Are You 42% More Likely to Achieve a Written Goal?

Every January the same figure comes round. It sits on a slide at the company kickoff, in the opening line of a productivity newsletter, on a laminated card inside a coaching workbook: write your goals down and you are 42 per cent more likely to achieve them. What gives the sentence its authority is that this time there is a university behind it, and a professor, and a control group.

All of that is true. The study exists, you can download it in a minute, and it was run by a working psychologist who designed it because the older goal stories about Harvard and Yale graduating classes turned out to have no paperwork behind them. On the shelf of self-improvement statistics, having an experiment at all puts this one near the top.

What the study does not contain is the number 42.

Search the document for the digits

The source everyone points at is a three-page document called Goals Research Summary, by the American psychologist Gail Matthews, then a professor in the Department of Psychology at Dominican University of California, hosted on the university’s own servers (Matthews, Goals Research Summary). It opens by dismantling the Yale and Harvard legends, describes who took part, sets out five conditions, prints two bar charts and ends with three conclusions.

Search it for a percent sign and you get two hits, both inside the legend being demolished: the 3 per cent of a graduating class who supposedly had written goals, and the ten times more they supposedly earned. No percentage is attached to Matthews’s own finding. There is no 42 in the file.

What the 149 finishers were asked to do

Two hundred and sixty-seven people signed up, drawn from businesses, organisations and business networking groups. One hundred and forty-nine finished. They were aged 23 to 72, thirty-seven men and one hundred and twelve women, from six countries, working as entrepreneurs, teachers, healthcare staff, artists, lawyers, bankers and managers. Their goals were the ordinary furniture of working life: selling a house, writing a chapter, securing a contract and, memorably, preventing a hostile takeover.

Each person was randomly assigned to one of five conditions and given four weeks. Random assignment is the part the older legends never had.

Group	What they were asked to do	Mean achievement score
1	Think the goal through and rate it, unwritten	4.28
2	Type the goal in and rate it	6.08
3	Write the goal plus action commitments	5.08
4	Write both, send them to a friend	6.41
5	All of that plus weekly progress reports	7.60

At the end of the four weeks, in the summary’s own words, “participants were asked to rate their progress and the degree to which they had accomplished their goals”. That sentence is doing an enormous amount of work. The outcome variable is a number a person typed about themselves. No colleague confirmed it, no contract was checked, no house sale was looked up.

Note also who is missing. Of 267 starters, 118 never came back, a dropout of 44 per cent over four weeks, and people whose month went badly are not equally likely to fill in the closing survey.

Where 42 comes from, arithmetically

Take the second row of the table and divide it by the first. 6.08 ÷ 4.28 = 1.4206. The written-goal group’s average self-rating was about 42 per cent higher than the unwritten-goal group’s. That is the whole derivation, and no other pair of numbers in the study yields 42: the four writing groups pooled score 6.44, which over 4.28 gives 50 per cent, and the top group’s 7.60 gives 78.

So the figure is arithmetically honest and semantically wrong. “Forty-two per cent higher average rating” and “forty-two per cent more likely to achieve” describe different quantities. Likelihood is a probability of an outcome; a mean is the centre of a pile of scores. Two hotel guests who rate a stay 4.28 and 6.08 differ by 42 per cent in their ratings, and nobody concludes that the second was 42 per cent more likely to have had a holiday.

There is a second wobble in the table that the popular version never mentions. Group 3, which wrote goals and action commitments, scored 5.08: a full point below Group 2, which wrote goals alone. If writing is the active ingredient, writing more of it should not go backwards. Matthews’s own list of significant differences says nothing about Group 3 beating anyone.

The scale nobody names

Here is the awkward part for anyone checking the division. The summary never says what the achievement rating ran from or to. The vertical axis of its bar chart stops at 8 and the highest group mean is 7.60, so the scale reached at least that far. Beyond that, we are guessing.

That gap in the record is worse than pedantic, because a ratio of two scores has no fixed meaning until you know the span they live on. On a 1 to 8 scale the 1.8-point gap covers about a quarter of the usable range, a respectable difference. On a 0 to 10 scale it looks smaller. The 42 per cent depends on where the zero sits, and the zero is not disclosed.

Nor are standard deviations, confidence intervals, exact p values, or how many people ended up in each condition. With 149 finishers across five groups, each cell holds roughly thirty people.

Who did the division first

The earliest dated page we could open carrying the sentence belongs to the American author and former publishing executive Michael Hyatt, in a post titled 5 Reasons Why You Should Commit Your Goals to Writing. Web archive captures of that page from June 2011, June 2012, September 2013 and November 2013 contain neither the number nor Matthews’s name. The capture from 6 January 2014 contains both: “She found that you are 42 percent more likely to achieve your goals just by writing them down” (Hyatt, captured 2014). So the phrasing entered wide circulation between November 2013 and January 2014, more than a year before Dominican University issued its own press release about the study.

Two different versions of Group 1

Dominican’s press office wrote the study up in January 2015 (Gardner & Albee, 2015). That release converts the ratings into pass rates twice, inconsistently.

