Can Architectural Design Counteract Friending Bias?

Architects may say so, but scientific research does not (yet).

~1,700 words, a nine-minute read

Architecture made a curious appearance in the slew of recent articles that came out after the latest Raj Chetty-led megastudy on socioeconomic mobility. The papers can be downloaded for free (see part I and part II here), and in the interest of brevity, I’ll quickly summarize it so I can move onto the design-related issues, which are unsurprisingly not the main focus of papers themselves.

In short, based on a massive dataset pulled from 72 million Facebook users, the study’s most headline-worthy conclusion is basically that: 

. . . economic connectedness (EC) — the share of high-SES (socioeconomic status) friends among low-SES people—is strongly associated with upward income mobility, whereas other forms of social capital are not.

So the thinking goes, and as well explained in summaries of the research like this one in The New York Times (NYT) morning newsletter: create better connections between otherwise economically separated class groups, especially for school-age kids, and upward mobility will improve. It’s a straightforward and logical enough idea, but the papers point out two challenges related to increasing EC. The first is with “exposure”, or the difficulty in merely providing a setting where there is realistic potential for lower and higher SES groups to connect with each other. The second is “friending bias”, or the tendency of lower or higher SES individuals to associate only with other people in their group. 

Given the enormity of the study — scholarly analysis of massive datasets can sometimes lead to statistically significant yet overly generalized, abstract conclusions — and the nature of the subject matter, it was something of a surprise when I read this excerpt from the NYT write-up linked above:

The school’s open architecture also encouraged serendipitous socializing. “Accidental, unstructured interactions between students was a very high priority,” John Diffenderfer, one of the school’s architects, said.

This bit was part of a larger discussion about one of the rarer cases of a high school in Fairfield, California where there is evidence of higher amounts of cross-SES interactions. Before the point about the open architecture, it’s explained that the district also has an unusual boundary that includes a mixture of richer and poorer areas. With that in mind, a critical, built-environment related question naturally follows: can architectural design have an impact on cross-SES socializing? “Friending bias” is pretty clearly some form of human characteristic, but “exposure” is more of an environmental property. By including a diversity of incomes, the district boundary is certainly a key part of this, but are there tweaks that can be made in the architecture to further catalyze the exposure effect?

Perhaps unlike Mr. Diffenderfer, I have reservations about the extent to which building design influences complex human behaviors such as cross-class interactions.

This is not to say by any means that I think design and architecture are unimportant for many other (important) reasons. Indeed, those reasons could fill (and have filled) libraries worth of elaboration. If I were to properly expand on this point, it would take at least an entire Substack article. For now, just this little paragraph will have to do. I make this hedge explicit as it is extremely important that my position not be misunderstood in the midst of an article which otherwise takes a more critical stance. 

What I am saying, though, is that scientific evidence on the relationship between design and human interactions — and specifically, the cross-SES interactions that Chetty and his team discuss — does not appear to support the implicit claims made by some design professionals. Of course, basic provisions in school buildings like wide hallways and flexible outdoor areas are obviously conducive to “serendipitous socializing”. But almost all schools have these things, and it does not take an extraordinary feat of design excellence to incorporate them. They’re basically required components of a school by definition. 

Beyond that, the excerpt about building design from the NYT piece above may perhaps lead one to believe that it’s something about an “open” design strategy that EC-seeking schools ought to be striving for. As an adjective, “open” is sufficiently vague to leave room for interpretation that may best be clarified by actual design drawings, which are not available for the Fairfield school in question — perhaps due to understandable safety concerns. But in the absence of a more satisfactorily precise definition, what do scientific studies say about open design and human interactions? For the purposes of this post, I will look at two sources.

Let’s start with the Chetty et. al papers themselves, where the word architecture appears exactly twice in about 61 pages of dense text. That would be near the end of the second paper, in the discussion section. Here it is noted that while “interventions to reduce friending bias have been studied less extensively”, strategies around spatial planning are identified as one of three interventions that may benefit from further study.

