The first key aspect related to being a good “space roommate” is being tidy and picking up after oneself. There is substantial overlap in both mutual workspaces and mutual personal spaces as a crew lives and works in the same confined habitat. Simply keeping common areas clean and organized reduces stress in others as they have less clutter to sort through to find what they need and fewer distractions in their field of vision. Clutter can negatively impact well-being and productivity at home and work (Roster et al., 2016; Roster and Ferrari, 2020). Space habitats, especially space stations that have been occupied for multiple decades, such as the ISS, are cluttered. Stowage is an ongoing stressor, as numerous pieces of equipment are added to the station with each launch and upgrade, accumulating in translation paths and blocking areas intended for habitability and work (Gore et al., 2021). Clutter can be dangerous: hard-to-find equipment can delay addressing off-nominal events, or objects can float into a crewmember and cause injury (Greene et al., 2018). Tidiness meant to prepare a work area and maintain it for others is procedural, saving time onboard for mission objectives. Tidiness reduces spatial conflict of objects and reduces the interpersonal tensions that might arise from the owners of said objects. Stated one crewmember, “I pick up a little after them, but I am at the point now where I am leaving soon, and it is easier to let it go and let someone else worry about it. X leaves tools out and cables dangling in the middle of a module. Nothing big, but these little things are the difference between making work easy or hard” (Stuster, 2016, p. 64). Tidiness focused on personal items can ensure, for example, that odorous exercise clothes are tethered in agreed-upon areas so that smells are contained and expected. An individual may vary how they organize work versus personal items, but in a confined space, the lines between work and personal overlap. Importantly, decluttering and organizing have been identified as a skill, and training in such skills can have a lasting positive influence on managing clutter and behavioral health (Aso et al., 2017; Aso, 2018).

Consideration of others’ preferences stems from understanding that others’ backgrounds, cultures, personal needs, and situations are diverse. Having sensitivity to these issues requires empathy and perspective-taking cognitive abilities to identify and alter behaviors to accommodate other people, which can be a positive force in a cooperative team environment (Ku et al., 2015; Longmire and Harrison, 2018). Research indicates that dyads high on empathetic perspective-taking are more accurate in synchronizing actions, and empathetic perspective-taking enables anticipation of others’ behavior and improved coordination (Novembre et al., 2019). Key example issues to be aware of may include sensitivity to personal space and to typical routines or ways of working and engaging in personal activities. This item also captures an individual’s motivation to be open to different preferences and find out more about their crewmates. For multi-cultural crews, typical on ISS, cultural awareness and sensitivity are viewed by space psychology experts as particularly important to this aspect of group living. Past studies on Mir and the ISS have found evidence of national, occupational, and organizational cultural differences between American astronauts and Russian cosmonauts, which can complicate living and working together (Boyd et al., 2009). Multi-cultural crews add another layer of complexity to the complex system of spaceflight as crews navigate differences in hierarchies and power distance norms, collectivism versus individualism, gender norms, and how performance is assessed and rewarded in-mission (Bell et al., 2015; Käosaar et al., 2022). Others’ preferences referred to here can be broad as they encompass both work and non-work issues, which are reflected in the item phrasing.

Cooperation and adaptability can also be supported by the appreciation of others’ knowledge, skills, and abilities for work and non-work actions. In a dangerous operational environment with a small crew, each individual is a critical part of supporting the crew’s health and safety and mission objectives. Teams are carefully composed to bring a unique and needed skillset to the mission. The team is also composed and trained so that they are able to deploy those skills efficiently and effectively together. Roles and responsibilities are carefully planned for mission tasks, but team members are also trained as backups or to provide proactive support when another crewmember acts as leader due to position or expertise (Landon et al., 2018). Pre-mission training as an intact team allows a transactive memory system to develop so everyone learns who knows what. A shared mental model develops for taskwork, teamwork, and the general norms of living and working together. Both forms of team cognition support task performance, affect within the team, and belief in team viability (Bachrach et al., 2019; Niler et al., 2021; DeChurch et al., 2023). Spaceflight is a high-consequence environment, so relying on others for safety is another way crews recognize the importance of complementary skillsets. Appreciation is ideally developed during pre-mission preparation periods, including training, and is reinforced as they accomplish tasks in flight and during team debrief opportunities. Interviews of NASA Shuttle astronauts indicated that skepticism related to a non-NASA career astronaut’s (i.e., payload specialist) predicted performance and integration into a crew was reduced when payload specialists were observed working hard, performing tasks successfully, and caring for other crewmembers (NASA, 2010). Appreciation can also stem from informal pre-mission interactions, highlighting the importance of intact and in-person (or virtual synchronous) events that allow for conversations and relationships to form between formal training time. Appreciation can be demonstrated formally and informally via specific recognition and symbols (e.g., awards) or by creating a culture where the team norms appreciative behavior and practices it consistently without prompting. The item is broadly written to recognize any type of appreciation for others’ knowledge, skills, and abilities that might be demonstrated by another crewmember.

