Every claim in our books carries the number of people it was tested on. Every page tells you which part of it is oversold. Including ours.
We sell books and workbooks about how a mind works, for every age from a baby's first week to the years after work ends. That's a field full of confident sentences with nothing behind them. The only honest way to stand apart from it is to show our work and let you check it.
So this page gives you five questions for checking any claim about a mind. Then it asks you to use all five on us. If you find something here that's wrong, tell us. We date this page and we revise it.
Last reviewed 23 August 2026 · we keep revising this page
Use them on anyone who tells you something confident about your own mind. Then use them on this page.
Question 1
A finding from 42 families and a finding from 617,466 people are not the same kind of thing. They get quoted the same way anyway.
Question 2
Almost anything beats nothing. The question worth asking is whether it beat something else that also takes effort.
Question 3
A striking first result is a hunch. It counts once somebody else runs it again and gets the same thing.
Question 4
People who stand to make money from a result report good news much more often. That includes us. It's why we print the numbers instead of describing them.
Question 5
Almost everything in psychology is a small shift in an average, measured on people who aren't you. That's still useful. It's still not a promise.
A ruler for effect sizes. In a study, "significant" means "probably not a fluke". It does not mean "big". Picture a room of 30 people. A large effect (around d=0.65) moves most of the room a step you would notice. A small one (around d=0.19) moves the room less than you could spot standing in it. Real, worth having, and nothing like a transformation.
A large effect · around d=0.65
Most of the room moves a step you'd notice standing in it.
A small effect · around d=0.19
Still real. Still worth having. You wouldn't spot it in the room.
These show up in our books and workbooks with the correction printed right next to them, not left out. Our standing rule: where the correction is itself disputed, we write contested, never "debunked". Overstating a correction costs us more than the original myth ever did.
The original finding came from n = 42 families (Hart & Risley, 1995). A later study that sampled differently did not find a gap that size (Sperry, Sperry & Miller, 2019, Child Development). That study is itself rebutted by Golinkoff, Hoff, Rowe, Tamis-LeMonda & Hirsh-Pasek (2019), who argue it did not sample the same extremes. The direction survives. The number does not.
The brain-imaging work people cite for it was done on adults, not on the small child the phrase is normally aimed at. We keep the practice, because putting a word on a feeling helps. We drop the brain-science story around it.
Later, bigger analyses found that much of the original link was explained by family background rather than by a child's self-control. It's a useful demonstration. It isn't a measure of character.
It's a tidy number stuck on a slow process that has no biological deadline and runs differently in different people. Useful as a reminder that adolescence is longer than the law says. Not a fact about a birthday.
It didn't come from a habit study at all. The closest real figure is a median of 66 days with a range of 18 to 254, measured on roughly n ≈ 40 people (Lally et al.). We print the range every time we print the number. The finding worth having from that work is the other one: missing a single day did not do meaningful damage to habit formation.
A re-analysis found that deliberate practice explained roughly 26% of the difference between people in games, 21% in music, 18% in sport, 4% in education and under 1% in professions (Macnamara and colleagues, disputed by Ericsson). What survives is narrower and more useful: practice aimed at one named weakness, with something that tells you the same day whether you got it right.
Troy, Shallcross & Mauss (2013, Psychological Science) found that reframing a problem helped with the ones people could not control, and went with worse outcomes on the ones they could. Correlational, samples in the low hundreds, so it shows a pattern and not a cause. It's why our adult material asks whether a thing is fixable before it offers you a way to make peace with it.
Later work (Killingsworth, Kahneman & Mellers, 2023, PNAS) found the ceiling shows up mainly in roughly the least happy 15 to 20% of people, not in everyone.
Andrés Zuzunaga drew it as a "Venn diagram of purpose" in 2011. British blogger Marc Winn relabeled it ikigai in 2014, which Winn has publicly confirmed. It is not Japanese, and it is not from the longevity research it gets attached to. The real idea (Kamiya, 1966) is smaller and more everyday. In the Ohsaki study n = 43,391 adults were asked one question about it, and the answer predicted death rates (Sone et al., 2008, Psychosomatic Medicine).
