Psychological job insecurity climate (PJIC) has been defined as a set of individual descriptions of worries and concerns regarding the future of one’s job shared by the employees within a common organizational context. Targets of PJIC perceptions may be either quantitative (e.g., potential job loss) and qualitative (e.g., potential loss of valued job resources and features). Previous research highlighted important implications of PJIC for employee health and adjustment (Låstad et al., 2015), and some evidence showed that PJIC perceptions are rather stable over time and partially shaped by individual perceptions of job insecurity (Låstad et al., 2016). Despite these findings, some important research questions regarding different aspects of the PJIC validity remained substantially unexplored.
Purpose: First, it is still unclear whether PJIC perceptions capture trait-like and state-like characteristics of individual responses, and how these different components may affect the actual construct meaning and PJIC measurement quality within organizational settings. Second, the unique information provided by one PJIC dimension (e.g., quantitative) over the other (e.g., qualitative) has been only investigated in terms of cross-sectional convergence among these two latent facets, which does not allow to distinguish among their ephemeral and stable characteristics. Third, trait-like and state-like correlates of PJIC perceptions have not been clarified. With this regard, as per other psychological facet-specific climates, PJIC perceptions have been mainly interpreted as outcomes of cognitive processes and mechanisms. However, there are reasons to believe that PJIC perceptions may be affectively charged (James & Tetrick, 1986) and they may display strong association with both stable and ephemeral components of job affect (James et al., 2008).
Methodology: Using a short lagged (i.e., one month) two-time points research design (T1=385 employees, T2=299 employees), the present paper was aimed at addressing the aforementioned research questions. Specifically, we administered at both measurement occasions a popular measure of quantitative and qualitative PJIC (Låstad et al., 2015) and a shortened version of the Job-related Affective Well-being Scale (JAWS, Van Katwyk et al., 2000).
Results: Results suggested that PJIC is a rather trait-like construct, although it showed a relevant proportion of within-individual fluctuations (state variability) across the two waves. Second, the stable component of PJIC perceptions may be meaningfully decomposed into quantitative and qualitative features of this construct. Noteworthy, a large part of trait-like features of PJIC was shared with the stable component of job affect. Finally, within-individual fluctuations of PJIC and JAWS were significantly and positively associated, suggesting that PJIC and job affect may be associated not just in terms of stable individual differences, but also within the same individual over time.
Theoretical and practical implications: First, results clearly indicated that PJIC and job affect are strongly associated both between and within employees. This evidence may suggest that the PJIC-job affect association is not limited to stable (chronic) characteristics concerning the relationship between employees and their work environment, but it applies also to their momentary (short-lagged) changes. Second, PJIC perceptions assessed with the Låstad et al. (2015) measure capture both trait-like and state-like features of the PJIC construct, and this instrument allows for a meaningful distinction of four components of total PJIC scores (i.e. trait, specific, state and measurement error components), with a consistent predominance of general and specific (quantitative and qualitative) trait-like features influencing the total score. Third, practitioners that may want to measure PJIC should be aware that quantitative and qualitative PJIC perceptions may provide different (but complementary) information and that these perceptions depend on both stable and transient characteristics concerned with the relationship between employees and their surrounding work environment.

Research goals. To prevent job insecurity and its negative consequences, individuals, organizations, and society as a whole would benefit from strategies to cope with job insecurity or, preferably, strategies to minimize the development of job insecurity in the first place. While previous research on coping with job insecurity has increased our understanding on how to minimize the negative consequences of job insecurity (e.g., Menéndez-Espina et al., 2019), very little is known about whether and how workers can minimize the experience of job insecurity itself. The current study aims to address this gap by investigating whether proactive coping can function as a tool to manage the experience of future job insecurity. Additionally, we examine whether proactive coping behaviors that are expected to minimize job insecurity can also function in a reactive manner to buffer the negative consequences of job insecurity.
Theoretical background. Our proactive coping hypotheses are built upon Aspinwall and Taylor’s (1997) conceptual framework of proactive coping. Proactive coping refers to future-oriented coping that tries to detect and proactively manage stressors before they can fully develop and can be divided into five components: Recognition, initial appraisal, preliminary coping, elicitation and use of feedback, and resource accumulation. These components are argued to diminish the development of potential stressors. In the current study, we have translated the first four of these conceptual components into proactive coping behaviors that may be beneficial in the context of potential job insecurity. Consequently, we expected that the amounts of weekly career planning, scenario-thinking, career consultation, networking, and reflection relate negatively to future job insecurity. The fifth component of proactive coping, resource accumulation, denotes that the more resources one possesses, the likelier it is that one will be successful in the other components of proactive coping. We examined resources as a between-person moderator variable with the expectation that workers with a higher amount resources, are more successful in their proactive coping behaviors than workers with a lower amount of resources. Additionally, we explored whether the behaviors discussed to formulate our proactive coping hypotheses, may in fact also be useful in a reactive manner by moderating the relationship between job insecurity and psychological strain. We tested both this reactive coping model and the prior discussed proactive model for four types of job insecurity: Cognitive quantitative job insecurity, affective quantitative job insecurity, cognitive qualitative job insecurity, and affective qualitative job insecurity (cf. De Witte et al., 2010; Jiang & Lavaysse, 2018).
Design. We collected weekly survey data through five online surveys targeting workers from different sectors and occupations. The final sample comprised of 266 participants of which 248 filled in all five surveys (93.2%). The resulting final dataset consisted of 1.280 weekly surveys. We measured demographics and resources at baseline and we measured the amount of proactive coping behaviors and experienced job insecurities weekly.
Results. Multilevel path modelling results showed that weekly proactive coping was generally related to an increase in job insecurity instead of the expected decrease. That is, career planning, scenario-thinking, career consultation, networking, and reflection were all positively related to job insecurity at the within-person level. Notably, we found this positive effect of proactive coping for all types of job insecurity (i.e., cognitive quantitative insecurity, affective quantitative insecurity, and affective qualitative insecurity), except for cognitive qualitative job insecurity. For the latter, we found the expected negative effect of most of the proactive coping behaviors on job insecurity. Additionally, resources did not moderate the relationship between weekly proactive coping and job insecurity for any of the coping or job insecurity types. Regarding our exploratory question, we found that generally none of the coping behaviors moderated the relationship between any of the job insecurity types and psychological strain.
Limitations. Our study was limited to the five-week timespan of our data collection. To investigate our proposition that proactive coping may reduce the development of job insecurity in the long term, more within-person research with longer time lags is necessary.
Research/Practical Implications. Proactive coping may have adverse effects in the short term due to consumption of resources, but beneficial effects in the long term due to gain of new resources. Therefore, it is advisable not to cease proactive behaviors to prevent temporary discomfort, but to keep engaging in proactive behaviors while trying to minimize discomfort (e.g., by emotional coping or recovery strategies).
Originality/Value. By using a within-person longitudinal design this study increased our understanding of job insecurity as a malleable process.

2.Theoretical background
Many organisations regularly claim that they want to be diverse and inclusive (Jonsen, Point, Kelan, & Grieble, 2019; Ozbilgin, Jonsen, Tatli, Vassilopoulou, & Surgevil, 2015), and hiring more diverse and inclusive staff is an obvious starting point to achieve that goal (Derous & Ryan, 2019; Ozkazanc-Pan, 2019). Using artificial intelligence (AI) in hiring is often heralded as way to create more diversity and inclusion because it has the potential to remove human bias (Raisch & Krakowski, 2020; Strange, 2018). The assumption is that AI is automatically increasing diversity and inclusion because traditional human biases no longer play a role.

