Anyone searching for signs that their friends merely tolerate them is already working from a premise the research on this exact question tends to complicate. Before looking for evidence that a friendship is one-sided, it is worth knowing what psychologists have actually found when they measured how much people think their friends like them against how much those friends really do.
The short version: most people are wrong, and wrong in a specific, consistent direction. They underestimate it.
The liking gap, and why it shows up almost everywhere researchers have looked
In a 2018 paper in Psychological Science, the psychologists Erica Boothby, Gus Cooney, Gillian Sandstrom, and Margaret Clark had strangers hold short conversations, then asked each person how much they liked their conversation partner and, separately, how much they thought their partner liked them. Across several studies, people consistently rated how much they were liked lower than their partner’s actual rating. The researchers called this the liking gap, and found it in first meetings between strangers, in first-year college dorm mates getting to know each other, and in relationships that had continued for several months.
A related 2021 study, by Adam Mastroianni, Gus Cooney, Erica Boothby, and Andrew Reece, extended the same question to groups rather than one-on-one conversations, testing a simulated lab group, a college engineering team, and a work team inside a company. The pattern held in all three settings: people in the group reliably underestimated how much their fellow group members liked them, and the researchers found this had a measurable effect on how satisfied people felt about working with their team.
This is a consistent finding across several separate studies rather than a single result, though it draws mainly on self-report measures collected shortly after the interactions being studied, and it does not measure whether the same bias holds up identically for friendships that have lasted for years or decades rather than months. Taken together, though, it points to a specific problem with trying to read “signs” of being merely tolerated: the mental instrument most people use to gauge how much a friend likes them appears to run consistently low, not neutral.
A separate, more measurable problem: friendships that genuinely are not mutual
None of this means every friendship is actually as warm as both people secretly feel. A different and more concrete line of research shows that a meaningful number of friendships really are one-sided, in a way that does not depend on anyone’s felt impressions at all.
In a 2016 paper in PLOS ONE, the researcher Abdullah Almaatouq and colleagues mapped out friendship networks by asking participants to name their friends directly, rather than asking them to rate how close a friendship felt. Comparing each person’s nominations against the nominations made by the people they named, the researchers found that only about half of the friendships people assumed were mutual were actually named by both people involved.
This is one dataset, drawn from specific study populations rather than a representative sample of friendships in general, and it measures something different from how a friendship feels day to day: whether both people would independently list each other as a friend at all. But it establishes something the liking gap research cannot: an objective, network-level rate of non-reciprocity, measured from what people actually reported about their own social ties rather than inferred from a mood or a missed text.
Why this matters for anyone hunting for signs
Put these two findings side by side and the honest picture is more complicated than either “you’re probably just anxious” or “trust your gut.” The liking gap research suggests that the felt sense of being merely tolerated, read from tone, response speed, or a friend seeming distracted, is exactly the kind of signal that tends to be biased toward the pessimistic reading, across strangers, dorm mates, and work teams alike. The Almaatouq findings suggest that real non-reciprocity does exist at a meaningful rate, but that it shows up most clearly in structural facts, who actually reaches out, who remembers plans, who is named back when asked directly, rather than in impressions formed from a single interaction or a passing mood.
Neither study offers a checklist for diagnosing a specific friendship from a handful of anecdotes, and neither was designed to settle whether any particular relationship a reader has in mind is genuinely mutual. What the research does support is a narrower and more useful distinction: the vague, ongoing feeling of being tolerated rather than liked is a poor guide on its own, given how reliably that feeling runs low even among people who like each other quite a lot. Patterns in actual behaviour over time, who initiates contact, who shows up, who is named back as a friend when it counts, are a more solid basis for judging a specific friendship than a private sense of unease.