Ask a room how good they are at doing two things at once and most hands go up. In one study that actually put the question to people, 193 out of 277 rated themselves above average.
Then the researchers measured it. The correlation between how good people thought they were and how good they turned out to be was 0.08, which is another way of writing no relationship at all.
Who multitasks
That study, published in 2013 by David Sanbonmatsu, David Strayer, Nathan Medeiros-Ward and Jason Watson at the University of Utah, set out to ask not whether multitasking is bad for you but who does it.
The Artful Age
A weekly letter on aging well, family across generations, and the creative life after the kids leave home.
They gave 277 students a questionnaire measuring how much of their media time involved more than one medium at once, asked how often they used a phone while driving, measured impulsivity and sensation seeking, and then tested actual ability with the Operation Span task, which makes you hold a sequence of letters in mind while verifying arithmetic problems. Two real tasks, running at once, scored separately.
The self-ratings were inflated. The average person placed themselves at the 63rd percentile, against an anchor where 50 meant exactly average. They were also, as above, unrelated to performance.
And the real correlation ran the wrong way. Multitasking activity was negatively related to multitasking ability, r = -0.19. Phone use while driving, negatively too, r = -0.15. Comparing the top and bottom quartiles of multitasking activity, span scores came in at 40.3 against 48.7.
What predicted multitasking was not capability but temperament. Three things came through in the regression: believing you are good at it, attentional impulsiveness, which is the difficulty of holding attention on one thing, and the disinhibition strand of sensation seeking.
The authors put it flatly. “The persons who chronically multi-task are not those who are the most capable of multi-tasking effectively.”
But doesn’t the multitasking cause that?
This is where the familiar story usually arrives. The multitaskers score worse because the multitasking wore them down.
That claim comes from one paper. In 2009, Eyal Ophir, Clifford Nass and Anthony Wagner published a study in the Proceedings of the National Academy of Sciences comparing heavy and light media multitaskers across a battery of laboratory tasks. The heavy group did worse whenever the task required filtering out something irrelevant, and the authors concluded that heavy media multitaskers “are more susceptible to interference from irrelevant environmental stimuli and from irrelevant representations in memory.”
The effects were large. Change detection, d = 0.68. A continuous performance task with distractors, d = 1.19. Task-switch cost, d = 0.96.
It is the most cited result in this area and the engine under most monotasking advice written since. One disclosure: the 2009 paper itself is not freely available, and every figure and quotation from it here is taken from the replication paper discussed next, which tabulates Ophir’s effects and quotes that conclusion directly.
What happened when people checked
In 2017, Wim Wiradhany and Mark Nieuwenstein at the University of Groningen ran two replication studies and then gathered every published test of the same question.
The replications comprised 14 tests, with an average power of 0.81, so they were well placed to find the effect if it was there. Five came out significant in the predicted direction. Two of those five held up under a stricter Bayesian analysis.
The meta-analysis pooled 39 effect sizes from 12 published articles plus their own two experiments. It produced a small but significant association, d = 0.17.
Then they checked whether the literature was lopsided, and it was. Small studies reporting large positive effects were abundant; small studies reporting nothing were conspicuously absent. Egger’s test put the asymmetry at p = 0.005, and the shape of the gap pointed at reporting bias rather than real variation between studies, since their moderator analyses had already ruled out differences by task, population or method of analysis.
Correcting for it moved the answer. Trim-and-fill gave d = 0.07, no longer significant at p = 0.81. A regression-based correction gave d = 0.001.
Across all 39 effects: 25 pointed toward heavy multitaskers being more distractible, 11 pointed the other way, and 3 found nothing.
Their conclusion is measured. The findings “lead us to question the existence of an association between media multitasking and distractibility in laboratory tasks of information processing.”
The parts that argue against that conclusion
Two things in the Groningen paper cut against its own headline, and the authors raise both themselves.
Their failed replications still tilted in Ophir’s direction even when not significant. They explain why that is weaker evidence than it looks: the seven measures within each study are correlated across the same participants, so one chance difference between groups shows up seven times over.
And they offer the alternative reading without hedging it away. The association may be real and simply very small. On the uncorrected estimate, detecting it reliably would take 428 people per group. Almost every study in this literature has used a few dozen.
One more complication
The two papers here come from different universities in different countries and share no authors. They do share an instrument: both use Ophir’s Media Multitasking Index.
And Wiradhany and Nieuwenstein criticize how that index is built. It captures the proportion of your media time that is multitasked, not how much of it there is. Someone who spends one hour a day on a laptop with the television on can score the same as someone who does it for sixteen. That objection lands on the Utah correlations as squarely as on anyone else’s.
What is actually left
The Utah study is correlational and cross-sectional. It cannot say whether weaker executive control pushes people toward multitasking, or whether something upstream produces both. Its participants are undergraduates at one American university, and the multitasking and driving measures are self-reported. The correlations are small, around 0.15 to 0.20, even where the quartile contrasts reach medium effect sizes.
So neither half of this supports the version of the case that usually gets made. Nobody has shown that your attention is broken, and nothing here can tell any individual reader anything about their own executive control. These are population averages with a lot of scatter, and a person in the heaviest quartile can be perfectly capable.
What is left is narrower and a little strange. The pull toward a second screen tracks temperament more clearly than it tracks talent. The people doing it most are, on average and by a small margin, slightly worse at it. And the one instrument that cannot help you here is your own sense that you are the exception, because that sense has been measured against performance and found unrelated to it. That last part is not an accusation. It applied to everyone in the study, including the people who were good.
Which makes the argument for doing one thing at a time less dramatic and more usable. Not a repair for damage that may not exist. Just an acknowledgment that the urge to split your attention is not itself evidence you can afford to, and that you are not well placed to referee the question.