Using Your Phone on a Date: What the Evidence Actually Shows

The phone on the table is the claim that did not survive

Start with the one everybody knows, because it is the one this site got wrong.

In 2013 Andrew Przybylski and Netta Weinstein published two experiments in which pairs of strangers talked with either a mobile phone or a notebook sitting on a nearby table. Nobody used the phone. Its mere presence, they reported, was enough to lower closeness, connection and conversation quality, and the effect was clearest when the topic was personally meaningful. That result is the origin of the advice you have read a hundred times: do not put the phone on the table.

It has not replicated. John Courtright and Scott Caplan pooled every mere-presence study they could find — six of them, Przybylski and Weinstein's two included, with a combined 601 participants. The average correlation was -.20 and it was not statistically significant (p = .359). Weighted by each study's reliability it fell to -.17 (p = .371). The heterogeneity test came back flat, meaning there was not even enough variation between the six results to go looking for a moderator. Separately, Carmen Linares and Anne-Laure Sellier ran a direct replication with 356 participants across two studies — groups of strangers talking for ten minutes with a phone or a notebook visible — and reported that no original relationship measure moved: in their dyads, all ps > .45.

The honest reading is narrower than 'debunked'. Six studies is a very small pile, and Courtright and Caplan calculate the statistical power of their own mere-presence test at roughly .11 — a test that weak would miss a real effect most of the time. So the finding is not that a visible phone is harmless. It is that after a decade of trying, nobody has been able to show it does anything, and the one large direct replication came back empty.

Two pages on this site said otherwise. The first-date mistakes piece called the phone on the table the one item on its list with real data behind it rather than folklore, which was exactly backwards, and the piece on a man's early signals treated face-up versus face-down as something you could read. Both were corrected the day this article went up, and both corrections are logged on our corrections page, which exists because our editorial policy commits us to saying it out loud rather than quietly patching it.

Picking it up during the conversation is the part that holds

The behaviour that does have evidence behind it is not the phone's location. It is picking it up while the other person is talking. The literature calls it phubbing, a portmanteau of phone and snubbing, and it has been measured well enough now to put a number on it.

The same Courtright and Caplan paper pooled 37 studies of phone use during an interaction, 7,515 participants in total. The average correlation with negative perceptions of the person using the phone was -.35, or -.29 once weighted by study reliability; their discussion rounds the first figure to -.34. Unlike the mere-presence set, this one was strongly heterogeneous — the studies genuinely differ from each other — but the direction is consistent across a decade of surveys and experiments.

The larger and more recent synthesis is narrower in scope and lands lower. Ni, Ahrari, Zaremohzzabieh, Zarean and Roslan pooled 52 studies across 58 samples and 19,698 participants, all of them about phubbing between romantic partners specifically. Their result for relationship satisfaction was rz = -.219 (95% CI -.26 to -.175). Intimacy quality came in at -.267, romantic relationship quality at -.201, perceived partner responsiveness at -.292, jealousy at +.289, and conflict over phone use at +.573, which is the largest thing in the table by a distance.

Translate the headline number. A correlation around -.22 is somewhere between a small and a moderate association. It is not nothing, and it is not the relationship-destroying force the coverage describes. It also carries a limitation the authors state plainly: the field is dominated by cross-sectional designs, so which way the arrow runs is an open question. People who are already unhappy with a partner may simply notice the phone more, and no meta-analysis of surveys can separate those two stories.

The effect lives in the perception, not in the behaviour

This is the finding that reorganises the whole topic, and almost nobody writing about phones and dating has caught up with it.

Katherine Carnelley, Laura Vowels, Sarah Stanton, Abigail Millings and Claire Hart ran a dyadic diary study on 100 couples in the UK: a lab session, then fourteen consecutive daily surveys from both partners, then a follow-up two months later. Crucially, they measured two different things. Perceived phubbing, meaning how much you thought your partner ignored you for their phone that day. And enacted phubbing, meaning how much your partner said they had actually done it.

Perceived phubbing predicted lower relationship quality the same day. The coefficient was B = -0.06, 95% CI -0.11 to -0.01, p = .021. That is a real effect and a very small one — in the raw correlations, perceived phubbing and daily relationship quality sat at r = -.05.

Enacted phubbing predicted nothing. Not the actor's own reported phone use, not the partner's, not on the day and not two months later. The correlation between a person's self-reported phubbing and their partner's relationship quality was .00. And the perceived effect itself did not survive to the two-month follow-up.

Michal Frackowiak, Peter Hilpert and Pascale Sophie Russell had found something compatible with 133 people diarised over seven days: participants did not report lower satisfaction on days when phubbing happened, and once two mediators were accounted for there was no direct daily effect at all. What moved was the appraisal — on high-phubbing days people rated their partner as less responsive and judged the behaviour more harshly, and those judgements, not the phone use, tracked end-of-day relationship quality.

