Can asking more questions make people like you?

In short text chats, prompting more questions modestly increased partner liking.

But the evidence does not support a quota, an interview technique or a shortcut for manufacturing charm. The experimental benefit was modest and belonged to the person receiving the questions. In the field data, follow-up questions—not questions in general—carried the useful signal.

The distinction is simple: a question matters when the next turn shows that the answer changed where the conversation went.

Two experiments changed how many questions people asked

Huang and colleagues studied short getting-acquainted conversations between strangers.

In the first experiment, 398 people formed 199 pairs. One person in each pair was randomly instructed to ask many or few questions during a 15-minute text chat. The other person did not know about the instruction.

The uninstructed partners liked high-question askers more than low-question askers. The difference was statistically detectable but modest.

A second experiment analyzed 338 people in 169 pairs. This time, both partners received independent instructions to ask many or few questions. Again, the number of questions asked by a person’s partner predicted liking. Asking more oneself—and whether the two assigned levels matched—did not produce the same result.

The experiments therefore support a bounded causal answer: changing question asking changed how the recipient evaluated the asker in these short text chats.

The benefit was experienced, not simply observed

A separate group of 612 people read transcripts from the second experiment. These outside observers did not like high-question askers more.

That null result did not support the interpretation that asking more made someone look generally more likable to an outside reader. The measured advantage appeared inside the interaction, for the person whose answers received the attention.

Participants who asked more were also perceived as more responsive: listening, understanding, validating and caring about what their partner said. Responsiveness statistically explained part of the liking difference.

That analysis supports a mechanism. It does not prove that responsiveness was the only cause or reveal an asker’s private intention.

Follow-up questions carried more information than the count

The researchers classified questions by what they did in the conversation.

A follow-up question stayed with something the partner had just said. A full-switch question moved to a new subject. Mirror questions returned a question without necessarily developing the answer. Introductory and rhetorical questions served still other jobs.

Follow-up questions were most consistently associated with responsiveness and partner liking.

That is why ask more questions is incomplete advice. Nine disconnected prompts can feel like a survey. One precise follow-up can signal attention to an answer.

Speed dating found the same pattern without proving its cause

The field study used face-to-face speed-dating transcripts from 110 people and 1,961 yes-or-no second-date decisions.

Daters who asked a higher proportion of follow-up questions were more likely to receive second-date interest. Other question types did not show the same relationship.

This was a behavioral outcome in a real social setting. It was not a randomized test. Question habits, attractiveness, compatibility, conversational skill and other characteristics could move together. The authors explicitly warned that the speed-dating association cannot be interpreted causally.

The field result is consistent with the experiments. It does not convert them into a promise about any date.

The 2025 correction changed numbers, not the answer

In March 2025, the journal published a correction after an independent integrity and reproducibility audit found several minor reporting errors.

The correction states that every substantive hypothesis-test result remained unchanged, the paper’s conclusions remained valid and the audit found no reason to doubt the integrity of data collection.

That matters because much of the current search coverage still repeats the 2017 headline without mentioning the updated record.

It also needs one boundary. The audit materials document limits around some source logs and human annotation steps, and an audit is not an independent replication of the social effect. The right conclusion is neither dismissal nor blind certainty: the corrected analyses preserve the original bounded result.

A question can still reveal self-interest

A separate 2025 paper tested boomerasking: asking someone a question, letting them answer and immediately turning the conversation back to one’s own answer.

People perceived boomeraskers as less sincere and preferred partners who disclosed directly. The question created an expectation of interest; the instant return exposed a different motive.

This later study does not replicate the 2017 experiments. It supplies the most useful counterexample. More questions are not automatically more responsive.

Primary sources

1. Huang et al., 2017
2. APA correction, 2025
3. Independent audit materials
4. Yeomans et al., 2019 author reply
5. Brooks and Yeomans, 2025 boomerasking experiments

Independent editorial summary. The authors are not affiliated with LifeScore.