And it’s under this bullet point that another quote from an architect pops up, discussing a design strategy whereby a school replaced three smaller cafeterias with one large cafeteria. The idea was that placing everyone in one big space would alleviate the natural sorting that three smaller rooms would necessitate:

One of the architects described the project’s goals as follows: “shrink income-based inequalities in education by designing schools that improve the way students learn and socialize”, noting that “though students may still split into their own cliques [...], they’ll have more opportunities to cross paths and interact with peers from other social groups”

That brief statement reads like the perfect setup for a completely feasible, real world study on the impact of going from three cafeterias to one. To be fair, the architect’s website does indicate some anecdotal evidence on an improved social scene at the school. Still, the question remains: did students still “split into cliques”? In the absence of a truly scientific study with a systematic methodology and a published paper, we are once again left to simply take the architects word for it. 

Unfortunately, studies of that sort on design and student interactions in K-12 environments are scant. Given the thoroughness of the Chetty et. al studies, one would expect to see such studies cited there if this were not the case. This may be due in large part to Institutional Review Board (IRB) requirements that have (understandably) especially stiff restrictions on involving children as research subjects. Still, given the attention that these new papers have gotten, perhaps more work will be done in the future. For now, we must take a slightly different angle into the search for evidence on architectural openness and human interaction. 

Now it nearly pains me to say it, because it’s pretty solidly achieved cliché status by now. But folks, it’s time to talk about the future of work and office design. I will do my best to make it exciting.

Because it is, actually! (I think anyway.) It turns out that a very interesting study on open office layouts may prove to be instructive on these matters. Published in 2018, “The impact of the ‘open’ workspace on human collaboration” opens with an epically clear introductory setup question (emphasis mine):

Boundaries between ‘us’ and ‘them’ have long captured human interest. Yet even as social scientists continue to study the value of a vast array of boundaries, in an era in which the nature of work is changing, managers and organizational scholars have increasingly framed boundaries as barriers to interaction that ought to be spanned, permeated or blurred to increase collaboration. In the most physically salient and concrete example, ‘spatial boundaries’  at work—such as office or cubicle walls—are being removed to create open ‘unbounded’ offices in order to stimulate greater collaboration and collective intelligence. Does it work?

In the interest of brevity, here’s the punchline. After studying the impact of interoffice collaboration in two field studies where corporate headquarters transitioned from traditional walled or cubicle-based configurations to open office arrangements without dividers, the researchers found that (emphasis mine):

…the volume of face-to-face interaction decreased significantly (approx. 70%) in both cases, with an associated increase in electronic interaction. In short, rather than prompting increasingly vibrant face-to-face collaboration, open architecture appeared to trigger a natural human response to socially withdraw from officemates and interact instead over email and IM.

For you, the skeptical reader, this may raise logical follow up questions in relationship to the original issue at hand. Can conclusions about office design with office workers be logically extended to school design with school kids? What about the electronic communication thing, isn’t that its own form of interaction? What exactly is the state of electronic communication between kids during school hours anyway? (Some parents who would like to know!!) Should we be studying interventions on digital interactions in schools too? Classrooms aren’t at all divided in the way cubicles divide an office, so what defines an “open” school environment anyway?

Subsequent research may tell. In the meanwhile, rather than ending with the safe, neutral, and trusty “more work should be done on this issue”, I’m tempted to reflect on a slightly more intellectually stimulating idea. And that would be the age-old “people versus place” dichotomy.

Take the term “church” as a classic example, a word which Christian scholars are quick to point out has biblical origins in the Greek word “ecclesia”, which refers to a gathering of people. As in, not the physical building that nowadays more immediately comes to mind for most. And speaking of now, the technical distinction between church-as-building and church-as-congregation appears to have collapsed with both definitions being officially valid, according to the first and third definitions of the word in the Oxford Advanced Learner's Dictionary (sorry, it’s the best I can do since I don’t pay the annual $100 for the proper OED) :

church noun

1) [countable] a building where Christians go to attend services, pray, etc.

3) [countable] a particular group of Christians

So to the extent that a vision of increased upward mobility is pursued in practical, implementable terms, and given that the Chetty et. al papers have not cited scientific research on architectural design interventions, the logical conclusion would seem to be that policies operating in the vein of the latter, or “people-based” definition may be more effective. Of course, that probably assumes that classrooms remain medium-sized open rooms, cafeterias remain large-sized rooms, and wide hallways remain used as wide hallways. From what I remember of school anyway, I did a hefty amount of spontaneous collaborating in those chaotic 5 minutes between classes when the whole place dumped all of us into the corridors. But don’t take my word for it! I’m simply an architect eagerly awaiting more hard evidence.

© 2022 James Carrico