We don't build on them. The fair version is that the best-known critique is still an unpublished preprint, and its author is disputed. Two separate lines settled it for our purposes. Okinawa fell from 1st to 26th of Japan's 47 prefectures by 2000, and to 43rd by 2020 (Poulain et al., 2024, Journal of Internal Medicine). And a researcher who co-declared Nicoya later published that people born after 1930 show no advantage (Rosero-Bixby, Demographic Research, 2023).
All three are widely reported as overturned, and we treat them that way in our books. Full disclosure: our sourcing for these three is second-hand. We have not gone back and checked them against the original studies ourselves. They're on our list to check, and we'd rather tell you that than imply a rigor we haven't applied.
A selection, sorted by the stage of life the book is written for. Every one of these appears in a product with its sample size printed beside it.
What predicts secure development isn't never missing a cue. It's coming back reliably after you do. Healthy pairs are out of step a lot of the time.
Tronick, Als, Adamson, Wise & Brazelton (1978); Mesman, van IJzendoorn & Bakermans-Kranenburg (2009), Developmental Review, 80+ studiesBedtime fading: for three nights, write down when they actually fall asleep. Set bedtime to that, then move it earlier a little at a time. No crying involved. Limit line, printed above the steps: infants 6 to 16 months whose parents already reported a sleep problem. Nothing in it applies to newborns.
Gradisar et al. (2016), Pediatrics, randomized trial, n = 43A predictable bedtime routine leads to better sleep.
Mindell et al. (2009), randomized, n = 405Having other teenagers in the room changes how much risk a teenager takes. It does not do that to adults. That's why our advice is about changing the situation rather than lecturing the person.
Gardner & Steinberg (2005), Developmental PsychologyAsking someone directly about self-harm or suicide does not plant the idea.
Dazzi, Gribble, Wessely & Fear (2014), Psychological Medicine, review of 13 studiesSpecific "when X, I do Y" plans move people from meaning to do a thing to actually doing it. That's the mechanism behind every scripted first line in our books.
Gollwitzer & Sheeran (2006) meta-analysis, 94 tests, 8,000+ participantsBetween 47% and 54% of people who genuinely mean to do something never do it. The failure is at getting started rather than at wanting it.
Sheeran (2002), European Review of Social PsychologyTalking to yourself in the third person cuts down on rumination, the going-over and over of the same thought. A real nudge and nothing more: the pooled effect is g = 0.19.
Murdoch et al. (2023), Stress and Health, 25 experiments, N = 2,397Going over and over a problem predicts depression more strongly than how bad the original problem was. That's why our material works on the loop rather than on what the worry is about.
Nolen-Hoeksema, Wisco & Lyubomirsky (2008), Perspectives on Psychological ScienceBurnout is something that happens at work. It is not a diagnosis. There's no validated cut-off, no tested line between having it and not, and it overlaps heavily with depression. We print it as a reason to go talk to someone, never as a verdict on you.
WHO ICD-11 (2019), from Maslach & Jackson (1981)Retirement is relief from bad work. The health benefit was largest for people leaving poor work environments, and absent in people leaving high-grade, low-demand, satisfying jobs. Contradicting evidence, printed alongside it: Behncke (2012) found harms, and Garrouste (2022) concludes there is no consensus.
Westerlund et al. (2009), The Lancet 374:490-501, n = 14,714Volunteers were less likely to die during the study (hazard ratio 0.78 for men, 0.77 for women, where 1.00 would mean no difference). Their non-volunteering spouses were not (1.01 / 1.00). Same house, same income, same street. That comparison is what answers the obvious objection: that healthy people simply volunteer more.
O'Reilly, Rosato, Moriarty & Leavey (2017), International Journal of Epidemiology 46(4), 617,466 individualsPurpose predicts lower death rates just as strongly in young, middle-aged and older adults. Neither age nor retirement status changed it. This is the single finding that justifies one method told at seven stages.
Hill & Turiano (2014), Psychological Science 25(7), MIDUS, 14-year follow-up, 569 deathsThe five questions above, turned on ourselves.
Finding the right line for your country…
Persistent withdrawal alongside loss of interest, or a change in sleep or appetite, warrants a doctor rather than a worksheet. Asking someone directly about self-harm does not plant the idea. It gives them somewhere to put it.
If we've misused a citation, got a sample size wrong, or missed a finding that's been overturned since we last reviewed this page, write to hello@masterfulmindsbooks.com. We'll fix it and date the fix. That's the whole point of publishing this.