By Lee Flanagan
✨ AI Summary:
- Communication cues and delivery style predict first impressions far more strongly than resume content or GPA, creating lasting bias even in structured interviews.
- First impressions formed in the first few minutes correlate 0.51 a month later, and averaging scores across all interview questions lets early rapport-driven answers unfairly influence final hiring decisions.
- Structured interview scorecards that don’t distinguish between rapport-building and substantive questions fail to remove first-impression bias and should be audited for which answers carry equal weight.
- Rate candidates frequently throughout the interview, use multiple raters, and explicitly score first impressions only when impression-making is genuinely part of the role; otherwise discount early exchanges from the average.
Picture the scene Brian Swider’s own research describes. A candidate exchanges a few minutes of small talk with the hiring manager, then faces a structured interview built from the same 12 questions, asked in the same order, scored on the same scale, every time. On paper, that is as fair as an interview process gets. Swider’s data says the small talk beforehand was already strongly linked to how the interview went.
The Resume Lost, And The Researcher Did Not Expect It
Swider, a professor at the University of Florida’s Warrington College of Business, led a meta-analysis of 204 independent samples across 145 studies, spanning different jobs, countries and education levels. His team set out to compare three inputs into a first impression: communication style, physical appearance, and content, meaning the actual substance of what someone says or what sits on their resume. Swider expected content to win.
“If you asked me before the study, I would have said content, things like a candidate’s GPA or what’s listed on their resume, would be the strongest, most consistent predictor of a first impression,” he told HRD Canada. “It turns out that was the weakest.” Communication cues, the verbal and nonverbal register of tone, facial expression, smiling, leaning in and hand movement, came out strongest. Appearance landed in the middle, weaker and more inconsistent, partly because it is not relevant to every job and partly because people disagree more about what looks good than about what sounds like effective communication.
A First Impression That Outlives The Interview
This is a study of how impressions form in general, not a hiring study alone. The leap to recruitment is our interpretation of what it means for interviews, not Swider’s own claim, but the durability numbers make that leap hard to avoid.
Swider’s team found a 0.51 correlation between an impression formed in the first few minutes and the same impression at least a month later, a figure that clears the threshold researchers use to call an effect strong.
The impression’s power to predict actual outcomes, hiring decisions and performance ratings, does soften over time: from roughly 0.5 in the first few minutes to roughly 0.3 after a month. Swider called that a modest decline given how much extra information people gather in that window. Read plainly, the impression barely moves. Only its accuracy drifts.
Structure Did Not Stop The Leak
Here is the finding that should give pause to anyone who treats “we use structured interviews” as a closed case. Swider’s team tracked a fully structured process from a separate 2016 study: 12 identical questions, identical order, identical rating scale, for every candidate.
“What we found was that the first handful of questions were really strongly related to the impression that they made in their little chitchat rapport-building conversation before the interview starts,” Swider said. “It was basically unrelated to the last handful of questions.”
Because interview scores are averaged across all 12 questions, that early bump or dip carries real weight. Candidates hovering near the cutoff for the top 20 percent, the threshold for a second interview, got pushed over it or knocked below it based on nothing more than how the small talk went. Structure standardizes the questions. It does not standardize what happens to the answers once they are averaged into a single score.
The Exception That Should Worry You, Not Comfort You
Swider offers one counterexample worth sitting with. One organization with heavily customer-facing roles formally scored candidates’ first impressions as part of the interview, on the logic that making a good impression is part of the job. Its own data showed candidates who scored well on that measure tended to perform better once hired.
This is one company’s fix for one narrow role profile, not a formula the rest of us should copy. Most employers are better served limiting how much weight a first impression carries than embracing it. The sharper lesson is this: that organization’s approach succeeded because it made the criterion explicit, scored it deliberately, and checked it against outcomes. Most interview scorecards do none of that. They let a first impression color every answer that follows, unexamined, and call the result objective because a scorecard existed.
What The Audit Actually Needs To Check
Swider’s own fix is reasonable as far as it goes: rate candidates more frequently through the conversation rather than once at the end, bring in multiple raters, and discount the earliest exchanges when they are clearly rapport rather than substance. Even he is clear that none of this removes the effect.
Scorecards almost never note which question came before rapport had settled and which came after. That gap means nobody can tell whether an early answer and a late one carry equal weight in the final average. If your scorecard cannot answer that question, treat every score on it as suspect until you can.
“We’re cognitive misers,” Swider said. “We want to use as little brain power as possible. Thinking is hard. First impressions give us those shortcuts.” That instinct does not switch off because a scorecard exists. It moves into the average, where nobody is watching for it.
Our read is that a scorecard averaging a rapport-driven opening question with a substantive closing one is deciding who moves forward on the strength of an impression that has nothing to do with the job.
Original reporting: hcamag.com.
Frequently asked questions
What did Swider’s meta-analysis actually measure?
It analyzed 204 independent samples from 145 studies across different jobs, countries and education levels to compare how communication cues, physical appearance and content, meaning a resume or GPA, predict the first impression one person forms of another. It is a general study of impression formation, and the hiring implication is our interpretation, not the study’s own claim.
Does using a fully structured interview eliminate the halo effect?
No. Swider’s separate 2016 study of a 12-question structured interview found the earliest questions were still strongly tied to the rapport-building chat before the interview began, while the later questions were basically unrelated to it. Because scores get averaged, that early bump or dip moved candidates across the cutoff for a second interview.
Is there a legitimate case for scoring first impressions directly?
Swider describes one organization with heavily customer-facing roles that formally scored candidates’ first impressions as part of the interview, and its own data showed those scores tracked later performance. This works as an outlier rather than a template, since most jobs do not build impression-making into the role itself.
What counts as ‘content’ in this research, and why did it perform worst?
Content refers to the substance of what a candidate says or what appears on paper, such as GPA or resume detail. Swider expected it to be the strongest predictor of a first impression before running the numbers; the meta-analysis found it was the weakest, with communication cues dominating instead.
What fix does Swider recommend, and does it fully solve the problem?
He recommends rating candidates more frequently throughout the interview rather than once at the end, using multiple raters instead of one person’s judgment, and discounting the earliest exchanges when they are clearly rapport-driven rather than substantive. Swider is explicit that none of these tactics remove the effect entirely.