However, despite the promise that AI can reduce bias in hiring, AI may replicate and amplify such bias and embeds it into technology. It has been shown that AI repeats and magnifies biases, which is called algorithmic bias (Benjamin, 2019; Noble, 2018). Automated systems give the impression that everyone is treated the same, which removes the bias of the individual manager, but these systems are designed by humans and bias is thus embedded in technology itself (Pasquale, 2015). Instead of bypassing bias, AI streamlines discrimination by embedding it into technology (Benjamin, 2019; Pasquale, 2015). Society and technology is mutually constitutive (MacKenzie & Wajcman, 1999), which means that many biases that exist in society will be embedded into AI. The purpose of this presentation is to explore how inclusion can be embedded in AI for hiring by developing the notion of algorithmic inclusion. This is central for the ethical use of AI.

3.New Perspectives/Contributions
The presentation explores exclusion and inclusion in AI-supported hiring by focusing on three interrelated areas: data, design and decisions. Data is a central requirement for AI-supported hiring and is key for replicating and avoiding bias. Data can be exclusionary and it is suggested that the notion of fit, categories and intersectionality require special attention. AI is designed by a homogenous group of individuals, and approaches to inclusive design are still in infancy. The predictions rendered by AI create various challenges for ethics and monitoring in relation to decision-making. The presentation contributes an understanding to how algorithmic inclusion in AI-supported hiring can be created. Algorithmic inclusion means to ensure purposeful design and use of AI-supported hiring.
4.Research/Practical Implications
This presentation poses some significant questions for further research. In regard to design, it needs to be explored how categories can be formed to allow for greater inclusion and how intersectionality can be embedded in AI. It is also important to question how notions of fit are perpetuated and potentially magnified through the use of AI in hiring. Further research should also explore how design methodologies for inclusive AI design might look and how they can be adopted and enhanced by organisations. There are many open questions in regard to decisions, such as how hiring managers interact with AI to ensure that hiring is inclusive. The presentation suggests that there is an urgent need to conduct empirical research on the use of AI in hiring. While other research has called for more quantitative research (Köchling & Wehner, 2020), there is a particular need for qualitative and particularly ethnographic research to analyse the design and use of technology. Such approaches are common in science and technology studies (Hess, 2002; Woolgar, 1991). In depth, ethnographic case studies would be ideally suited to analyse how algorithmic inclusion can be created.

The presentation helps practitioners in two ways. First, it provides practitioners with an understanding of how exclusion and inclusion may occur with regard to AI. Second, it outlines ways in which practitioners can become involved in building more inclusive AI. Practitioners should also be centrally involved in assessing the impact of AI over time and correcting any exclusionary practices.

5.Originality/Value
The presentation shows how AI used for hiring is amplifying existing biases, unless data, design and decisions are attuned to spotting those biases. The implication of the article is research needs to go beyond documenting algorithmic bias by focusing on how inclusion can be created through data, design and decisions to ensure that AI used for hiring is ethical.

Theoretical background
In the last few years the digitisation of assessment globally has proceeded at a rapid pace which was only accelerated by the pandemic situation. Almost all assessment providers deliver their testing in online and digital format, and many of them are developing Artificial Intelligence (AI) add-ons and functionality. AI is a lively topic of conversation in the global media and is the subject of intense legal, political and ethical scrutiny. In April 2020 the European Commission presented regulatory text to protect citizens from harmful AI, through a requirement for organisations to demonstrate safe use and informed decision making. Those breaching the regulation could be subject to fines of up to 6% of their annual turnover. The financial and reputational risks of getting AI wrong are huge and our profession has arguably been too slow in joining and contributing our perspectives to the debate. We have exercised our usual, and often justified, caution and strong commitment to ethics, but is this allowing technology firms to lead on innovating in AI without our valuable input?
This paper is delivered by a team of psychologists who provide advice to clients on a range of assessment technologies and are increasingly being asked to give opinion on whether to “switch on” the AI functionality that some assessment providers offer, and make recommendations for providers who specialise in AI. The team do not own or sell any proprietary assessment products but instead work with a wide range of assessment providers and partners to construct end to end processes that use technology wherever appropriate. As such they have a unique and objective view of the range of products on offer.
The advice that is given to clients relating to AI is often in the context of a keen focus on, and targets around, Diversity, Equity & Inclusion (DEI). Clients are often very concerned about the risks, referring to familiar cautionary tales of AI going wrong – the much-publicised Google algorithm horror story, the recent exam marking fiasco in the UK, HireVue’s facial recognition functionality etc. Further, public sentiment towards AI indicates that perspectives on the use of AI range from being sceptical to outright hostile. Surveys and polls support this - in 2018 the Royal Society of the Arts in the UK commissioned YouGov to poll people on attitudes to the use of Automated Decision-Making Systems (ADS), and 60% of respondents opposed it’s use in recruitment and criminal justice.
This paper takes that stance that Work and Organisational Psychologists have a role in calming the debate and opening a more reasoned conversation to consider how best to use AI and to contribute their thinking to the development of new solutions. With increasing volumes of candidates in some markets, skills deficits in some areas, and fluctuating pipelines of talent, many Talent Acquisition (TA) functions are starting to see the potential benefits of increased automation, in particular the inclusion of automated CV screening, candidate ranking, and on-demand video interviewing with AI functionality to support hiring decisions. They want to attract candidates into their recruitment processes, assess them quickly to retain them in the process, give them a positive branded experience, and select them fairly – and AI promises to be able to help deliver this. But clients are also incredibly wary of what is on offer and the potential damage it could do to the reputation of their businesses if it goes wrong.
Research /Practical Implications
To enhance their understanding of the concerns and perceptions in their own client base, the speaker and her team are conducting a small-scale qualitative review, interviewing and surveying a range of clients about their attitudes, perceptions and concerns in relation to the use of AI in selection and assessment. This work will “represent” the client concerns anonymously as a means of ensuring a balanced perspective on the issue in the symposium.
Specifically, the qualitative research sets out to define client perceptions and perspectives in the following areas:
•How do they define AI? What do they think AI is?
•What ADS and AI functionality have they considered?
•What are their primary concerns about AI? What are their biggest barriers?
•What feedback have they had about AI from candidates and hiring managers?
•What advantages do they see for using AI?
•What would increase their confidence that AI in assessment was safe and beneficial?
•How ready is their organisation for AI? What would need to change internally?
Though not all of this research has been completed at the time of this submission, the final paper will represent clients’ views to a practitioner and academic audience and, combined with the perspectives of the other symposium contributors, provides a coherent picture of the current landscape for AI. The speaker will also describe how best to interface between clients and providers to ensure the appropriate and ethical use of AI and how to support them to manage risks and issues and ensure appropriate outcomes, particularly in respect of DEI. They will share case studies and concepts that can be used to inform practice and make suggestions for areas of further research for academics.

Artificial intelligence (AI) and its subset, machine learning, have been a hot topic in employee selection in recent years. AI-based tools have great potential in automating and optimising the employee selection process. When developed or used inappropriately, however, AI-based tools can produce bias.

Human decision-making biases are well documented. The selection process is no exception (Knight, 2017). Some researchers have demonstrated how applicants with an accent can be rated less favourably by interviewers than the applicants without an accent (Segrest et al, 2006). In CV screening, it has been found ethnic minority applicants can drastically increase their chances of being offered an interview if they ‘mask’ their ethnic name from CVs (Decelles et al, 2017). By explicitly taking such sources of bias out of the decision-making process, well-designed AI tools have the potential to help reduce human biases in many situations. At the same time, increasing evidence suggests that AI tools can also embed human and societal biases and amplify them at scale. For example, an AI-based CV screen tool, developed on a dataset consisting of mostly males, was found to be producing male-favouring results (Dastin, 2018).