Early in the release: “more than 70 percent of the participants who sent weekly updates to a friend reported successful goal achievement (completely accomplished their goal or were more than half way there), compared to 35 percent of those who kept their goals to themselves, without writing them down”. Further down the same page: “only 43 percent of Group 1 either accomplished their goals or were at least half way there”, against 62 per cent for Group 4 and 76 per cent for Group 5.

The unwritten-goal group is therefore 35 per cent in one paragraph and 43 per cent in another. No figure at all is given for Group 2, the written-goal condition the 42 is supposed to describe. Had one been given, and had it sat 42 per cent above Group 1’s 43, it would have to land near 61.

Looking for it where studies are indexed

Dominican’s own repository is candid about what this is. The record lists Document Type as «Podium Presentation» and Sponsorship as the 87th Convention of the Western Psychological Association, held in Vancouver (Matthews, 2007). It is filed under faculty conference presentations, not articles. The 2015 press release adds that the findings were due to be presented again that May at the Ninth Annual International Conference of the Psychology Research Unit of the Athens Institute for Education and Research. Two conference talks, eight years apart, one dataset.

Here is what four bibliographic indexes returned on 5 September 2026. The queries are printed so you can rerun them.

• Crossref, author Gail Matthews with written goals goal achievement in the bibliographic field: exactly one record by this author on this topic in the ranked results, DOI 10.1037/e656862007-001, container PsycEXTRA Dataset, and Crossref’s own type field for it reads dataset.
• OpenAlex, filter=title.search:”written goals”: 7 works. Two are Matthews’s, the repository record above and the same PsycEXTRA entry; the rest are three nursing papers, an undergraduate thesis and a book chapter.
• OpenAlex, filter=title.search:”goal achievement”,raw_author_name.search:Matthews: 1 work, whose source is listed as a repository.
• PubMed, “written goals”[Title]: 0 hits. PubMed, “written goals”[Title/Abstract] AND “goal achievement”[Title/Abstract]: 1 hit, a 2019 paper in Psychological Science about planfulness and exercise, unrelated.
• Europe PMC, TITLE:”written goals”: 0 hits. Europe PMC, the full title The Impact of Commitment, Accountability, and Written Goals on Goal Achievement: 0 hits.

The one live identifier resolves to the American Psychological Association’s PsycEXTRA collection, where the association files material that never went through a journal (Matthews, 2007, PsycEXTRA). Nineteen years after the talk, no peer reviewer has looked at the analysis and no reader can see the numbers behind the bars.

Being unpublished is not the same as being wrong; plenty of sound work stays in a drawer. It does mean the checks a journal performs, on the statistics, on the missing scale, on the 44 per cent attrition, were never performed here.

What 95 pages say, and the division none of them does

We pooled thirty-two literal searches, from 42% more likely written goals and Gail Matthews goal study to 42 percent goals study debunked and Gail Matthews study never published peer reviewed. They returned 142 unique addresses; 47 are excluded from the count because they refused the request, returned a missing page, served unrelated content or delivered no article body.

Of the 95 that remained, 49 state the claim as an established finding, from “people who write down their goals are 42% more likely to achieve them than those who don’t” to a garbled “33-42%”, a “39.5 percent” and a “50.4% improvement”. Two qualify it: one notes that the study “relied on self-reports”, the other that “I don’t believe the results of this study were published”. One calls it “a real study, but it’s routinely overstated”. The remaining 43 stand next to the claim without weighing it: the press release’s 70 per cent against 35, the Harvard legend, the raw group means.

Six of the 95 say the study was a conference talk or was never published, and ten name the 149 people who finished or the self-rated outcome. Two upgrade it to a journal that never carried it, one crediting Psychological Science and one crediting the Journal of Applied Psychology. Three pages print the group means, 4.28 and 6.08; none of the 95 divides one by the other.

Goal research with error bars

None of this leaves you empty-handed. Goal setting has one of the longest measurement records in work psychology, built from hundreds of studies rather than one.

The founding review came from the American organisational psychologists Edwin Locke and Gary Latham with Karyll Shaw and Lise Saari, covering laboratory and field work from 1969 to 1980: specific, challenging goals led more often to higher performance than easy goals, do-your-best goals or no goals, with 90 per cent of studies positive or partially positive (Locke, Shaw, Saari & Latham, 1981). Two decades later Locke and Latham set out the machinery: goals direct attention, mobilise effort, stretch persistence and push people to invent strategies (Locke & Latham, 2002). Neither paper singles out the physical act of writing.

For a modern number with error bars, British health psychologists Tracy Epton, Sinead Currie and Christopher Armitage pulled 384 effect sizes from randomised controlled trials, covering 16,523 people, and asked what setting a goal adds on its own. Their pooled answer: d = .34, confidence interval .28 to .41, with bigger effects when the goal was difficult, set publicly, or set as a group (Epton, Currie & Armitage, 2017).

