DiBenigno & Kellogg (2014). Beyond Occupational Differences: The Importance of Cross-cutting Demographics and Dyadic Toolkits for Collaboration in a U.S. Hospital

Authors:
Julia DiBenigno – MIT Sloan School of Management
Katherine C. Kellogg – MIT Sloan School of Management

Interviewers:
Saheli Nath – Northwestern University
Katikeya Bajpai – Northwestern University

Article link: http://asq.sagepub.com/content/59/3/375

Question 1. Your project examines the processes underlying cross-occupational collaboration and its success and failure. In our reading, the persuasiveness of the study is greatly enhanced by the excellent choice of cases, which very neatly embody the differences in demography and subsequent performance outcomes. In addition, we found the use of ethnography to examine aspects of occupational demography to be quite different and novel from studies in this tradition. Could you give us more details about the process of conducting an inductive study of this scope? For example, how were these two cases selected, how did you refine your theoretical lens during the process of data collection, and what were the challenges of the fieldwork process?

We’d be happy to provide more detail on how we conducted our study. First, in terms of how the two cases were selected, finding matched cases was very important to our study design. Once we gained access to the first medical-surgical unit (Unit A), we next sought a second unit that was comparable on the dimensions that previous scholars had found mattered for cross-occupational collaboration. We did this by first asking the Unit A nurses and PCTs (Patient Care Technicians) which unit they believed was most similar to theirs when they occasionally “floated” to other units throughout the hospital when they were short-staffed. They all hands-down responded, “Unit B,” the other similarly-sized medical-surgical unit at the hospital. We then needed to convince the Unit B nurse manager to give us permission to also study her unit, which we were able to do.

One of the reasons the hospital was interested in our study was that they were having problems with nurse-PCT collaboration, specifically as it related to promptly answering patient bedside calls for help. The first few months of observations were of all shift times on Unit A and B, not just the night shift. It was only on the Unit A night shift, the shift staffed by nurses and PCTS with cross-cutting demographics, that we observed effective collaboration. The other shifts on Units A and B were staffed by nurses and PCTs with consolidated demographics—where occupational characteristics correlated with status-laden demographic characteristics so that nearly all nurses were white, young, and U.S. born while nearly all PCTs were non-white, older, and foreign-born— and had strained nurse-PCT relations. Since we observed variation in collaboration outcomes in the night shift only, we decided to shift DiBenigno’s fieldwork to the night-shifts to compare the effects of these very different occupational demographics on collaboration. DiBenigno’s husband certainly did not like this arrangement! But, studying all those night shifts was worth it to study such interesting, well-matched cases.

Second, in terms of how we refined our theoretical lens during the study, this occurred in two phases. One was during data collection; the other during the review process. During data collection, we realized early on the starkly different levels of collaboration between nurses and PCTs on the Unit A and B night-shifts. However, after scouring the cross-occupational collaboration literature, we could not find any leads in prior research regarding why this difference was occurring since the units were so well-matched on all the factors previously found to affect cross-occupational collaboration. The only substantive difference between the units was the demographic characteristics of the nurses and PCTs. This led us to the diversity and demography literatures, since no cross-occupational studies (to our knowledge) explicitly discuss the role demography plays in shaping cross-occupational collaboration. However, the literature here was initially not much help to us. This literature was very well-suited to explaining the failed collaboration on Unit B as due to the consolidation of positional (occupational) and nominal (demographic identity characteristics) known to many as strong “faultlines”, but less well-suited to explaining how cross-cutting demographics on Unit A generated conditions favorable to collaboration.

This led us to the second phase of refining our theoretical lens. This occurred during the review process, when one of the reviewers made the insightful comment that “weak faultlines” (which we previously used to account for the successful collaboration on Unit A) were insufficient to explain successful collaboration, since it’s “hard to theorize and illustrate how the absence of something (i.e., lack of strong faultlines) works.” This comment pushed our thinking and led us to another round of data analysis in which we focused extensively on what the nurses and PCTs on Unit A were doing to collaborate successfully. This analysis led us to see how the cross-cutting demographics on Unit A loosened the occupational status and identity order so that nurses and PCTs could draw upon alternative forms of expertise, status, emotion rules, and meanings from shared non-occupational identities—what we call “dyadic toolkits”—to collaborate successfully. Interestingly, some of these identities were obvious and visible, such as being “Haitian immigrants,” while others were less visible but just as powerful sources of connection, such as being “working mothers” or “sports fans.”