With AI-based selection tools being increasingly used by organisations, I/O practitioners must consider pragmatic ways in which machine learning can be applied to reduce biases in employee selection and crucially consider how human intervention can prevent biases from being baked into AI tools. These approaches can be grouped into the following three areas.

Getting your data right
The fairness and accuracy of any AI-driven selection tools depend on the quality of ‘training’ data – the dataset upon which algorithms and models are developed. Training data should be diverse and include as many relevant demographic groups as possible. The dataset should be multidimensional, with predictors, outcome variables, and information on protected characteristics being accurately collected. Additionally, human intervention is required to mitigate or remove biases if they are found to be evident in the training data.

Debiasing your algorithm with a fairness lens
In the machine learning process, computers learn patterns and relationships from data, and iteratively derive different algorithms, before arriving at a statistically optimal model. This process offers a great opportunity for human involvement to safeguard fairness. A machine learning process may pick up and subsequently make use of variables that are illegal to use in selection practices. or variables that act as proxies for protected characteristics such as ethnicity, gender, and age. Without human judgement, it would be extremely difficult for computers to determine whether a relevant predictor should be removed from the algorithm. Human judgment is also required in deciding the optimal predictor weighting and any trade-off between validity and fairness.

Testing and monitoring
As a minimum, machine-learned algorithms should be tested using a testing dataset that is separate from the training data. The purpose of this is to ensure that the algorithms are stable enough to be used in real-world scenarios. Where possible, it is often desirable to pilot the algorithms alongside human decisions. This offers unique opportunities to examine possible discrepancies and improve the performance of the algorithm.

A worked example
As part of a research project, we developed an algorithm that automatically scores video interviews as an aid to support recruiters who then make the final decision.

The data (n=3284) consisted of pre-recorded video interviews from real job applicants. The videos were rated by trained human raters, and candidates’ gender and ethnicity information were also available. The dataset was split into two, training and test data, with appropriate sample distribution.

In the first step, the videos were transcribed into text so that the machine learning process can focus on what the candidates said, therefore eliminating the potential impact of gender, ethnicity, accent, or non-verbal cues such as facial expressions.

In step two, a set of predictive models were derived based on the training dataset. The models were compared and reviewed by I/O psychologists. Variables contained in the resulting model were further scrutinised based on their level of prediction and fairness. Variables that are highly connected with gender and ethnicity were removed from the model.

The resulting model was tested on the testing dataset in step three where fairness and accuracy were validated. As can be seen, the algorithm has resulted in improved fairness against ethnicity.

Adverse Impact Ratio
(4/5th rule)
Human rated video interview score AI-based video interview score
Female/Male96%97%
Asian/White79%109%
Black/White70%102%
Mixed/White113%103%

Conclusion
AI systems can be designed and applied ethically to reduce bias and alleviate the inefficiencies of the selection process. With AI playing an increasingly important role in society, AI bias mitigation will become increasingly crucial. We believe fairness should be treated as one of the central success criteria in any AI-based selection system design, not as a by-product or afterthought. More broadly, integrating humans and AI decision-making offers the optimal approach, as the role of AI is to assist human decision-making and not replace it.

Research goals and why the work was worth doing: Flexible working hours are becoming increasingly available to many workers. In 2017, nearly 7 out of ten employees in Sweden had some influence over their working time. This number has probably further increased, not least because of the coronavirus pandemic. Generally, high work-time control (WTC), a special form of work-time flexibility, seems to have favourable effects on health and work-life balance. However, some studies found that very high flexibility at work can lead to blurred boundaries between work and private life. Men seem to be especially at risk of working more overtime hours with higher WTC, whereas women report improved work-life balance. However, these findings might be explained by the fact that men and women often work in different occupations and consequently in different work environments. Here, we examined associations between WTC and subsequent overtime, work-life interference and exhaustion in a sample of knowledge workers and tested if gender moderates the mediating role of overtime.

Theoretical background: Two main mechanisms have been suggested regarding buffering effects of WTC on health. First, WTC might benefit recovery from work, both within and outside of working hours, by facilitating the taking of breaks when needed and the scheduling of work to fit current recovery needs from strains at work or in private life. Second, WTC buffers against work-life interference by allowing better alignment of work and non-work responsibilities, which in turn is associated with better health outcomes. Although these theories predict that WTC promotes health overall (or prevents ill-health), some findings point towards a difference between men and women in the extent to which they benefit from WTC.

Methodology: The sample contained 2248 Swedish knowledge workers drawn from the Swedish Longitudinal Occupational Survey of Health. Employing hierarchical, multiple regression modelling, we examined the effects of the two sub-dimensions of WTC, control over time off/daily hours, on subsequent overtime hours, work-life interference and exhaustion in general and gender-stratified samples. Using conditional process analysis, we tested whether gender moderates the indirect effect via overtime on work-life interference and exhaustion.

Results: Control over time off was related to less work-life interference for women (β=-0.259; 95% CI: -0.397 – -0.122) and lower exhaustion for both genders (βmen=-0.207; 95% CI: -0.357 – -0.056; βwomen=-0.232; 95% CI: -0.407 – -0.057). Control over daily hours was not associated with work-life interference or exhaustion, but with increased overtime among men (β=0.617; 95% CI: 0.149 – 1.084). Men worked on average 42 minutes/week more overtime than women. We could not confirm that gender moderated the indirect effects of control over time off/daily hours on work-life interference and exhaustion, via overtime. Independent of gender, effects of control over time off on work-life interference were partly explained by working fewer overtime hours.

Limitations: Despite many strengths, the study also has some limitations. First, sample participants worked about three hours of weekly overtime, which may have been too low to actually observe stronger effects on work-life interference and exhaustion, particularly given that we included part-time workers in the sample. Second, we cannot rule out bias due to unmeasured confounding such as personality traits. Third, SLOSH data involve selection processes of participants, particularly for those responding several times, which may introduce selection bias. Fourth, attrition and missing data may have added a degree of bias to our results; we tried to limit this by utilising full-information maximum likelihood (FIML) estimation for data missing at random. Fifth, data used are mostly self-reported which are subject to common-method bias and measurement error.

Practical implication: Results presented here show that control over time off might have particularly beneficial effects: more control over time off was related to less work-life interference (particularly for women) and lower exhaustion. Based on our results, employers seeking to improve male or female knowledge workers’ work-life balance and counteract feelings of exhaustion should focus on increasing individual control over when not to work (taking breaks, running private errands during work and scheduling vacation/other leave) while at the same time preventing employees from further increasing their working hours.

Originality: Previous research indicates that men with higher levels of WTC tend to work more overtime hours while women appear to benefit in terms of work-life balance, but increase total workload/domestic labour. These results may however be biased by gender-segregated occupational sectors with differing opportunities to self-determine working hours and work overtime. We surmounted these shortcomings by focusing on a homogenous occupational group in a relatively egalitarian society.

Research goals and why the work was worth doing
More and more employees no longer have rigid and fixed working hours set by their employers, but can control them to a certain extent themselves. Several studies have shown that work-time control (WTC), which is an employee’s autonomy regarding how much and when to work (Val-cour, 2007) is positively related to employees’ health. However, little is known about how WTC leads to improved health, because only few studies investigated underlying mechanisms of this relationship. Some of these studies showed employees’ work-home interference to mediate the relationship of WTC and health. To further understand this relationship, and more specifically the relationship of WTC and exhaustion, this study aims to investigate internal work-to-home interference (IWHI) as a mediator. IWHI is a sub-dimension of work-home interference that focuses on employees’ psychological preoccupation with work during off-job time, such as thinking about work related problems (Carlson & Frone, 2003). In addition to previous studies, we also considered the other direction of interference, namely internal home-to-work interfer-ence (IHWI).