So the variable carrying the damage is the reading, not the act. Which is uncomfortable in both directions. It means your date is not reacting to your phone, they are reacting to what they concluded from it. It also means the conclusion is doing work that the behaviour by itself does not support.

You are a poor witness to your own phone use

That gap between perception and behaviour gets wider once you know how badly people estimate their own phone use.

Sally Andrews, David Ellis, Heather Shaw and Lukasz Piwek installed a logging app on the phones of 29 participants — 23 made it into the analysis — and recorded two weeks of real use. Participants unlocked their phones a mean of 84.68 times a day (SD 55.23). Asked to estimate the same number, they said 37.20. Not only was the estimate less than half the truth, it did not correlate with the truth at all: r = .11, p = .610. Duration estimates were far better; the number of uses was essentially uninformed guessing. Fifty-five per cent of all those uses lasted under thirty seconds.

Two caveats, because the study is small and old. Twenty-three Android-owning students at one British university in 2015 is not a population estimate, and phones have changed since. Treat 84.68 as a demonstration that the gap exists, not as the number of times you personally check your phone.

An earlier version of this site's article on early signals opened with the line that the average person checks their phone around ninety times a day, sourced to nothing. That figure is gone. The finding worth keeping from this literature is not a count. It is that the person best placed to know how much they are on their phone is measurably bad at knowing it — which is why 'I barely touched it' and 'you were on it all night' can both be sincere.

Why it lands: it reads as exclusion, not as rudeness

The mechanism has been tested, and it is not etiquette.

Varoth Chotpitayasunondh and Karen Douglas showed participants a three-minute silent 3D animation of two people talking and asked them to imagine being the one with their back to the camera. In one condition the other character never touched their phone. In the second they used it part of the time. In the third, most of the time. Of 153 people recruited, 128 passed the attention checks.

As phubbing intensity rose, relationship satisfaction fell steeply (r = -.72) and communication quality was rated worse. The four fundamental needs measured in the ostracism literature — belonging, self-esteem, meaningful existence and control — were all threatened, with correlations running from -.39 to -.74. But only belonging mediated the effect on communication quality and satisfaction. The other three did not. Being phubbed did not primarily register as disrespect. It registered as being left out.

The limitations are real and the authors name them: cartoon figures rather than people, an imagined conversation rather than a lived one, and 114 of the 128 participants were women. It is a clean demonstration of a mechanism, not a measurement of what happens at dinner.

One detail of the design is worth more than the headline. In the control condition — the no-phubbing baseline — the character sits down, puts the smartphone on the table, and never picks it up for three minutes. The experiment that established how much being phubbed hurts used a visible phone on the table as its picture of nothing happening. Which is a second, accidental argument for the first section of this article.

What it predicts over time is smaller and disputed

Same-day effects are one thing. Whether phone use predicts where a relationship will be in six months is a different question, and the answer is genuinely unsettled.

Carnelley's couples showed no association between diary-averaged perceived phubbing and relationship quality two months later. The authors set that directly against two studies that did find longer-run effects: one on US married couples across a year, one on Chinese married couples across three months. They offer several reasons the results might diverge — their own measure averaged across fourteen days rather than sampling a single one, and it covered phones only rather than phones, tablets and television — and they do not claim to have settled it.

The other durable ambiguity is direction. Much of the foundational work is correlational. Brandon McDaniel and Sarah Coyne's technoference study, which named the phenomenon, surveyed 143 married or cohabiting women at a single point in time and found that more reported interference went with more conflict over technology, lower relationship satisfaction, lower life satisfaction and more depressive symptoms. James Roberts and Meredith David's founding paper built its partner-phubbing scale on 308 adults and tested the model on 145 more, finding that the route from phubbing to dissatisfaction ran through conflict about phone use. Both are informative. Neither can tell you which came first.

Even the most-cited field observation has this problem. Shalini Misra and colleagues watched 100 pairs of people talking in coffee shops and coded what their devices did; the meta-analysts who later included it point out that with no experimental control, whether the phone soured the conversation or the souring conversation invited the phone cannot be recovered from the data.

A first date is a worse sample than any of these. One glance at a screen, on one evening, is a single observation of a behaviour that varies enormously by day, by mood and by what is actually on the phone. If you are trying to read someone, the accuracy numbers for reading romantic interest at all are sobering before you even start, and what body language actually predicts covers how few of the classic tells survive measurement.

The move with the most support behind it is naming it

If the appraisal is what carries the effect, then whatever changes the appraisal is the lever. The research points at this consistently, though it has to be said as an inference rather than a trial result: nobody has randomised people into explaining their phone use and measured what happened.