And for the part that most resembles reaching for a notebook, a Sheffield-led team examined 138 experiments with 19,951 participants on monitoring progress towards a goal. Monitoring promoted goal attainment at d+ = 0.40, confidence interval 0.32 to 0.48, and the effect grew when the outcome was reported or made public and when the information was physically recorded (Harkin et al., 2016).

Translate those into something you can picture. An effect of d = 0.34 means the average person given a goal finishes ahead of about 63 out of 100 people who were not, where chance alone would put them ahead of 50. An effect of d+ = 0.40 moves that to about 66 in 100. Line up thirty colleagues and you could not pick the difference out by eye, but across a company or a clinical trial it is a repeatable shift.

Notice which moderator earned its place: physically recorded. Writing does show up in the well-measured literature, as a way of tracking progress worth a modest fraction of a standard deviation, rather than as a 42 per cent multiplier on your chances.

Round numbers with a research address come apart the same way elsewhere: a claim that six minutes of reading cuts stress by sixty-eight per cent, which traces to a report named after a chocolate bar, and the ninety-six per cent behind the military sleep method, which lost six weeks of daily practice on its way to the internet. The twenty-five minute working interval has a shorter trail and exactly one experiment: what happened when somebody tested the Pomodoro split. Everything we have checked is listed in the registry of checked claims. The older and grander goal story, the Harvard and Yale classes, is traced in was there ever a Harvard or Yale study on written goals.

Sources

• Matthews, G (2007). Goals Research Summary. Dominican University of California. Three-page summary of the study hosted by the university; 267 recruited and 149 completed, aged 23 to 72, 37 men and 112 women, five randomly assigned conditions over four weeks, outcome a self-rating of progress and accomplishment; group means 4.28, 6.08, 5.08, 6.41 and 7.60, and 6.44 for groups 2-5 pooled; contains no percentage attached to written goals dominican.edu
• Hyatt, M (2014). 5 Reasons Why You Should Commit Your Goals to Writing. michaelhyatt.com, Internet Archive capture of 6 January 2014. Earliest dated page we could open that carries the sentence: she found that you are 42 percent more likely to achieve your goals just by writing them down; captures of the same page from June 2011, June 2012, September 2013 and 1 November 2013 contain neither the figure nor Matthews's name web.archive.org
• Gardner, S., & Albee, D (2015). Study focuses on strategies for achieving goals, resolutions. Dominican Scholar, Press Releases. Dominican University press release of January 2015; gives 70 percent versus 35 percent in one paragraph and 43, 62 and 76 percent for groups 1, 4 and 5 further down, with no figure for group 2; the PDF at cgi/viewcontent.cgi returns 403 to a command-line request, the landing page returns 200 scholar.dominican.edu
• Matthews, G (2007). The Impact of Commitment, Accountability, and Written Goals on Goal Achievement. Dominican Scholar, Psychology Faculty Conference Presentations. Repository record; Document Type listed as Podium Presentation, Sponsorship as the 87th Convention of the Western Psychological Association, Vancouver, British Columbia, publication date 2007 scholar.dominican.edu
• Matthews, G (2007). The Impact of Accountability and Written Goals on Goal Progress. PsycEXTRA Dataset. American Psychological Association grey-literature record, the one live identifier for the study in Crossref and OpenAlex; Crossref types it as dataset and gives no volume, issue or pages doi:10.1037/e656862007-001
• Locke, E. A., Shaw, K. N., Saari, L. M., & Latham, G. P (1981). Goal setting and task performance: 1969-1980. Psychological Bulletin. 90(1), 125-152; review of laboratory and field studies finding that specific, challenging goals led more often to higher performance than easy goals, do-your-best goals or no goals, with 90% of studies showing positive or partially positive results doi:10.1037/0033-2909.90.1.125
• Locke, E. A., & Latham, G. P (2002). Building a practically useful theory of goal setting and task motivation: A 35-year odyssey. American Psychologist. 57(9), 705-717; summary of the goal-setting programme and its mechanisms: directing attention, mobilising effort, increasing persistence and prompting strategy development doi:10.1037/0003-066X.57.9.705
• Epton, T., Currie, S., & Armitage, C. J (2017). Unique effects of setting goals on behavior change: Systematic review and meta-analysis. Journal of Consulting and Clinical Psychology. 85(12), 1182-1198; 141 papers, 384 effect sizes, N = 16,523 from randomised controlled trials, random effects d = .34, CI [.28, .41], larger when the goal was difficult, set publicly or a group goal doi:10.1037/ccp0000260
• Harkin, B., Webb, T. L., Chang, B. P. I., Prestwich, A., Conner, M., Kellar, I., Benn, Y., & Sheeran, P (2016). Does monitoring goal progress promote goal attainment? A meta-analysis of the experimental evidence. Psychological Bulletin. 142(2), 198-229; 138 studies, N = 19,951; progress monitoring promoted goal attainment at d+ = 0.40, CI [0.32, 0.48], with larger effects when outcomes were reported or made public and when the information was physically recorded doi:10.1037/bul0000025