Every January someone writes that only 8 percent of people keep their New Year's resolutions, and credits the University of Scranton. I checked that sentence the way I'd check a citation in a paper. Follow each link to the page it cites, open that page, find the number, and see whether the number is on it. Then do the same for whatever that page cites, until a page cites nothing. For the 8, every chain I followed ends at the same web page, and that page doesn't name a paper.
The 8 percent can't be traced to any study. It's the share of people in a 2005 telephone poll, run by a self-help author alongside his book, who said they always achieve their resolutions, and an aggregator site later filed it under "University of Scranton, Journal of Clinical Psychology." The Scranton papers are real and say something else. Norcross and Vangarelli (1988) followed 200 resolvers for two years and found 77 percent still keeping their pledge at one week and 19 percent at two years. Norcross, Mrykalo and Blagys (2002) found 46 percent of 159 resolvers continuously successful at six months, against 4 percent of 123 comparable people who wanted to change the same kinds of problem but made no resolution.
Below is the chain, dated and quoted, plus the other two January numbers (80 percent quit by February, and Strava's "Quitter's Day") and two independent studies. I didn't re-run anything. Every number is copied from the page that published it, and every page is archived with its capture date.
Every chain ends at Statistic Brain
The earliest news story I found is Forbes, January 1, 2013, by Dan Diamond: "University of Scranton research suggests that just 8% of people achieve their New Year's goals." The words "8% of people" are a hyperlink. It goes to statisticbrain.com/new-years-resolution-statistics, not to a journal. Within weeks the claim was on the Association for Psychological Science's "Members in the Media" page (January 8, 2013, "From: Forbes"), and BU Today had it on February 5, 2013, already mutated: "only 8 percent keep them for longer than a few weeks."
Balance Bar's press release on PR Newswire, December 19, 2013, gave it a footnote. "According to a recent study in the Journal of Clinical Psychology, over 40% of Americans make New Year's resolutions each year, but only 8% actually achieve those goals." The footnote reads "University of Scranton, Journal of Clinical Psychology; December 2012." It names an institution, a journal and a month, with no author, title, volume or page.
The footnote's wording is Statistic Brain's source line, so I pulled the page's history from the Wayback Machine. The label is already on the first capture, January 19, 2012: "Source: University of Scranton. Journal of Clinical Psychology," with "Date Verified: 12.26.2011." The "Published: 12.13.2012" date appears in the January 2, 2013 capture. It's a yearly restamp. The 2014 capture reads "Published: 12.13.2013," 2015 reads "Research Date: 1.1.2014," and 2016 reads "Research Date: December 27th, 2015." The table under the date is identical in all six captures from 2012 to 2016.
Norcross's CV lists three resolution studies he led, in Addictive Behaviors (1989), the Journal of Substance Abuse (1989 on the CV, 1988 on PubMed) and the Journal of Clinical Psychology (2002). Only the 2002 paper is in the journal the label names, and its headline numbers are 46 and 4. None of the three reports 8 percent.
The 8 is a poll answer
"Percent of Americans who usually make New Year's Resolutions 45%," then 17 percent infrequently and 38 percent never. "Percent of people who are successful in achieving their resolution 8%," then 49 percent with infrequent success and 24 percent who never succeed. Then two age rows, 39 percent of people in their twenties and 14 percent of people over 50.
Those rows are Stephen Shapiro's, and none of them is from a Scranton paper. Shapiro posted "How to Make Resolutions That Work" on December 30, 2005. It says "But 92% of resolutions fail" and links to his year-end write-up of "a survey of 1012 Americans." That write-up calls it a "random telephone survey ... conducted by Shapiro, author of 'Goal-Free Living,' with the assistance of Opinion Research Corp. of Princeton N.J.," and gives a margin of error of 3 percent. His summary was "only 8% of Americans say they always achieve their New Year's resolutions." A December 2008 post has the whole breakdown. Of the people he asked, 8 percent always succeed and 24 percent never do. Another 19 percent "achieve their resolutions every other year," and 49 percent have infrequent success.
Shapiro's four answers add to 100. Statistic Brain's three add to 81, because the 19 percent every-other-year row is missing. The age rows changed too. Shapiro wrote "39% of those in their twenties achieve their resolutions every year or every other year" and "Less than 15% of those over 50." Statistic Brain wrote "achieve their resolution each year," dropped the every-other-year part, and turned "Less than 15%" into 14%.
The page also carried "People who explicitly make resolutions are 10 times more likely to attain their goals," which is the 2002 paper's 46 against 4, rounded. And it carried a "Length of Resolutions" block (75 percent through the first week, 71 past two weeks, 64 past one month, 46 past six months), which is the 2002 paper's survival series.
In 2017 Statistic Brain replaced the table with its own survey ("1,273 online respondents, 216 phone, and 73 in-person," conducted January 1, 2017). Its rows are 41 percent who usually make resolutions and 9.2 percent who "felt they were successful." That's where the 9 percent and 41 percent variants come from. Psychology Today, January 9, 2017, quoted "9.2 percent of 1273 respondents." Inc., January 8, 2022, wrote "only 9% were successful" and named no study.
What the Scranton papers measured
Norcross and Vangarelli, "The resolution solution" (Journal of Substance Abuse 1(2), 127 to 134), tracked 200 resolvers prospectively for two years. The abstract gives two survival numbers: "Seventy-seven percent maintained their pledges for 1 week but only 19% for 2 years." The intermediate points, as Oscarsson and colleagues summarize the paper in 2020, are 55 percent at one month, 43 percent at three months and 40 percent at six months. The abstract also reports that 53 percent of the people who succeeded slipped at least once, with a mean of 14 slips over the two years.