Theoretical background
According to job demands-resources theory (Bakker & Demerouti, 2017), WTC can be seen as an important job resource that can buffers adverse effects of job demands on employees’ health. We thus assumed WTC to reduce employees’ exhaustion. Since WTC enables employees to bet-ter align (conflicting) work and non-work responsibilities, we also assumed WTC to help em-ployees not to think about work during non-work time and vice versa. IWHI and IHWI imply sustained cognitive activation, which according to the cognitive activation theory of stress (Ursin & Eriksen, 2004) might lead to exhaustion in the long term. Overall, we thus assumed IWHI as well as IHWI to mediate the effect of WTC on exhaustion.

Methodology
We used data from three waves of the BAuA-Working Time Survey (2015, 2017, and 2019; Wöhrmann et al., 2020), a panel study that is representative of a large part of the German work-force. For this study, only participants that took part in all three waves of data collection were included (balanced panel). We further restricted the sample to dependent employees aged 65 years or younger. Thus, our final sample included 3,990 employees (48% women; average age of 47.35 years, SD 8.60). We used path analysis, more specifically cross-lagged panel models, to test the associations between WTC, IWHI, IHWI, and exhaustion. IWHI and IHWI were ana-lyzed separately to reduce model complexity.

Results
Regarding IWHI, both paths of causal mediation were significant and we found a small, but sig-nificant indirect effect of WTC on employees’ exhaustion via IWHI. Regarding IHWI, we found a positive effect of WTC on IHWI, but no significant relation between IHWI and employees’ exhaustion. Thus, we found IWHI but not IHWI to mediate the effect of WTC on exhaustion. In addition, our results partly suggest reverse and reciprocal relationships of the examined con-structs.

Limitations
Like any study, this study is not without limitations. Since our data is based on self-report, we cannot exclude common method bias. Moreover, we did not differentiate between the two di-mensions of WTC, namely control over time off and control over daily hours.

Practical Implication
Non-exhausted and healthy employees are highly desirable for organizations. Our results sup-port the assumption of WTC as an important job resource that promotes psychologically switch-ing off from work during non-work time, helping to protect employees from exhaustion. Thus, organizations should provide their employees with a certain extent of temporal autonomy to help them not to think about work during free time, and thus promote their employees’ health.

Originality/Value
Previous studies have shown that WTC positively affects employees’ health via reduced work-home interference. We extend this result and show that the sub-dimension of IWHI also acts as a mediator. In addition, to our best knowledge, this study is the first investigating the potential mediating role of home-to-work interference.

Research goals and why the work was worth doing
Due to demographic change and rising old-age pension eligibility ages many employers already see a need to provide work environments that supports the maintenance of older employees’ work ability. One important characteristic of the work environment is the organization of working time. For example, long working hours have been linked to ill health, while work time control is often regarded a resource to adapt working hours to individual needs. Furthermore, the possibility to separate work and private life is important for psychological detachment from work. However, employees differ with regard to their preferred working time arrangements. Therefore, the aim of the study was to shed light on the role of person-environment fit regarding working time arrangements for older employees’ prospected work ability.

Theoretical background
According to the model of successful aging (Kooij et al., 2020), the work environment has to fit the individual older worker’s preferences and abilities to ensure the maintenance of work ability. Thus, it is reasonable to assume that the person-environment fit perspective is applicable to working time arrangements. Based on earlier research we assumed that working time arrangements have main effects on prospected work ability to be better when working hours are short, work-time control is high, and segmentation of work and private life is possible. Based on person-environment fit theory (Edwards et al., 1998) it can also be assumed that the prospected work ability is higher when workplace supplies regarding working time arrangements meet or exceed the preferences of older employees.

Methodology
We used data from the BAuA-Working Time Survey 2017. This panel study is representative of most of the German workforce. We used a subsample of N = 4,322 employees aged 50 and above. To determine their prospected work ability, the participants indicated the age until which they expected to be physically and mentally able to continue working in their current job. Participants specified the number of their actual and preferred weekly working hours, the extent of their actual and preferred control over the start and end of their workday as well as their possibilities and their preferences regarding work-private life segmentation.

Results
Polynomial regression and response surface analyses confirmed all hypotheses regarding the main effects. Shorter weekly working hours, higher supplies in terms of control over the beginning and end of workdays, and higher workplace segmentation supplies were related to better prospected work ability. However, examination and testing of surface parameters did not reveal the expected congruence effects. Inspection of the response surface plots confirmed these results since the surfaces did not assume shape of an inverted parabola above the line of incongruence (which would be characteristic of congruence effects). Despite the absence of a congruence effect, exploratory analyses point towards a higher prospected work ability among employees with longer preferred weekly working hours, lower preferred control and lower work-life segmentation preferences.

Limitations
Cross-sectional self-report data was used for the analyses, thus selection effects and reverse or reciprocal relationships cannot be ruled out. We cannot foreclose that in the estimation of prospected work ability many other factors may have been taken into account including possible adaptations to working time arrangements. Furthermore, polynomial regression does not allow for modelling measurement error that is likely when relying on single items.

Research/Practical Implications
Our findings indicate that prospected work ability seems to be rather related to the design of working time arrangements with respect to scientific ergonomic principles allowing for sufficient recovery. The perfect fit of these working time arrangements with personal preferences, however, does not seem to be a precondition for older employees’ prospected work ability. This challenges assumptions of the person-environment fit perspective as – for example – applied in the model of successful aging. Future research should look more closely into the role of individual preferences and the needs-supplies fit with regard to work design. Also, more specific outcome variables measuring current work ability of older workers could be investigated.

Originality/Value
Over the last years, research efforts with respect to work design that considers the needs of older employees have increased. With our study, we contribute to this line of research by combining it with a person-environment fit perspective and considering the individual older employees’ needs. Using large-scale national data as well as a response surface analysis technique allows us to unravel the complex interplay between working time arrangements and older employees’ preferences that may add to their prospected work ability.

Research goals and why the work was worth doing
Duration and timing of work are essential factors for employees’ wellbeing. Presumably, the relevance of these dimensions has accelerated against the backdrop of the pandemic with many working parents facing challenges in terms of homeschooling and lockdown of childcare facilities. Meanwhile, in Germany, we observe extensive use of short-time work, prolonged work hours in system-relevant occupations and an increase in mobile work and telework with an associated flexibilization of work hours. Our study aims at examining the role of working time duration, work time control and working time preferences for employees’ health and work-life balance in a life-phase sensitive approach in light of the pandemic.

Theoretical background
The demand-control model (Karasek, 1979; Karasek & Theorell, 1990) classifies jobs according to the interplay between the demands of a work situation (job demands) and employee’s autonomy (job control). We apply this model to the context of working time by examining work hours (as a type of working time demands) and work time control. Furthermore, person-environment-fit theory (French et al., 1982; Harrison, 1978; Voydanoff, 2005) suggests that a fit between people’s needs and the supplies a workplace offers predicts employees’ well-being. Thus, we also take into account employees’ working time preferences and needs, i. e. preferred work hours and perceived sufficiency of work time control.

Design/Methodology/Approach/Intervention
We use data from a repeated cross-sectional survey from two time frames (November 2018 to January 2019; April to June 2021) among a representative sample of 2,000 employees living and working in North Rhine-Westphalia, the largest federate state in Germany. Data is collected by professional interviewers via computer-assisted telephone interviews. Analyses focus on the joint and interactive effects of work hours, work time control and working time preferences on employees’ health and work-life balance among employees with and without childcare demands before and during the pandemic.