Astronauts point to humor as an important way to form connections, alleviate stress and interpersonal tensions, keep connections strong, and maintain morale (Stuster, 2016 Lungeanu et al., 2023). The positive influence of humor has been documented in analogous contexts such as healthcare teams, firefighters, and the military (e.g., Mesmer-Magnus et al., 2012; Sliter et al., 2014; Vivona, 2014). The key to team-supportive humor is the appropriateness for that particular team and context, determined by the individuals giving and receiving jokes. It should be inclusive, rather than exclusive or meant to denigrate others in order for it to maintain team cohesion. A spaceflight analog study found that crewmembers with an affiliative style of humor used positive humor over time, while those with an aggressive style used more negative humor over time (Vazquez and Bell, 2023). Findings showed that the relationship between an affiliative humor style and use was stronger during unstructured, social activities (i.e., lunch) versus a structured, work-oriented decision-making task, suggesting humor is context-dependent. A past study on humor as a coping mechanism in spaceflight found that astronauts and cosmonauts used it more on long-duration flights, and cosmonauts used less affiliative humor than astronauts (Brcic et al., 2018). These findings suggest that cultural nuances and considerations of others are tied to the successful use of team-appropriate humor. The study also found that the use of self-deprecating humor decreased during the mission compared to pre- and post-flight, likely to maintain positivity and morale. Thus, humor is important for group living skills, but it must be deemed “appropriate” to be supportive of the team. The definition of “appropriate” is up to each crew member to decide which drives the phrasing of that survey item.

Long-term commitment to living with colleagues is salient in isolated, confined, long-duration work environments. The final item about going on another mission with a particular individual or crew captures aspects of long-term team viability, that is, “a team’s capacity for sustainability and growth for success in future performance episodes” (Bell and Marentette, 2011, p. 276). It is slightly different from the other five items on the survey because it is future-focused, while the other items consider behaviors a respondent observed being performed by another crewmember. This item is more speculative and asks respondents to imagine what living and working on another extreme mission might be like and—importantly—if they believe the other person or crewmember would contribute positively to the predicted scenario. This item targets the frame of mind in psychologically committing to each other and mission success on long-duration missions. Based on the sociometric approach used in Russian isolation studies (e.g., Vinokhodova et al., 2012), earlier versions of the item added duration details [e.g., 3-month space station tour (McFadden et al., 1994); 3-year mission (Lungeanu et al., 2023)], but duration on the GLS Survey is unnamed such that respondents can consider the unique context of their mission experience with that person or crew.

The ISS is in low Earth orbit, which carries with it the complexity of living and working in microgravity, a high-tempo workload, and a hostile environment with real danger. The typical six-month ISS crew of six astronauts experiences a crew rotation of three members approximately every 3 months (with occasional one- to two-week visitors), resupply and care packages, close coordination with experts in mission control (usually with real-time communication), readily accessible communication with family and friends on Earth, and meaningful work such as spacewalks or extravehicular activities (EVAs) and research tasks. Astronauts are rigorously selected to be team-oriented, resilient, adaptable, and self-sufficient (Landon et al., 2017; Beven et al., 2018; Schmidt and Spychalski, 2021). They are also selected for expertise such as advanced science/technology/engineering/mathematics (STEM) degrees, military and extreme environment experience, as well as passing a medical qualification and psychiatric screening. They undergo years of training in technical and team skills, including training and feedback in stress management, conflict management, teamwork, and group living skills (Landon et al., 2018; Sipes et al., 2021). An assigned crew will engage in an approximately 18-month pre-mission training program to prepare them for a specific mission together. While only some of that pre-mission training is with their intact crew, crewmembers often know each other for several years before the mission.

Astronauts on the ISS completed a digital version of GLS Survey approximately every 30 days as part of a spaceflight standard measures study (PI = G. Clement), only answering items with the crew-referent as the target. Measures of cohesion and performance (as well as other behavioral health and team measures) were completed at the same time. These are short forms of the measures used in the spaceflight analogs (i.e., HERA and SIRIUS) to reduce crew time burden (see Appendix A for items and Table 4 for psychometric properties).