Norcross, Mrykalo and Blagys, "Auld lang syne" (Journal of Clinical Psychology 58(4), 397 to 405), is the one that has a comparison group. 159 people made a resolution. 123 "comparable nonresolvers interested in changing a problem later" didn't. The groups didn't differ in demographics, problem history or goals, and both were followed for six months by telephone, and outcomes are self-reported. At six months 46 percent of resolvers were continuously successful and 4 percent of nonresolvers were. 46 divided by 4 is 11.5.
Norcross described both studies on NPR's Science Friday on December 28, 2012. "In two of our longitudinal studies, 40 to 46 percent of New Year's resolvers will be successful at six months." And of the comparison group, "0 to 4 percent." When Ira Flatow said the papers report a success rate "10 times higher" for resolvers, Norcross answered "That was our last study, that's exactly right." Forbes ran the 8 under his university's name four days later. I found no statement by Norcross about the 8 in anything I fetched.
A Psychology Today post on December 22, 2025 says "by the end of the first week into the new year, 77% of resolutions will have already failed (Norcross, 1988)." The paper says 77 percent maintained their pledge at one week.
The two numbers that are real
The closest real relative of the 8 is Richard Wiseman's. His Quirkology project "tracked over 3000 people" through 2007. At the start 52 percent were confident of success, and a year later 12 percent had achieved their goal. The result is published on the Quirkology site, and the page cites no paper. Its 12 is half again larger than the 8.
The other is Oscarsson, Carlbring, Andersson and Rozental (PLOS ONE, 2020), which its authors call "probably the largest and most comprehensive study on New Year's resolutions conducted thus far." They randomized 1,066 people into three groups (active control, some support, extended support) and followed them through 2017, at the end of January, June and December for the control group and at the end of every month for the other two. At one year 55 percent of responders considered themselves successful. Participants with approach-oriented goals did better than those with avoidance-oriented goals, 58.9 percent against 47.1 percent. The paper states that success was self-reported and "not objectively confirmed," the 55 percent counts only those who responded at one year, and three of the four authors receive royalties for a book on New Year's resolutions based in part on the study.
So the published figures run from 12 percent at a year (Wiseman) to 19 percent at two years (Norcross 1988) to 46 percent at six months (Norcross 2002) to 55 percent at a year among responders (Oscarsson).
Quitter's Day has a date, and it moves
The 80 percent line ends at an opinion piece. U.S. News, December 29, 2015, by Joseph Luciani: "By the second week of February, some 80 percent of those resolution-ers are back home with a new kind of remorse staring back at them in the mirror." No study is cited. Inc. repeated it in 2020 ("Approximately 80 percent of people who make New Year's resolutions have dropped them by the second week of February") without adding one.
Strava computed Quitter's Day from its own activity uploads and published a date. The first was Friday, January 12, 2018, reported the same day by The Independent: "By analysing over 31.5 million global January activities, Strava was able to pinpoint the second Friday in January as 'Quitters' Day'." So "the second Friday" is Strava's own 2018 wording. The next releases didn't keep it. In 2019 it was Thursday, January 17, in the US (Women's Running, from "more than 108 million activities in the United States and almost 685 million activities worldwide") and in the UK (road.cc, from 98.3 million UK activities). A news.com.au piece run by Fox News on December 31, 2018 had given Saturday, January 12, from the "31.5 million online global activities last January," the 2018 count. In 2020 it was Sunday, January 19, from "more than 822 million activities."
In 2022 Strava named Monday, January 17, and then, per The Independent's updated piece, "the company has now said that the date is no longer accurate." A 2025 TechRadar article says "Strava says that Quitter's Day falls on the second Tuesday of January each year." I couldn't find that wording on any Strava page. Searching press.strava.com and stories.strava.com for "quitter" returns nothing, and the three Year in Sport releases since January 2024 contain neither "Quitter" nor "Tuesday." National Today lists Quitter's Day 2027 as January 8, which is the second Friday.
The same Strava reports carried two other numbers. Strava's 2018 Year in Sport, from 36 million athletes in 195 countries, found 88 percent of runners who set mileage goals still pursuing them six months later, and road.cc's summary of the same data says "94% of athletes who set goals remain active nine months later."
How the 8 kept its label
The Huffington Post, reposted by APS on January 4, 2016, wrote "A whopping 92 percent of people who set resolutions don't succeed, according to University of Scranton research."
At each hop the citation got shorter. Shapiro named his pollster, sample and margin of error. Statistic Brain named a university and a journal. Forbes named a university and linked the aggregator. Balance Bar footnoted the aggregator's label as a journal citation with a month. By 2016 the label was all that was left, and "University of Scranton research" was attached to 92 percent, the number Shapiro's 2005 post opened with ("92% of resolutions fail").
Limits
Forbes, Inc. and U.S. News block fetches from my network even in a headless browser, so those three were read from Wayback captures, not live. Shapiro's question wording and field dates aren't published anywhere I found, and I found no wire release for the 2005 poll. The "8 percent always achieve" figure is his reported result, not a quoted question. The 1988 paper's one-month, three-month and six-month figures come from the 2020 paper's summary of it, since the 1988 abstract gives only 77 and 19. The "second Tuesday" wording has only TechRadar as a source.
If you cite a resolution number this January, the one with a paper and a comparison group behind it is 46 percent of resolvers still succeeding at six months, against 4 percent of people with the same goals who didn't make a resolution.