Finally, what were the challenges of the fieldwork process? There were many challenges which, in our experience, is typical of all field studies. First, DiBenigno conducted the fieldwork during her first and second year of graduate school, so like many of the PCTs studied, she was working “two jobs” taking classes and then “working nights’ conducting fieldwork for portions of the year. This was physically challenging since DiBenigno is not a night owl by nature, and her experiments with coffee, Red Bull, and five-hour energy led to greater empathy and appreciation for the workers keeping a hospital running 24/7. The second challenge was developing informant relationships, particularly with the lower-status PCTs. DiBenigno followed Kellogg’s advice to always “start at the bottom” by beginning the study shadowing PCTs and observing collaboration with nurses from their point of view before switching to shadow nurses, since the lower status actors (PCTs) might not trust her if it seemed like she was hanging around with the higher status actors (nurses) who were superordinate to the PCTs. This turned out to be great advice, as the PCTs quickly warmed to sharing their experiences with DiBenigno and were touched that anyone was interested in their work experience given that they were largely seen as “the invisible help.” Once DiBenigno switched to shadowing nurses, it became more difficult to maintain PCT relationships on Unit B because they often saw her spending time with the nurses and became concerned about whose side she was on. However, while the lower status PCTs were highly aware of where DiBenigno was spending her time, the higher status nurses were not. In fact, during the time when DiBenigno shadowed the PCTs, the Unit B nurses had hardly noticed her and some reported upon being asked to be shadowed that they had never seen her before even though she had been on the unit for months shadowing PCTs (this also reflects the constant influx of rotating staff, including interns, residents, phlebotomists, hospitalists, and floating nurses and PCTs regularly passing through the units). This gave us a first-hand appreciation for the invisibility of the PCTs on Unit B, who spent most of their time in patient rooms out of sight of most other hospital staff.

Question 2. A striking aspect of your project was the strict policing of deviant behavior in Unit B, where occupational roles were strongly correlated with demographic group membership i.e. strong faultlines. We were interested in the nature of power within such groups. Our reading of the study suggests that this was driven by strong in-group norms and practices. However, we were curious as to whether these were also accompanied by leadership aspects, as in were there certain group members who would lead or particularly enforce these norms? Conceivably, without such active reinforcement, there could be greater utilization of shared social identities even in cases like Unit B.

The question of power dynamics is an interesting one. Our data suggest that the consolidated demographics on Unit B generated poor cross-occupational collaboration regardless of leadership by particular group members. For example, as you suggest, there certainly were some nurses on Unit B who were more influential than others and served as informal leaders on their units. Yet, we do not believe these leaders were driving or exacerbating the failed collaboration we observed for two reasons. First, nurses and PCTs worked only three nights a week, and sometimes rotated which nights they worked, so there was typically a different mix of nurses and PCTs each night, minimizing the impact that one leader had on group dynamics. Second, we did not observe variation in cross-occupational collaboration practices that correlated with the presence of any of these more influential nurses working the shift.

And what about leaders among the PCTs on Unit B? In this context, PCTs had extremely little power to influence group dynamics, especially with the nurses. They were considered a low-cost, low-skill, replaceable labor pool. PCTs who reported having “fought” mistreatment by nurses at one time or another told us that they had quickly learned that the word of a nurse was privileged over that of a PCT, and that any PCT complaints to managers were typically ignored or backfired to reflect poorly on the PCT. Our discussions with the other staff showed that the word of PCTs was typically discredited by managers, nurses, and physicians, who regularly believed the word of patients over PCTs (while a nurses’ word was typically believed over a patient’s word), so the PCTs were very vulnerable and typically became withdrawn and quiet around nurses and physicians to avoid saying anything that could be held against them. For example, we saw a nurse find out from a patient that they wet the bed and then ask whether the PCT had failed to ask the patient if they needed to use the bathroom during one of the PCT’s checks. And even though the PCT had indeed asked the patient, and told the nurse as much, the word of the patient (many of whom suffered from dementia or were on powerful medications that affected their cognition) was believed over the word of the PCT.

Question 3. In an interesting illustration presented in the study, you show how nurses in Unit B rarely expressed positive emotions with the PCTs, while PCTs in Unit B suppressed their negative emotions with nurses and expressed positive emotions only in their interactions with other PCTs, indicating a lack of mutual liking and trust between the two groups. From your observations, did dyadic interactions over a length of time only serve to entrench this dislike and distrust or were there instances of dyad members managing to make a deeper connection but being afraid to express this due to the existing norms? What organizational toolkit might be useful for dismantling such norms? For example, were there reward systems in place where PCTs could nominate nurses for nursing excellence awards?

