Our sources

In brief: every number on this site traces back to a named primary source, listed below, and every widely-repeated figure we could not trace is listed too, with the reason it was turned away.

We only publish numbers we can trace to a primary source: the original study, official statistics, or the organization's own reporting. Secondary retellings, influencer polls, and unattributed "studies show" claims are rejected at the fact-gate.

What makes a source primary?

A primary source is the original report, dataset or dictionary entry a number comes from, rather than an article or listicle that quotes it secondhand. For instance, when this site cites Pew Research Center, 2020, the link goes to Pew's own report page, not to a roundup that mentions Pew.

A probability sample is a survey built so every person in the population has a known chance of being asked, which is what lets a result generalize beyond the people who happened to answer. An opt-in panel is the opposite: people choose to take part, usually because a brand invited its own customers, and the result describes only that self-selected group. Consider a dating app surveying its own users about how long couples waited to meet each other's kids — a real number, but not one that describes daters in general. We reject figures shaped like that; see the rejected list below.

Which sources do these numbers come from?

Below is every publisher currently cited on this site, deduplicated by URL. For example,Pew Research Center, 2020 is the source behind the "friends with benefits" awareness figure quoted above, and it appears once here even though several pages reference it.

• Dictionary.com (2026)Dictionary entry read directly on 2026-08-21; definition, 'Informal' label and origin note ('First recorded in 2005–10; situation + (relation)ship') quoted verbatim from the page.
• Dictionary.com (2026)Dictionary entry read directly on 2026-08-21; definition quoted verbatim.
• Dictionary.com (2026)Dictionary entry read directly on 2026-08-21; the relationship sense (noun and verb) quoted verbatim. No origin note present for this sense.
• Dictionary.com (2021)Editorial article '10 New Dating Slang Words To Know In 2022', published 18 Nov 2021, read directly; wording quoted verbatim.
• Oxford University Press (2023)Publisher's own announcement, dated 4 December 2023, read directly. States the four finalists were 'rizz, Swiftie, prompt, and situationship' and that 'over 30,000 language lovers around the world got involved'.
• Pew Research Center (2020)American Trends Panel Wave 56, 4,860 US adults, fielded Oct 16–28 2019.
• Pew Research Center (2020)American Trends Panel Wave 56 topline, question DATEVOCAB item b, ASK ALL, n=4,860, fielded Oct 16–28 2019.
• Pew Research Center (2020)American Trends Panel Wave 56 topline, question DATE10YR, ASK ALL, n=4,860, fielded Oct 16–28 2019. Distribution: harder 47, easier 19, about the same 33.
• Pew Research Center (2023)Survey of 6,034 US adults, fielded July 5–17 2022, published Feb 2 2023.
• Pew Research Center (2023)Survey of 6,034 US adults, fielded July 5–17 2022, published Feb 2 2023.
• YouGov (2024)YouGov Surveys: Serviced. Online survey of 1,110 US adults, fieldwork 24–25 January 2024, weighted to US Census Bureau population estimates. Age-band subgroup of that sample.

What did we reject, and why?

This list matters as much as the one above. These are figures that circulate widely in writing about situationships, the talking stage and modern dating slang, and that we could not trace to a primary source, so they do not appear anywhere on this site.