The HERA spaceflight simulation missions occur in a habitat at NASA Johnson Space Center in Houston, Texas (Vessey et al., 2017; Cromwell and Neigut, 2021). Each Campaign is a series of four 45-day missions with the same set of studies, tasks, and analog environment designs, with only a change of crew members between missions. Changes may occur between campaigns, such as studies added or subtracted, and the habitat, tasks, and stressor protocols may be altered. Thus, each HERA campaign is treated separately in the analyses. Crewmembers are selected to be astronaut-like (i.e., 30–55 years old, advanced STEM degrees, psychologically resilient, Class 3 flight physical, no disqualifying medical or psychiatric conditions), and crews are usually mixed gender (see Table 2). All campaigns aim to be spaceflight-like, with living and working conditions similar to the ISS (see Table 3). The crews undergo a two-week pre-mission training period on mission tasks and team skills. They coordinate with mission support personnel, but this is communication delayed up to 5 min one-way.

HERA participants completed the GLS Survey via tablet device every 10 days during the mission. Campaign 4 (C4) crewmembers did not complete the GLS Survey with the crew-level referent, but C5 and C6 crewmembers completed the next iteration of the GLS Survey with individual-level referents and the crew-level referents. All HERA crewmembers completed other behavioral health and team measures on approximately the same mission day (see Appendix A for items and Table 4 for psychometric properties). Participants completed the team cohesion and team performance measures daily in all campaigns and during all missions. Observations taken the same day as the GLS Survey were merged into separate datasets (e.g., GLS and team cohesion as one dataset, whereas GLS and team performance are another dataset) to retain the highest number of observations. The n for each dataset varies due to the removal of careless observations or missing at random datapoints. Team cohesion and team performance measurements taken on non-GLS Survey days were not included in this analysis.

The Scientific International Research In a Unique terrestrial Station (SIRIUS) analog missions occur in the NEK chamber (Nezemnyy Eksperimental’nyy Kompleks) in Moscow, Russia (NASA, 2023b). Crewmembers are selected to be astronaut-like with similar medical and psychological guidelines to HERA, and the mission scenario is also spaceflight-like with a crew of six (see Tables 2, 3). The crews undergo a pre-mission training period of up to 8 weeks on mission tasks and team skills. The crews coordinate with mission support personnel similar to HERA, also with a communication delay of up to 5 min one-way. Some studies and tasks were altered, added, or removed between the two SIRIUS missions. Also, the durations of the two missions were significantly different (120 days, 240 days), creating more distinctness between them. For these reasons, SIRIUS-19 and SIRIUS-21 are considered separately in the analyses.

Due to the longer duration of the SIRIUS missions and concerns for crew time burden, the sampling frequency was changed to collect the GLS Survey every 20 days in-mission. All SIRIUS crewmembers completed the GLS Survey via tablet device with the individual-level referents and with the crew-level referent. They completed team cohesion and team performance measures and other behavioral health and team measures at approximately the same time (see Appendix A for items and Table 4 for psychometric properties). Participants completed the team cohesion and performance measures with an approximate sampling frequency every 3 days. Observations taken the same day as the GLS Survey were merged, and those close to the same day were utilized to fully represent each crew. The range of target administrations varied from measurements taken 4 days later; however, most administrations were within 2 days of the GLS Survey. As with HERA, each measure was merged into a separate combined dataset to address the removed careless observations and missing at random datapoints.

We examined the reliability and validity of a team-level measure of group living skills by (1a: All individuals with self) aggregating crew members’ ratings of all crewmembers including their self-rating to the team level (Chan, 1998), (1b: Peers) aggregating crew members ratings of all other crewmembers excluding their self-rating to the team level, and (1c: Crew Referent) examining the team mean of the responses to items with the group-level referent level.

To examine reliability at the item level, we calculated omegas as a measure of internal consistency (McNeish, 2018). To examine the appropriateness of aggregating the scores to the team level, we calculated the proportion of total variance attributable to team membership [i.e., intraclass correlation, ICC(1)]. We also calculated the reliability of the team mean ICC(2) (Chen et al., 2004; LeBreton and Senter, 2008). To examine the consistency of scores over time (i.e., reliability of change), we conducted variance decomposition analyses that included item, time (mission day), and individual, then again at the team level with item, time, and team/crew (Cranford et al., 2006).

We examined the construct validity and factor structure of the GLS Survey with a multi-level confirmatory factor analysis (CFA; Fabrigar et al., 1999; Mathieu and Taylor, 2006). We also tested for measurement invariance over time (Putnick and Bornstein, 2016; Winter and Depaoli, 2022). Finally, we examined the extent to which GLS scores were related to team cohesion, team viability, and team performance (Kozlowski, 2015; Tannenbaum and Mathieu, 2015; see Appendix A).

