Hiring managers often assess prospective employees for signs that a new hire’s attitude and values line up with those of the organization—so-called “culture fit.” But not only can this demand for alignment based on personal values risk racial, gender, and class stereotyping, new research suggests it’s not the only important way that employees can fit in on the job.
A new paper by Berkeley Haas professors Jennifer Chatman and Sameer Srivastava and Stanford’s Amir Goldberg, along with Richard Lu, PhD 19, shows there’s another side to culture fit that’s also important to success at work. And, in contrast to cultural fit based on values, it can be learned over time.
“You could take a gamble on someone who’s good at perceiving what’s going on and is willing to emulate it,” said Chatman, a pioneer of organizational culture research. “That opens up all kinds of options for changing past hiring practices.”
The paper, forthcoming in the journal Organization Science, explains that “culture fit” appears to consist of two separate components, each associated with different job outcomes. The first aspect is what the authors termed “value congruence,” which refers to the match between an employee’s values and those of an organization. Value congruence—which is what people most often mean when talking about “culture fit”— appears to influence deliberative choices such as whether to stay at a job or leave.
But there is another force at play, which the researchers call “perceptual congruence.” This is the degree to which employees are able to accurately perceive an organization’s culture and adjust their behavior to fit in. Perceptual congruence has more bearing on a person’s day-to-day behavior, like whether they use similar words and expressions as their colleagues.
Both types of culture fit are linked to career outcomes. But when it comes to work performance—as measured by monthly bonuses—perceptual congruence is especially important.
The findings could complicate how managers use the concept of culture fit to make and explain their hiring decisions.
Melding two measures
Past research has consistently shown that higher levels of culture fit are associated with increased productivity, stronger job commitment, and lower turnover. Both Chatman and Srivastava, who together founded and co-direct the Berkeley Culture Center, have made it a research focus. Chatman’s research in the area began with her dissertation, which explored culture fit at the top accounting firms. She went on to co-develop a tool for quantitatively assessing culture fit that became among the most widely adopted research tools in the field.
As popular as the survey tool is, it is labor intensive—requiring time from researchers and employees with every round of data gathering.
The impetus for the new paper was, in part, to test a new methodological approach to measuring culture fit that uses the power of machine learning based on linguistic analysis of digital trace data—such as emails. Srivastava created the linguistic analysis tools in close partnership with co-author Goldberg of Stanford, with whom he co-directs the Computational Culture Lab.
Their analysis tool, called the Interactional Language Use Model (ILUM), funnels the words in emails into one of 64 linguistic categories and then examines the degree of overlap in linguistic styles between each individual and their peers. “It basically asks such questions as, ‘Are you somebody who uses a lot of swear words? Are you somebody who uses a lot of emotional language or are you very logical in your expressions?’” Srivastava explains. “It turns out that those styles are influenced by organizational culture, and those who fit in behaviorally are able to align their styles with those of their peers.”
Chatman agreed that Srivastava’s methodological approach to studying culture came with big benefits—like the ability to capture variations in culture fit over weeks or months without the burden of repeated survey collection—but she had her doubts as to what his tools were really measuring. “I challenged their approach because I was worried that they weren’t necessarily picking up culture—it could be that the words people use in emails may fail to accurately represent organizational culture. I wondered whether language embeddedness is, in fact, a reasonable representation of culture fit.”
The research team worked to bring both approaches together and attempted to cross-validate them. To do so, they wielded multiple tools, administering parts of Chatman’s survey to the employees of a mid-sized technology firm and also analyzing eight years of internal emails from the same firm using Srivastava’s and Goldberg’s linguistic analysis methods.
“There was great convergence between the two approaches, which I think suggests that the natural language processes approach is a valid way of assessing organizational culture,” Chatman says. “There’s great promise in doing that, because you can use email data unobtrusively, and it opens up this world of data collection in the culture domain that’s really important.”
Srivastava adds that bringing the separate culture-fit measures together also allowed for some early validation of a new machine-learning model that can start to infer employees’ culture-fit trajectories over time based on, say, their responses on a single survey.
The Cost of Faking It
This cross-validation of the distinct measures of culture fit was how the researchers discovered that two different types of culture fit mattered, each influencing different work-related behaviors.
Chatman admits she was surprised by the strong impact of perceptual fit, especially on employee success at work. In her mind, this finding opens a set of pressing questions.
For example, one implication is that “it is possible, given these findings, that a person could have very high perceptual fit and very low values fit, and stay and succeed at a job for a long time,” Chatman says. However, “That’s a kind of inauthentic way of existing at work.”
Although perceptional fit is linked to job success, Chatman predicts there’s a likely emotional cost to “faking it” at work. Indeed, in the male-dominated company that provided the data for the paper, women’s value congruence was significantly lower than that of men. “What are the long-term implications? Because the study was not that long, that question is still open. Do we know how long people can fake it?”