Results obtained or expected (if not available, it must be made clear when they will be)
The ongoing data collection during the corona pandemic will be completed by June 2021. Results will be available by the end of 2021.

Limitations
The survey design of repeated cross-sectional measurements does not allow for panel analyses following the same individuals over time. This limits causal interpretability of the results. Due to time restrictions during telephone interviews, all variables are measured via single items. Moreover, reliance on subjective self-report measures could affect results as a result of common method bias. Moreover, follow-up measurements will be needed to validate our findings in post-pandemic times.

Research/Practical Implications
The more flexible handling of work hours as a side effect of excessive use of telework and mobile work during the pandemic has fueled political debates concerning the implementation of the European Working Time Directive and a revision of the German Hours of Work Act. Our analyses shed light on the role of work hours, work time control and working time preferences on employees’ health and work-life balance before and during the pandemic, thereby highlighting chances and risks of autonomous working time regimes. Findings may give orientation for organizations and policy makers concerned with the design of “new work” practices that promote health and work-life balance in post-pandemic times.

Originality/Value
Drawing on demand-control model and person-environment-fit theory, we link two theoretical frameworks and apply these to the topic of working time. Using representative data collected before and during the corona pandemic, our analyses test the validity of these theories in the context of working time prior and after this caesura. Our findings may advance research on work hours and the work-home interface by generating novel insights and a better understanding of the interplay between work duration, work time control and individual working time preferences and needs. Thereby, we aim to contribute to a design of work schedules and working time legislation that secures employees’ health and fosters work-life balance beyond the pandemic.

Research goals and why the work was worth doing: Unemployment is both a micro-level (i.e., individual unemployment) and a macro-level phenomenon (i.e., unemployment in a geographic entity) with both types affecting not only the unemployed but also partners and children. Differentiating between micro- and macro-level unemployment, this review draws on research from different disciplines (i.e., psychology, management, economics, sociology, and epidemiology) and theoretical perspectives. We conceptually integrate the effects of unemployment on partners’ and children’s well-being, relationships, and career (prospects), thus providing a better understanding of the phenomenon.
Theoretical background: With the review, we integrate four theoretical perspectives on the partner and children effects of unemployment. The economic perspective suggests that unemployment affects the whole household by reducing resources that can no longer be allocated to certain investments and by changing decision-making. The stress perspective classifies unemployment as a stressor prone to elicit affective, behavioral, and/or cognitive strain reactions. The frustration—aggression perspective asserts that the frustration resulting from unemployment elicits aggressive tendencies. The homophily perspective suggests that unemployed individuals, partners, and children might have similar life experiences due to homogamy and the acquisition of the parental habitus (Bourdieu, 1996). These four perspectives sometimes align and sometimes compete with each when explaining numerous partner and children effects of unemployment.
Design/Methodology/Approach/Intervention: We conducted a database driven systematic review with predefined search terms and inclusion criteria. To be included, primary studies had to be peer-reviewed journal articles published in English in or after 1979. We applied no restrictions in terms of the studies’ geographical setting. The database search yielded 6,699 hits with 346 primary studies being successfully sifted.
Results: Supporting the economic and stress perspectives, micro-level unemployment relates negatively to partners’ psychological and physical well-being. In line with the frustration-aggression perspective, the relationship between partners deteriorates with micro-level unemployment. However, there is little evidence for a positive association of macro-level unemployment and the divorce rate, pointing to non-homologous effects. The economic perspective (added-work effect) and the homophily perspectives (coupled unemployment) predict opposite effects of micro-level unemployment on partners’ careers. Our review shows more support for the added-work effect: partners of unemployed (vs. employed) individuals are more likely to participate in the labor market. In line with the economic and stress perspectives, parents’ micro-level unemployment impairs children’s well-being. However, the effects of micro- and macro-level unemployment on children’s well-being are not homologous, as we identified research showing that regional unemployment rates were either unrelated or positively related to children’s well-being. Empirical research supports the frustration-aggression perspective, as both micro- and macro-level unemployment emerged as a risk factor for children’s relationship with their parents with the evidence base being stronger for micro-level unemployment’s effects. Finally, alluding to the economic and homophily perspectives, unemployment negatively affects children’s career prospects, on both the micro- and the macro-level.
Limitations: The results of the systematic review are potentially subject to publication bias which favors primary studies with significant (vs. non-significant) results.
Research/Practical Implications: We invite future research to investigate micro- and macro-level unemployment jointly to identify the micro-level processes that mediate the partner and children effects of macro-level unemployment. This approach would also allow modeling interactive effects of macro- and micro-level unemployment. When examining the cost-effectiveness of public policies to reduce macro-level unemployment, we suggest considering partner and children effects of unemployment. Interventions that target partners and children of individuals who are unemployed should be developed and evaluated.
Originality/Value:
We explicitly distinguish between the effects of micro-level (i.e., a household member being unemployed) and macro-level (i.e., regional/country-level) unemployment on partners and children. This is an important distinction because a priori homology cannot be readily assumed as our review shows. Reflecting on the multilevel character of unemployment allows us to compare relevant theoretical perspectives on how unemployment affects partners and children and to account for contradicting results. The study of unemployment is innately interdisciplinary. We thus advance unemployment research by reviewing and integrating studies from different disciplines as previously empirical research has often been caught in its respective disciplinary perspectives.

Research Goals:
Job insecurity constitutes one of the most critical stressors in people’s working life and holds harmful consequences for workers, such as burnout. Yet, research remains elusive regarding factors reducing or aggravating the impact of job insecurity on employees’ well-being.
We aim to uncover how and why burnout arises from job insecurity and to identify factors facilitating or hindering its development. In combining a career with a stress perspective, we place the process from job insecurity to burnout into one of stress research’s core theoretical foundations: The transactional stress model (Lazarus & Folkman, 1984). We thus examine (1) whether job insecurity has stronger affective consequences when people are more deeply rooted within their organizations (i.e., operationalized via career entrenchment) and (2) whether and how employees’ coping reduces burnout caused by job insecurity (operationalized via career adaptability and career planning).

Theoretical Background:
When experiencing job insecurity, employees feel exposed to potentially impending involuntary job and career changes. Job insecurity is divided into cognitive job insecurity (i.e., the perceived probability of job or benefit loss) and affective job insecurity (i.e., the fear or worry about possibly losing a job). While research diverges in treating job insecurity as a cognitive or affective phenomenon, both constructs connect to different outcomes. We thus consider both separately to ensure an accurate representation of the underlying mechanisms.
To shed light into the process unfolding from job insecurity to burnout, the transactional stress model offers a helpful framework (Lazarus & Folkman, 1984). According to the model, strain results from a (two-step) appraisal process: While a person assesses the relevance of the situation and its potential harm in a primary appraisal, a secondary appraisal addresses whether the person has the needed resources to deal with the situation adequately. Appraisals and coping influence people’s reactions to strain and determine whether they exert actions to counteract the hassles. Successful coping results in perceived changes in the situation or one’s resources. For this study, we specified two appraisals (i.e., career entrenchment and career adaptability) to determine how threatening employees perceive job insecurity. We proposed a model expecting employees’ perceptions of their jobs at risk (i.e., cognitive job insecurity; stressor) to cause threat perceptions (i.e., affective job insecurity; primary appraisal), particularly when lacking alternatives (i.e., career entrenchment). Depending on this and employees’ career adaptive responses (i.e., career adaptability; secondary appraisal), we expected employees to engage in coping (i.e., career planning; coping), which should lower their strain (i.e., burnout). While the transactional stress model assumes the phases as sequential, we aim to test whether this holds or whether the established link between job insecurity and burnout unfolds independent of the phases. Understanding whether the model unfolds sequentially provides guidance for future research and practice regarding the influence of appraisal and coping variables in the relationship of job insecurity and burnout.

Design:
We collected data via three online surveys, each 1.5 months apart using a cross-lagged design with 174 international employees (81 men, 92 women, ø age = 34.11) who rated their jobs as normally distributed from secure to insecure. We measured cognitive and affective job insecurity, career entrenchment, and control variables at time 1, career adaptability and career planning at time 2, and burnout at time 3. Data were analyzed using structural equation modeling in MPLUS.

Results:
Results supported our propositions: Cognitive job insecurity indirectly linked to burnout via affective job insecurity and career planning; cognitive job insecurity fostered affective job insecurity, particularly among highly entrenched employees. Affective job insecurity impaired career planning moderated by career adaptability. Planning reduced burnout. We found no direct effects of cognitive or affective job insecurity on burnout, nor did cognitive job insecurity have a direct effect on planning. Our findings suggest the transactional stress model’s phases to unfold sequentially.

Limitations:
Participants might not be representative of the average working population. While we conducted a three-wave design controlling for prior measurements, we cannot prove causal relations. We also cannot rule out reverse causation of certain effects.

Research/Practical Implications:
We contribute to job insecurity, strain, and career research by revealing career-related mediators and moderators involved in the relationship between job insecurity and burnout. We illustrate that employees who experience job insecurity undergo the strain phases sequentially. This knowledge provides a relevant framework for future research and practice.

Originality/Value:
We highlight how employees can reduce the fear of job loss during job insecurity by becoming less dependent on organizations and how their career adaptability can then help cope more via career planning to prevent burnout.

Research goals:
Financial worries are one of the most frequently named stressors among working adults, still their impact on work behavior and well-being is mostly overlooked (APA, 2019, 2020; Sinclair & Cheung, 2016). This is problematic, as financial worries often go hand in hand with mental health problems, with low living standards and other life stressors which can impact work performance. Ignoring financial worries in work psychological research renders models imprecise and runs the danger of a limited understanding of precarious work (Creed et al., 2020). This study has the goal to illustrate the impact of financial worries on wellbeing and work behavior (contract breach and turnover intentions) and to offer a valid theoretical framework for the study of financial worries and their inherently dynamic nature.

Theoretical background:
Conceptually this study draws on uncertainty management theory (Van den Bos, 2002, 2009) which it combines with novel understandings about the temporal dynamics of organisational behavior (Griep & Hansen, 2020). We argue that financial worries imply stressful situations full of uncertainties – to meet current and future financial demands. A financial situation however will be evaluated in reference – not only to other people but also to oneself previously. This is why it is important to not only take absolute levels of financial worries into account, but also their ‘velocity’ in comparison to previous time points – whether things improved or got worse (Hsee & Abelson, 1991). Furthermore we propose an interaction between absolute levels and velocity of financial worries: people who suffer from high worries on average, will be more affected by those changes in velocity, as changes to a vulnerable financial situation comes with more uncertainty, than people who have comparably low average financial worries.
We look at health outcomes, psychological contract breach and turnover intentions as typical outcomes in the context of uncertainty management. In times of uncertainty, people suffer in their mental health and strive for certainty – by judging how fair the situation is they are in (i.e. looking at their psychological contract) and by evaluating whether the certainty of their situation can be enhanced (i.e. by searching for alternative employment) (Van den Bos, 2009).

Design/Methodology/Approach/Intervention:
To test these hypotheses, data of a four-wave study conducted among n = 711 British employees across a variety of occupations was evaluated. Financial worries, emotional exhaustion, perceived contract breach and turnover intentions were measured with standardised, valid instruments. We control for job insecurity, which is a related but separate aspect of precarity (Creed et al., 2020).

Results:
Multilevel regression models conducted in MPlus confirmed most of the hypotheses. As expected, people who reported more financial worries reported worse mental health, more contract breach and turnover-intentions than people with less financial worries. Furthermore, people who experienced an increase in financial worries in comparison to a previous time point reported a worsening of mental health and increase in contract breach and turnover intentions. The interaction hypothesis was not supported: Although there were differences in the slopes, these were not due to different aggregate levels of financial worries. In other words, different to what was expected, people who had on average high financial worries reacted in the same way to an increase or decrease of worries as people with on average low financial worries.

Limitations:
This is the first study to show the dynamic impact of financial worries on mental health and work behavior. The effects in this study are robust, but would need replication in other samples. We only offer a subjective measure of financial worries, as previous studies show a close relationship to a person’s financial income (e.g. Klug et al. 2020).

Research/Practical Implications:
Researchers and practitioners eager to explain mental health problems or work behavior would be well-advised to include a person’s financial worries into their predictive models. The policy conclusion for our findings are very clear – in order to reduce mental exhaustion, feelings of contract breach and turnover intentions in organisations – why not pay people better?

Originality/Value:
This study is the first to evidence the impact of financial worries (absolute level but also velocity) not only on mental health but also on work behavior. By adopting uncertainty management theory it also offers a novel theoretical lens to analyse the meaning of financial worries at the workplace.

Theoretical background:
Driven by globalized markets and political deregulation, an erosion of structured and secure employment contracts has taken place in favour of flexible and unstructured forms of employment. Flexible employment contracts are associated with higher precarization risks because they tend to offer less income, security, and integration, as well as more social deprivation to employees. The consequences of precarious employment are not limited to work and the workplace but extend to impaired individual health, well-being, family formation, and social life in general.
If employment is to be a mechanism to lift people out of poverty, work must be remunerated adequately to allow for a decent standard of living. Fair or living wages must also be complemented by decent employment and working conditions to allow for capability development in and through work. Need satisfaction has been shown to mediate the transmission process from decent work to individual health-, work- and life-related outcomes.
The International Labour Organization’s (ILO) decent work agenda comprises fair wages, social protection, workplace rights, equal treatment, opportunities for self-development, as well as being recognised and heard. Psychological research on decent work is still at its beginning, with existing approaches building on either the ILO definition or the psychology of working framework. While models of human-centered job design also concern the issue of decent work, the relevance and applicability of these models to all workers remains disputed.

New Perspectives/Contributions:
In this contribution, we first discuss the merits and limitations of existing approaches to conceptualize decent work and then introduce a five-dimensional framework of decent work grounded in sociological research on precarious employment. This framework consists of reproductive–material, social–communicative, legal–institutional [participation], status and recognition, and meaningful–subject-related aspects and features living wages as a core element.
Furthermore, we propose the five dimensions of decent work contribute differentially to the satisfaction of hierarchically ordered sets of needs, which in turn promotes capability development. By considering individual, organizational, and country-level influences, we then clarify the context-dependent salience of the five dimensions of decent work for individuals.
Finally, we apply this framework to explain the characteristic shape of the curve that links income to capability development, showing that an integrated perspective on living wages, decent work and need satisfaction provides a context to better understand capability development.

Research/Practical Implications:
By focusing on subjective experiences associated with decent work, we aim to further stimulate the development of a psychological perspective on both living wages and the broader concept of decent work. Although our decent work framework builds on sociological research on precarious employment, it resonates well with existing psychological conceptualizations. We therefore argue for integrative work towards a more complete conceptualization of factors that comprise decent work.
Our framework yields a number of propositions that should be tested empirically. Such testing may not only improve the determination of a living wage that considers quality of life and work life as goal in itself, it could also help to identify work-related conditions that, much like living wages, boost human capability.
The framework presented here could also point to differences beyond the income level, such as inequalities concerning gender, nationality, age, education levels, or economic context. Finding solutions to mitigate such inequalities translates directly to tangible recommendations for policy-makers, organizations, and individuals.

Originality/Value:
Although psychological conceptualizations of decent work exist, our framework draws on its own theoretical basis and has unique features that offer a more complete perspective on decent work, strongly intertwined with living wages.
By linking specific dimensions of decent work to the satisfaction of specific hierarchically-ordered sets of needs, we are able to offer a new explanation of the characteristic curve that explains capability development as a result of increased income.
The explicit consideration of contextual factors stresses the need to paint a more contextualized picture of how decent work and living wages impact individuals.

Work hours have been gradually inclining in the U.S. in the past two decades (OECD, 2021). Managers often choose to work long hours for personal rewards and financial gains (Brett & Stroh, 2003). However, work hours have been related to impaired psychological and physical health (Ganster et al., 2016), and scholars warn that working long hours can be detrimental for organizations and employees in the long run (Lundberg & Cooper, 2011; Pfeffer, 2010). Whereas the literature on work hours and health is rapidly maturing, little is known about how working long hours affects the performance of managers and other professionals (with the exception of healthcare; Lockley et al., 2007). It is important to address this question since employee performance affects the bottom line of organizations (Motowidlo, 2003) and plays a key role in employees’ career success (Bozionelos et al., 2016). Therefore, the goal of this research is to investigate how daily work hours affect next day performance. Using a recovery angle (Fritz & Sonnentag, 2006), we contribute to the work hours literature by zooming in on the daily process whereby working long hours might potentially backfire.
Theoretical Framework
Conservation of Resources theory (COR; Hobfoll, 1989, 2002) is built on the premise that individuals value various resources, which they aim to protect and expand. Threats to resources, or actual resource loss, results in stress. COR theory has been used to explain the recovery process after work (Ten Brummelhuis & Bakker, 2012), suggesting that individuals use resources throughout the day, and that well-being is maintained when they replenish resources after work. Working long hours hinders this recovery process, as individuals have less time to recover and stay mentally connected to work longer (Sonnentag & Bayer, 2005), which impedes sleep (Åkerstedt et al., 2002; Hülsheger et al., 2014). When recovery is incomplete, the employee wakes up with drained physical (e.g., physical fatigue) and psychological resources (e.g., resilience), making it more difficult to perform optimally on the job (Xanthopoulou et al., 2009).
Hypothesis 1. Daily work hours are (a) negatively related to next morning resilience and (b) positively related to next morning physical fatigue through reduced sleep hours.
Hypothesis 2. Daily work hours are negatively related to next day work performance through (a) reduced morning resilience and sleep hours and (b) enhanced morning physical fatigue and reduced sleep hours.
Methods
Four organizations in North America participated in a study that asked participants to fill in a general questionnaire, wear a Fitbit (a wrist worn sleep tracker) and complete a morning log to report personal resource levels, and an end of the day log to register hours work on five workdays. On each workday, a coworker rated the participant’s work performance. We obtained data from 81 employees and coworkers reporting on 312 days. Sample descriptive and measures are available from the first author upon request.
We tested our hypotheses using multilevel regression analysis in MPlus (Muthén & Muthén, 2012). ). Level-1 predictor variables were person-mean centered (Enders & Tofighi, 2007; Ohly et al., 2010). For the indirect effect analysis, we used the indirect model command in Mplus and we calculated the 95% bias corrected confidence intervals of each indirect effect using the Monte Carlo method in R with 20,000 repetitions (Bauer et al. 2006).
Results
Daily work hours were negatively related to sleep hours (γ = -.257, SE = .059, p < .001), whereas sleep hours were positively related to morning resilience (γ = .082, SE = .036, p < .05), and negatively related to morning physical fatigue (γ = -.219, SE = .067, p < .001). The negative indirect effects between work hours and morning resilience (γ = -.021, SE = .010, p < .05), and the positive indirect effects between work hours and morning fatigue (γ = .056, SE = .021, p < .01) were significant, providing support for Hypotheses 1a and 1b. Whereas physical fatigue was not significantly related to work performance, morning resilience and work performance were significantly related (γ = .315, SE = .081, p < .001). The negative double indirect effect between daily work hours and work performance through reduced sleep hours and resilience was significant (γ = -.007, SE = .004, p < .05), supporting Hypothesis 2a, but not Hypothesis 2b. Discussion Our study shows that working long hours can be detrimental for work performance because employees get a shorter sleep than on days when they work fewer hours. Hence, their physical and psychological energy levels are suboptimal in the morning, hindering how effective they are at work.

Research goals and why the work was worth doing
The study aimed at ecologically exploring whether and how the short-term affective reactions to work stressors are moderated by the individual level of workaholism. The investigation of such micro-processes is fundamental to better characterize the reactivity profile of workaholic workers, as well as to design interventions aiming at preventing the negative effects that job design might exert on the health and wellbeing of these workers.
Theoretical background
Workaholism is recognized as a dysfunctional form of heavy work investment, characterized by preoccupation and compulsion regarding one’s work, difficulties in disengaging from it, and the tendency to work for very long hours. Although workaholism has been consistently associated with negative outcomes, most studies adopted between-individuals designs focusing on stable differences between workaholics and controls. Here, we adopted a within-individual perspective investigating the micro-processes of health deterioration activated by the interaction between workaholism and working conditions. Building on studies identifying emotional strain as the most immediate response to work stressors, we expected the individual level of workaholism to magnify work-related strain responses, and specifically the affective reactions to increased workload, and the linear growth of perceived fatigue over time, within a typical workday.
Design
A three-day experience sampling methods (ESM) protocol was conducted with a convenience sample of knowledge workers mainly involved in back-office activities. Following a preliminary questionnaire measuring workaholism, demographic and occupational variables, participants received six ESM surveys per day (each 90 min from 10:00 to 18:30) over three non-consecutive workdays (i.e., Monday, Wednesday, and Friday), measuring their current mood (negative affect, tense arousal and fatigue) and perceived workload.
Results obtained or expected (if not available, it must be made clear when they will be)
Results of multilevel modeling conducted on 1,484 observations from 147 participants (51.7% females, response rate = 56.1 ± 20.7%) showed a substantial relationship between workload and each mood dimension, as well as a positive linear trend of fatigue over time spent working. Workaholism was found to moderate both relationships, while being positively associated with higher mean levels of negative affect, tense arousal and fatigue. However, contrary to our expectations, higher levels of workaholism were associated with a weaker rather than a stronger relationship between workload and negative affect, and between time spent working and fatigue, even after controlling for sex, age, job position and family status.
Limitations
The limitations of this study include its relatively short length (i.e., three days might not be generalizable to stable work-related processes), and the use of a sample weakly representative of those occupations associated with higher workaholism levels (i.e., managers and employers only accounted for the 13.6% of the sample). Moreover, the lack of objective (e.g., physiological) strain indicators does not allow discriminating between different levels of the stress response.
Research/Practical Implications
Taking these limitations into account, a possible interpretation of these findings is that individuals with higher workaholic tendencies might be more resistant/tolerant to increased workload and sustained work time compared to their colleagues. Alternatively, workaholics might be just unaware of/unsensitive to the short-term effects of work stressors, eventually resulting in the same long-term negative outcomes affecting their colleagues. Research and practical implications of both interpretations are discussed.
Originality/Value
This is among the first studies characterizing the short-term responses of workaholics to work stressors. The ecological nature of its design, the high density of measurement occasions, and the consideration of both perceived (workload) and objective working conditions (time spent working) provide an additional value to the present study.