In general, we observed a negative reinforcing cycle of distrust among nurses and PCTs on Unit B leading to entrenchment over time, as your suggest. In our observations, attempts to break this cycle were few and far between, and this was related to the dynamics generated by the consolidated demographics on Unit B, including a negative view of traditionally devalued identities, high occupational in-group pressure, and feelings of dislike and discomfort between the nurse and PCT groups. As an illustration, in the paper, we describe the experience of one nurse who moved from Unit A to B. Nurse16, who was young, white, and U.S.-born and had recently graduated from a local nursing school near the hospital, worked on Unit A for a few months before switching to Unit B. On Unit A, she had collaborated well with most of the PCTs she worked with. We observed her make the transition from Unit A to Unit B. During her first week on Unit B, she wore colorful and festive scrub outfits typical of those worn by both nurses and PCTs on Unit A but only by PCTs on Unit B. During this week, we also observed her attempt to find commonalities with the PCTs she worked with, talking with one of the younger PCTs about a popular musical group. We saw her openly praise and thank one of the PCTs in front of the other nurses. Unfortunately for Nurse16, her actions were condemned by the other nurses on Unit B. We saw the other nurses on Unit B gossip about her “inappropriate” behavior, and we saw them avoid helping her in several situations even though they went out of their way to help other nurses. During this first week, she sat alone on the periphery of the nursing station, and the other nurses did not include her in social conversations. Over the course of several shifts on Unit B, we saw Nurse16 distance herself socially from the PCTs. She also started conforming to the norms of her Unit B occupational group and began exclusively wearing plain blue scrubs like the other nurses. When we left the field, she was demonstrating occupationally prescribed behavior and had gained acceptance from her fellow nurses on Unit B. Thus, even a nurse who had experience making meaningful connections around non-occupational identities with PCTs was unable to over-ride the strong occupational in-group pressure regarding how to engage with PCTs on Unit B.

In terms of what organizational toolkits could improve collaboration and relations between nurses and PCTs, one of our man findings was that the organizational tools the hospital had put in place to help nurse-PCT collaboration was insufficient for cross-occupational collaboration to occur. First, rules and routines were in place on both units that governed the order and timing at which certain patient care tasks needed to occur. These routines were designed so nurses and PCTs could understand when the other group’s workload was heavy so they could offer assistance during these times in order to meet ongoing deadlines. Second, boundary objects existed to facilitate collaboration between nurses and PCTs (e.g., patient charts, electronic patient records, pagers, etc.). Such boundary objects were designed so nurses and PCTs could be aware of when the other required help. The hospital also did not require nurses and PCTs to dress differently, which minimized status differences. Finally, common spaces existed to allow for informal information sharing between nurses and PCTs. Yet, despite having these organizational tools, collaboration failed on Unit B and succeeded on Unit A. It is possible that these organizational tools did not facilitate cross-occupational collaboration because of the widespread distrust between demographic groups not only inside the hospital but also in society more broadly. Our study suggests that one of the things the organization might do to facilitate cross-occupational collaboration would be to use creative selection and retention strategies to construct a unit structure that has cross-cutting demographics between nurse and PCT groups.

Question 4. In your ethnography, you found that 0% of the nurse-PCT dyad members in Unit B had shared race, 12% had shared age and 7% had shared immigration status. They did not use dyadic toolkit in work interactions, in sharp contrast to 81% using dyadic toolkit in Unit A where the percentages for shared race, shared age and shared immigration status were 39%, 55% and 61% respectively. One of the key takeaways from your work is that organizations should pay close attention to diversity across all levels of the occupational hierarchy. For managers strategically aiming to increase diversity across levels, what should be some of their key considerations? For instance, could low in-group pressure due to high degree of intersecting group affiliations have unintended effects on group cohesiveness?

On the one hand, our findings demonstrate that top managers who want to promote cross-occupational collaboration should consider constructing a cross-cutting social structure so that cross-occupational dyads can broaden their dyadic toolkits and use them to work across occupational differences. On the other hand, in practice, given strong occupational clustering by race, gender, age, and immigration-status, it may be difficult for top managers to do this. Thus, to promote cross-occupational collaboration, top managers may need to use innovative selection strategies not merely to attract occupation members with varying demographic characteristics, but also to ensure they are evenly distributed across different areas and levels of their organizations. The prescription here is not for top managers to “increase diversity” in general, but for them to attend to issues of status and hierarchy by creating organizations with cross-cutting demographics. In practice, this would mean not only paying attention to diversity at the top of the organization, but across all levels. For example, physician and nurse collaboration has a long history of being fraught with challenges, many of them due to consolidated gender dynamics. While great effort has been made so that today more than a third of all physicians are women, much less attention has been paid to recruiting men to go into nursing, which our research suggests could create cross-cutting demographics that could improve physician-nurse collaboration. Attending to cross-cutting demographics across all levels of the organization, not just at the top, could generate the improved cross-occupational collaboration essential for achieving good outcomes in many organizations.