• Any figure for how long a situationship lasts (average, typical, median, 'a few months') — No probability-sample or peer-reviewed measurement of situationship duration exists. Nothing was found on any primary source. Note that vague quantifiers ('usually a few months', 'most last under a year') are the same fabrication without the digits and are equally banned by content-spec §0. /situationship/ FAQ must answer 'how long do situationships last' by explaining that nobody has measured it.
• Any figure for how long the talking stage lasts or should last (e.g. '2–8 weeks', '1–3 months') — Nothing defensible exists. The circulating numbers are brand marketing: a 2023 eHarmony survey of its own users ('57% of singles spend 1–3 months from first contact to defining a relationship') and Helen Fisher's '12 dates / 6–8 weeks', which comes from Match's Singles in America — a survey commissioned by a dating company about its own market. Neither is measurement. /talking-stage/how-long/ — the top key_score page in the niche — must be built on this absence and answer through signs of progress.
• Stir / Match Group: '61% of single parents wait up to three months before introducing a new partner to their kids' — Product marketing, not measurement. Confirmed as a survey of 1,494 members of Stir, Match Group's own dating app, released as a launch press release (21 March 2022). A self-selected sample of one app's customers cannot describe single parents, and the figure circulates as if it were a norm. Also off-topic for this site's boundaries.
• Any figure for the share of situationships that turn into relationships — No source measures this. Anything shaped like 'X% work out' would also be a verdict dressed as data. /situationship/turn-into-relationship/ answers with mechanics, not odds.
• 'Merriam-Webster added situationship to the dictionary in September 2023' — Could not be verified at the primary source: merriam-webster.com returns HTTP 403 to automated fetching, and the evidence conflicts — the dictionary entry's own metadata surfaced an 'entry added' date of 24 February 2026, while secondary coverage of the September 2023 batch of 690 new words does not list situationship in any list actually read (Library Journal infoDOCKET's list omits it; the two syndicated wire copies were 404/403). Medium confidence needs two reputable secondaries reporting the primary; one could not be opened and one contradicts. Writers must use the Dictionary.com entry and the Oxford 2023 finalist claim instead, and must not assert any Merriam-Webster addition date.
• 'Delusionship is in the dictionary' / any dictionary status for delusionship — The only dictionary-branded page is a Collins 'New Word Proposal' — user-submitted content pending review, not an accepted entry. User submissions are not authoritative. /terms/delusionship/ must define the word in the site's own words and state plainly that no major dictionary has an entry for it. Reinforces risk boundary 4: the word is slang, not a condition.
• 'Benching' has a dictionary entry in the dating sense — Checked: Dictionary.com's entry for 'benching' carries only the sports/workplace sense. Only the editorial dating-slang glossary covers it, which is approved separately with an explicit 'not a dictionary entry' constraint.
• Ghosting's dating sense dates to 1955–60 — Misreads the source. Dictionary.com's 'First recorded in 1955–60' origin note attaches to the general noun 'ghosting' (as in a television image), not the dating sense. No first-use date for the dating sense was traceable, so none may be published.
• YouGov 2024: how situationships began — 55% in person, 17% via social media, 7% via dating apps — Rejected on base ambiguity. The published article states the figures both as shares of situationships and as shares of 'all Americans' in adjacent sentences, so the population the percentages describe cannot be pinned down from the primary page. The headline prevalence figures from the same survey (39% / 50%) are approved because their base is stated unambiguously.
• Thriving Center of Psych 2023 Gen Z/Millennial ghosting survey; Forbes Health 'ghosting impacts 76% of people who are dating' — Commissioned/branded opt-in web polls published as content marketing, with no probability sampling and no published weighting to a population frame. Pew's ghosting figures cover the same ground with a real methodology and are approved above.
• Ghosting statistics from aggregator sites (gitnux.org, leftunsaid.store, datingadvice.com round-ups, hilainie.com 'Situationship Statistics 2026') — Aggregators, several with the signature of machine-assembled statistics listicles: numbers without a traceable instrument, chained citations, and in places figures attributed to studies that do not contain them. Nothing sourced this way enters the gate.
• Freedman, Powell, Le & Williams (2019) and Powell et al. (2021) ghosting percentages (e.g. ~25% ghosted, ~20% have ghosted) — Peer-reviewed, but the samples are online convenience panels (MTurk-style), not probability samples of any population, and the papers' aim is testing implicit-theory predictors rather than estimating prevalence. Publishing their percentages as prevalence would misdescribe the method. Pew covers prevalence properly. The papers remain usable as conceptual background with no number attached.
• 'Escaping the Situationship' (Partners Universal International Innovation Journal) and similar low-barrier-journal situationship papers — Journal shows the markers of a pay-to-publish outlet with no verifiable peer review; the paper is a narrative synthesis with no original sampling. Not an authoritative source.
• Pew 2023: 'a quarter of adults under 30 have used a dating app in the past year' — Appeared only in a search-result summary; the figure was not located and read on the primary Pew page during verification. Rejected on the read-it-yourself rule rather than on plausibility. The two under-30 and past-year figures that WERE read directly are approved above and cover the same ground.
• Any clinical prevalence figure — rates of anxiety, depression, obsessive thinking or attachment disorder among people in situationships — YMYL boundary 2. Even a correctly sourced clinical figure would invite the site to diagnose readers, which is outside its lane. Pages route to a professional in words, with no number.
• Any figure that grades the participants — 'X% of situationships are toxic', 'X% of men never commit', 'X% of women want more than their partner does' — Tone boundary 1. A statistic is the easiest way to smuggle a verdict about people onto a page that is supposed to describe mechanics. Barred by policy regardless of source quality.
• A dictionary or glossary definition for 'haunting' (the post-ghosting term) — No entry found in any authoritative dictionary or publisher glossary that was opened. /terms/zombieing-haunting-orbiting/ may define haunting in the site's own words but must not attribute the definition to any publisher, and must not imply it is a dictionary term.

What rules govern a citation here?

Four rules, applied to every figure before it ships. Rule two is why a lexicographic entry like Dictionary.com, 2026 links straight to the dictionary page rather than to an article describing it.

1. Named source in the same sentence as the number, with the year.
2. Primary source only — we link the origin, not an article about it.
3. Dated verification: every statistic carries its verification date on the statistics page.
4. Unverifiable → unpublished. No exceptions.