Research Goals
Workaholism and work engagement are two forms of heavy work investment that can co-occur. However, research to date has rarely explored the way in which these constructs co-occur. In other words, we know very little about the different forms or configurations of workaholism and work engagement employees exhibit. Even less research has examined how these constructs (and their configurations) relate to health behaviors, such as diet, fitness, and wellness program participation. These are important questions to answer because health behaviors are a key proximal antecedent to health outcomes. Thus, a better understanding of what types of heavy work investors exist and how these heavy work investment profiles relate to both health behaviors and health outcomes is an important step towards targeted interventions to enhance the health of heavy work investors.
Theoretical Background
Workaholism is a multidimensional construct reflecting a compulsion to work, frequent thoughts about work, negative emotions when unable to work, and working beyond expectations. Work engagement is also a multidimensional construct, characterized by energy towards, commitment to, and being engrossed in work. Research suggests that workaholism and work engagement are moderately positively associated with each other (Clark et al., 2016), suggesting that they may co-occur. Moreover, preliminary work has examined what forms (i.e., profiles) these constructs co-occur (e.g., Gillet et al., 2018; Spurk et al., 2020). However, these efforts have collapsed across workaholism and/or work engagement dimensions. While these can be conceptualized and operationalized as higher-order constructs, research suggests differential relationships at the dimension-level (Clark et al., 2016, 2020; Halbesleben, 2010; Schaufeli et al., 2006). Finally, workaholism and work engagement have been found to interact with one another to influence health outcomes (e.g., ten Brummelhuis et al., 2017). However, we do not know the mechanisms through which this occurs. Thus, we seek to explore various health behaviors and perceptions as proximal outcomes in the relationship between heavy work investment profiles and health.
Design
Data were collected as part of an annual wellness survey at a state public health agency in the U.S. A total of 1,131 employees provided at least partial data for the variables of interest, which included participation in various wellness programs; perceptions of overall health, job stress, and health climate; exercise, diet, drug use, and sleep behaviors; and work limitations and chronic health conditions. Latent profile analysis was conducted to explore heavy work investment profiles. Key outcomes were then compared across the profiles.
Results
The best-fitting latent profile solution differentiated among 8 heavy work investment profiles. The profiles were both quantitatively and qualitatively distinct, including lower levels of all dimensions, high workaholism and low work engagement, high work engagement and low workaholism, higher levels of all dimensions, among others. Importantly and consistent with previous research at the dimension-level, there was variability across dimensions within the same construct and profile, which points to the importance of examining these profiles at the dimension-level. The profile with higher work engagement dimensions and lower workaholism dimensions tended to be the most active in various wellness programs, followed by the profile with both high workaholism and work engagement. There was not a significant difference between profiles on sleep quantity, but there was on sleep quality. Interestingly, profiles with relatively extreme levels of workaholism and/or work engagement had significantly higher monthly alcohol consumption compared to profiles closer to average levels of workaholism and work engagement. Finally, profiles with low engagement and/or high workaholism perceived greater job stress and lower health climate.
Limitations
The primary limitation of this study is the cross-sectional design, which does not allow for temporal/causal order inferences. Additionally, because the sample came from a public health agency with wellness initiatives and an annual wellness survey, it is possible that the profiles and/or relationships between profiles and outcomes would not generalize to employees in less health-conscious work environments.
Implications
This research is exploratory in nature and has greater implications for future research than for practice. This study provides evidence that workaholism and work engagement interact with each other in numerous, qualitatively distinct forms. Moreover, variability of dimension levels within constructs and profiles demonstrates the benefit of researching at the dimension-level for these constructs. Finally, significant differences between profiles on various health behaviors is helpful as a first step to target health behavior change among heavy work investors.
Originality
This is the first known research on heavy work investment to focus on health behaviors, which represent proximal antecedents to health outcomes. Additionally, it is the first known study to explore profiles of workaholism and work engagement at the dimension-level.

Research goals: The present study aimed to investigate the role of harmonious and obsessive passion for work among those workers who live their work with a high level of dedication, namely academic workers. The research was conducted during the Covid-19 pandemic to investigate the association of academics' work passion with their perceptions of work-family conflict (WFC) and insomnia. The analysis also considered the buffering role of detachment as a recovery strategy that might be particularly hard for this type of worker under the exceptional conditions of mandatory work-from-home and distance learning.
Theoretical background: Passion is defined as the love an individual feels for an activity considered important, to the point that it becomes an essential element of his/her identity. The Vallerand et al.’s (2003) Dualistic Model of Passion specified that depending on how the activity is internalized in the individual's identity, it is possible to live a harmonious or obsessive passion. Harmonious passion is characterized by a strong engagement in the passionate activity, which is in harmony with other aspects of the individual's life and is not pervasive in his/her identity. On the contrary, obsessive passion is characterized by intrapersonal or interpersonal pressures that drive the individual to participate obsessively in the activity, resulting in conflicts with other elements of the individual's life. The Model can also be applied to the work experience and can explain workers' perceptions of well-being or stress (Lavigne et al., 2012). Previous studies have shown that harmonious passion can reduce emotional exhaustion and increase job satisfaction (Pollack et al., 2020), while obsessive passion has been associated with high levels of cognitive symptoms, anxiety, conflict, and exhaustion. From this perspective, understanding the role of the two types of passion among academic workers, a profession characterized by high levels of dedication and engagement, is a key element in understanding how this inclination have affected their well-being during the Covid 19 pandemic and lockdowns. In particular, the relationship between passion for work and both WFC and insomnia problems was considered. In addition, the buffering effect of detachment has been investigated, in order to help identify recovery strategies that may protect against negative consequences for workers’ well-being.
Methodology: The study was conducted during the Covid 19 pandemic in Italy and involved 1119 participants from 11 different universities. Among participants, 54.8% were male; 64.3% were professors, while the rest were researchers; the mean age was 50.39 years (SD = 9.64). They were working from home an average of 3.61 days per week (SD = 1.46). All participants completed an online questionnaire after providing informed consent. SPSS 27 and Mplus 8 software were used to conduct the analyses.
Results: The structural equation model with the mediation of work-family conflict showed a good fit to the data [χ2(92)=267.01, p<0.001, RMSEA=0.04, CFI=0.98, TLI=0.97, SRMR=0.03]. According to the results, harmonious and obsessive passion for work showed a negative and positive relationship with WFC, respectively, which, in turn was positively related to insomnia. The significant indirect effect was negative for harmonious passion and positive for obsessive passion. In addition, detachment also showed a negative indirect effect to insomnia. The moderation of detachment in the association between passion for work and WFC was significant and positive only in the case of harmonious passion. In the analyses, we controlled for gender and role. Limitations: Although the study involved a large sample of academics, it has the limitation that a cross-sectional design and self-reported data were used, which precluded the possibility of drawing any conclusions regarding causal effects and increased the likelihood of common method variance effects. Research/Practical Implications: Future longitudinal studies should be conducted to provide empirical support for the model tested in the current study. Another interesting and useful direction for future research should also investigate the relationship between work passion and workaholism, a form of heavy work investment that implies several harmful health consequences, in academics. Practical implications relate to the recovery process, which appeared to be effective only in academics more characterized by harmonious passion, allowing them to prevent the risk of WFC and insomnia, while this was not the case in academics with obsessive passion for work. Academics should be aware of the risk stemmed from the obsessive passion for work, which can lead them to severely compromise their social life and health. Originality/Value: The study focused on the specific academic work and, accordingly to the recent research trend on "Healthy Universities", uncovered the psychosocial risks associated with a "tricky" professional profile that has been increasingly modified over time in Italy.