What there is: Frackowiak's diary found the daily effect running through perceived partner responsiveness and moral judgement of the behaviour. Follow-up work from the same group found that people who saw their partner as low on validation were the most upset on high-phubbing days, while those who felt understood were relatively protected. And Lynne Kelly, Aimee Miller-Ott and Robert Duran, asking 225 adults about their partner's phone behaviour, found it was rated as generally expected, neutral and typical — but rated more negatively when the person had been expecting attention at that moment, and more negatively in public than at home. Our piece on texting habits goes through the rest of that work.

Read together, those results describe an expectancy problem rather than a behaviour problem. The same thirty seconds on a phone reads as neutral in one context and as a snub in another, and the difference is what the other person had been led to expect. Saying 'my sister is in hospital, I need to keep half an eye on this' at the start of the evening does not change the phone use at all. It changes the only variable the research has shown to matter.

The same logic runs the other way. If you are the one being phubbed, the appraisal is happening in your head, on a first date, with essentially no information about this person's baseline. Asking is cheap. Concluding is not, and the difference between love and anxious attachment covers why some of us find it much harder to leave a question open.

Using the phone together is a different behaviour

One exception turns up whenever anyone looks for it, and it is worth stating because the etiquette advice almost never distinguishes it: pulling out a phone to show the other person something is not what any of this research measured.

Kelly, Miller-Ott and Duran's participants rated sharing something on the phone with a partner positively — it was among the few phone behaviours that came out on the right side of the ledger. Michal Frackowiak and colleagues have since argued explicitly that co-present phone use can be good, bad or neutral, and that treating every instance of it as partner phubbing has been the field's blind spot.

The mechanism explains why. If what makes phubbing hurt is the threat to belonging, then a phone both of you are looking at is not doing the thing that causes the harm. Nobody has been excluded from anything.

This is the weakest-supported section of this article and it should be read that way: a plausible mechanism, one survey finding and a conceptual argument, not a tested result.

When the conversation about phones stops being about phones

There is a version of this topic that is not an etiquette question, and it needs saying plainly rather than at the bottom of a listicle.

Joyce Schokkenbroek, Wim Hardyns and Koen Ponnet surveyed 346 people in relationships and traced a chain: being phubbed lowered perceived partner responsiveness, which raised anxiety, which predicted electronic partner surveillance — checking a partner's phone, messages or online activity. The authors are careful about what they found, and anyone quoting them should be too: the model explained roughly five per cent of the variance in surveillance behaviour, and they say directly that the findings apply to a small group of people. It is a documented pathway, not an explanation of why any particular person goes through a phone.

But the pathway is worth knowing, because it describes how an ordinary feeling turns into a behaviour that is not ordinary at all. Feeling ignored is ordinary. What happens next is where the line sits.

Being required to reply within a set time, or to account for a delay. Being asked to hand over a phone, or having one checked. Location sharing that is not mutual and not freely chosen. Consequences for a slow reply — escalation, punishment, silence used as a sanction. Being told who you may and may not message.

Those are not phone habits and they are not communication problems. Monitoring, restricting contact and demanding constant availability are recognised markers of coercive control. If any of that is familiar, from either side of it, the useful next step is a person and not an article: Find A Helpline, linked in the sources below, lists free and confidential services by country, and in the United States the National Domestic Violence Hotline is 1-800-799-7233, or text START to 88788, at any hour.

Where what sits underneath is jealousy rather than control — and the phubbing meta-analysis does find a jealousy association, rz = .289 — the psychology of jealousy works through that distinction at length. More in this vein under Mindset.

What all of this is worth on an actual date

Pulling the evidence together into something usable, with the confidence levels attached.

The table is not the variable. Well supported, in the sense that a decade of attempts has failed to show that it is. Put the phone wherever you like. The advice to bury it in a bag was never grounded in anything, and the little ritual of arranging it face-down is managing a problem nobody has demonstrated exists.

Picking it up mid-conversation is the variable. Well supported, across 37 studies of interactions and 52 of couples, with an effect somewhere between small and moderate and a causal direction that is still open.

What lands is exclusion, not rudeness. Reasonably supported as a mechanism, from an animation study with obvious limits. This is why a check that visibly has nothing to do with the other person still costs something, and why explaining it helps.

Say what you are doing. Inferred from mediation results rather than tested directly, but it is the one lever that touches the variable the research keeps naming.

Do not build a verdict out of one evening. This is the part the genre gets most wrong. A correlation of -.22 in a sample of nineteen thousand people tells you almost nothing about the person in front of you, your own count of your own phone use is uncorrelated with reality, and one date is a sample size of one. If the pattern is real it will still be there in a month, and it will be far easier to see then. The same logic governs what happens after the date, where the timing of the first message has now been measured directly and the popular rule lost: see the 72-hour rule. More in this vein under Dating, and if you want to know how much of your own reading is coming from your attachment system rather than from them, our attachment style test takes about two minutes and is honest about what it cannot tell you.