We measured the reliability and validity as a “Roommate Quality” measure by aggregating the scores for each individual as rated by their other teammates to look at how each individual on a team was perceived by others. Our analyses followed a similar progression as the previous set of team-level analyses.

For the GLS Survey, when all individuals (including self and crew) were included as referents, ICC(1) and ICC(2) were large [i.e., ICC(1) ≥ 0.25 (LeBreton and Senter, 2008)] and excellent [i.e., ICC(2) > 0.75 (Fleiss, 1986)], and the model had good internal consistency (total ω = 0.93; McNeish, 2018). For the GLS Survey when individuals rated peers, ICC(1) and ICC(2) were large and excellent, and the model had good internal consistency (total ω = 0.94). For the GLS Survey when referred to the whole crew as the referent, resulting ICCs were large and excellent. McDonald’s omega indicated good internal consistency (ω > 0.70): total ω = 0.91 (see Table 4).

Confirmatory factor analysis (CFA) was used to examine the underlying structure of the survey. We ran a two-level CFA to account for the complex and nested nature of the data with individuals nested within teams. Each time point was allowed to be independent, but each observation was clustered/nested within individual and crew. We used the two-level complex maximum-likelihood parameter estimates with standard errors and a chi-square test statistic (when applicable) that are robust to non-normality and non-independence of observations. The MLR standard errors were computed using a sandwich estimator. Mission day was added to the models to account for the temporal nature of the data collection, which occurred multiple times throughout the missions. The results did not meaningfully change; subsequently, mission day was omitted from the final models.

We evaluated models according to the model fit guidelines by Mathieu and Taylor (2006). For the model with all individual ratings of other crewmembers and self, the two-factor model (i.e., two factors nested within at the individual level and two factors between at the team level) had excellent comparative fit [χ²(8) = 20.42, CFI = 0.995, SRMR within = 0.014, SRMR between = 0.014, p < 0.01] and good absolute fit (RMSEA = 0.03; Fabrigar et al., 1999). See Figure 1 for the model, Table 8 for specific factor loadings, and Table 9 for model fit indices. For the model of individual ratings of their peers, the model rating the crew, and the model examining individuals’ roommate quality ratings, the same model structure held with excellent fit. We identified the two factors as “Tidy” and “Considerate.”

We next examined the relationship between GLS and team cohesion, team performance, and team viability. We conducted generalized mixed models to account for the data collected over time (i.e., mission day) and across campaigns. Our first model included mission day and campaign as predictors of either team cohesion, team viability, or team performance. Our second model added the GLS Survey as a predictor. The predictor utilized was the team-level score of each crewmember’s average rating of the five crew referent items (see Table 1) aggregated by their mission crew (e.g., HERA C4 Mission 1) and the crew mission day, labeled as GLS Crew Score Aggregated in the Tables 10–12. We then looked at the marginal R² change between the first and second models to estimate the relationship between GLS and the outcome (e.g., team cohesion, viability, or performance). Results suggested that group living was strongly related to team viability (marginal R² change = 0.40), team cohesion (marginal R² change = 0.28), social cohesion (marginal R² change = 0.25), and moderately related to task cohesion (marginal R² change = 0.18). GLS had a small relationship with team performance (marginal R² 0.08).

For the GLS Survey, when each crewmember was rated by their peers on roommate quality, the irrICC package in R was used to calculate Models 1B and 2B (Gwet, 2014) intra-rater reliability coefficients across multiple raters and time points on the target roommate. The resulting ICCs suggest moderate to excellent agreement. McDonald’s omega indicated good internal consistency (ω > 0.70): total ω = 0.96 (see Table 4).

For the model with all individual ratings of other crewmembers, aggregated to create a roommate quality score for each individual, the two-factor model (i.e., two factors nested within at the individual level and two factors between at the team level) had excellent comparative fit [χ²(8) = 16.42, CFI = 0.99, SRMR within = 0.016, SRMR between = 0.016 p < 0.01] and acceptable absolute fit (RMSEA = 0.057; Fabrigar et al., 1999; Furr, 2017). See Figure 1 for the model, Table 8 for specific factor loadings, and Table 9 for model fit indices.

We examined the relationship between roommate quality and team viability since it was also measured with an individual-level referent and could be calculated as a sociometric measure. We conducted generalized mixed models to account for data being collected over time (i.e., mission day) and across campaigns. The first model included campaign and mission day (time) as a predictor of team viability. Our second model added the GLS Roommate Quality Score. The score was calculated from each crewmember’s average ratings of the five individual referent items (see Table 1) with them as targets (e.g., crewmember-1, -2, -3 scores about crewmember-4). Results indicated a 0.15 marginal R² change, suggesting that roommate quality was moderately related to viability (see Table 14).