You also ask whether there are any unintended consequences of cross-cutting demographics in terms of reducing group cohesiveness that managers should be aware of. Diversity has been shown to reduce group cohesion but increase innovation. The general prescription from the diversity literature is that diversity is especially valuable when a group is charged with coming up with novel ideas, but that diversity can be a double-edged sword that may impair group performance on tasks requiring cohesion for communication and coordination unless certain conditions are present to offset these negative effects. Our research challenges the notion that diversity necessarily has negative consequences for performing work that requires high levels of communication and coordination, such as patient care. As noted above, cross-cutting demographics are not the same as diversity, and many diversity studies have not taken into account whether demographics are cross-cutting versus consolidated. Regarding cohesion, we did observe less cohesion among the Unit A nurses compared to the Unit B nurses. However, our research suggests a “dark side” to group cohesion, especially in a context where different groups (here, occupational groups) must work collaboratively to accomplish interdependent work. Cross-cutting demographics may provide an antidote to the intense in-group pressure generated by homogeneous groups, especially when the goals of an in-group conflict with those of the organization. While cohesion may have benefits, we think that it is important to consider from whose point of view cohesion matters. While for the Unit B nurses, the high levels of cohesion may have helped them enjoy their work more, from the organization’s point of view or the patient’s point of view or the PCTs’ point of view, such intense cohesion among the Unit B nursing group had negative effects. In this way, cross-cutting demographics can be thought of as an antidote to the formation of strong identity-based in-groups that may be detrimental to cross-occupational collaboration.

Question 5. What question have we missed? Please ask yourselves a question you wanted us to ask, and answer it.

One thing that surprised us when doing this study was how little attention cross-occupational scholars had paid to demography, to the point of rarely mentioning the demographic composition of the occupational groups studied. While our research suggests that demographic differences between occupations may exacerbate collaboration challenges, existing scholarship has suggested that cross-occupational collaboration challenges are due nearly entirely to differences in expertise, status, and meanings between occupational groups. One question worth thinking about is how prior research may require reinterpretation if demographic differences that generally correspond with occupational differences are taken into account. For example, we are never told the demographic characteristics of occupational groups in the cross-occupational collaboration between Barley’s (1986) radiologists and technicians or Faraj and Xiao’s (2006) trauma team members. And, while Kellogg, Orlikowski, and Yates (2006) and Bechky (2003a) refer to the demographic differences among members of the different occupations they study, they do not investigate the implications of these demographic differences for cross-occupational collaboration.

To address this issue, for one of our rounds of revisions, we went back and used Census data from the Bureau of Labor Statistics to estimate the demographic composition of various occupational groups that were the focus of prior important studies of cross-occupational collaboration. These data suggest that, although these scholars of cross-occupational collaboration have not investigated the impact of demographics on collaboration, demographics may have shaped the collaboration they observed. For example, gender may have impacted the cross-occupational collaboration between Barley’s radiologists and technicians; according to the 1990 census for the region of the Northeast where this study was conducted, 92 percent of radiologists were men while 82 percent of radiology technicians were women. Similarly, age and gender may have played a role in the collaboration among members of the medical trauma teams that Faraj and Xiao (2006) studied. According to the 2000 census and a study of US Physician Characteristics and Distribution (Freeman, 2001), during the time of their study, most trauma surgeons in the US were men over age 35, medical residents were nearly half men and half women under age 35, and most nurses were women over age 35. Gender differences may have shaped the cross-occupational collaboration between Information Technology and Creative workers in Kellogg, Orlikowski and Yates’ (2006) study. The authors mention that nearly all IT workers were men and invested in a particular form of “geek” masculinity while nearly all of the Creative workers were women (which aligns with the gender breakdown of 68 percent men in IT versus 66 percent women in Creative work from the 2000 census), but they do not explore how these gender differences affected the cross-occupational collaboration between these IT and Creative workers. Finally, race may have impacted collaboration between the engineers, technicians, and assemblers Bechky (2003b, 2003a) studied; the demographic breakdown that Bechky (2003b) provides shows that engineers and technicians were primarily white (60 percent for engineers and 75 percent for technicians) while assemblers were all non-white. But, Bechky does not investigate how these race differences may have affected cross-occupational collaboration between engineers and technicians on the one hand and assemblers on the other.

To be fair, since most occupations are highly segregated by demographic characteristics, researchers are likely to be studying cross-occupational collaboration in a consolidated setting, so it is difficult for them to observe or even conceive of how cross-occupational collaboration might be different if there were cross-cutting demographics. Our unique dataset allowed us to account for the impact of demography on cross-occupational collaboration because of the unexpected demographic makeup of one of the two hospital units we studied—Unit A—which was cross-cutting in terms of the race, age, and immigration-status of members of the different occupational groups.

Thanks for your interest in our findings and for the opportunity to answer these interesting and insightful questions!

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: