Ghosters Felt Worse Than the Ghosted Guessed in Two of Three Studies. The Live Test Found No Overall Gap.

Eight minutes into a text chat with a stranger, each person in the experiment got a choice: sit through a follow-up chat, or skip it without a word. Of 229 pairs that finished, 175 had one partner who skipped. Those silences are the data behind a paper in Personality and Social Psychology Bulletin, and they are the least tidy part of it.

The paper, by Sophia Li of the Haas School of Business at UC Berkeley, Coral Zheng of Cambridge's Judge Business School and Juliana Schroeder, asks something most ghosting research skips. Earlier work measures how it feels to be ghosted. This one asks how it feels to be the person who stops replying, and whether the person left waiting has any idea. Zheng is a PhD candidate in organizational behavior, and Europe PMC lists the paper as first published on August 31.

Two studies built on memory and imagination

Study 1 recruited 195 adults. Each recalled a specific online exchange with someone they knew who was not a close friend or immediate family. Half remembered ghosting someone, half remembered being ghosted. The ghosters rated their own negative feelings at an average of 2.27. The people recalling being ghosted guessed the other side felt 1.56. The gap was widest for unintentional silence, which Zheng described as people who "accidentally forget to respond," and the write-up on PsyPost adds that they may also miss the message or put off replying.

Study 2 moved to hypotheticals. Among 601 adults, those imagining they had gone silent expected to feel 3.23, while people imagining being on the receiving end predicted 1.94. Then came the fix. Add a short reply and the sender's expected discomfort fell to 2.33, while the receiver's prediction moved to 1.82, much closer. About a third of receivers assumed they had been annoying their friend after total silence. After a short reply, about one in five did.

The test with real people

Study 3 is the one that involved actual behavior, and it is where the tidy story frays. The researchers paired adults in an online chatroom for an eight-minute text conversation, then let one partner skip what came next without explanation. The abstract counts 350 people. PsyPost reports that people left waiting were told their partner might not return, which may have reduced the uncertainty that gives ghosting its sting.

The overall pattern did not hold. The only clear gap came from the subset who rated the behavior as ghosting: skippers scored 3.93, and the people waiting predicted 3.51. PsyPost lists that subset result among the study's limitations as an unplanned analysis. The abstract frames it more gently, saying the underestimation shrinks when communicators do not see their own silence as ghosting.

What the numbers can and cannot carry

Read the three studies side by side and the biggest gaps sit where nobody acted. Memory and imagination produced 2.27 against 1.56 and 3.23 against 1.94. Real behavior produced a smaller difference, and only after splitting the sample. Zheng's practical advice to the ghosted is that the silent person "probably feels worse than you expected." That may be true, but the evidence for it is strongest in what people say they would feel.

There are other limits. The live chats were short and between strangers, which cannot carry the weight of a long friendship or a stalled romance. Nobody in the experiment was asked to drop a date. They were asked to skip a chat with someone they had known for eight minutes, a lower bar for guilt. And none of the material reviewed here says how often ghosting happens, so the size of the problem is not measured in this piece.

Two things were not available. The full paper sits behind a publisher wall that blocked access, so the numbers here come from the abstract and the PsyPost write-up. And no outside researcher commented in the coverage reviewed, so the findings rest on the authors' own reading.

The part with the most practical pull is also the cheapest. A one-line reply cost the sender little in the imagined scenarios and softened the receiver's self-blame. Whether it works with real friends, in real silences, is a question this paper points at and does not answer.