It takes 50 hours to make a friend? The paper gives two different answers

The paper behind "it takes 50 hours to make a friend" reports two sets of threshold estimates, and neither includes 50. Jeffrey Hall's study, published online in 2018 in the Journal of Social and Personal Relationships, found that an acquaintance becomes more likely than not to be called a casual friend at about 94 hours together in one sample and 43 hours in the other. [1]

The familiar sequence of 50, 90 and 200 hours comes from a University of Kansas press release, which rounded Hall's combined reading of both studies. [2] That combined reading is reasonable. The numbers still mean something narrower than the way they are usually quoted, and the paper says little about a slow schedule such as a weekly dinner: its participants averaged 13 to 19 hours a week with the person they described.

What each number measures

Hall ran two studies of people who had recently moved.

• Study 1 analysed answers from 355 US adults recruited online (429 took part; 17% failed a screening check) who had moved at least 50 miles in the previous six months. Each described one person met since moving, excluding old friends, romantic partners and relatives. On average they had known that person for 11 weeks, with a range of a few days to 26 weeks.
• Study 2 followed 112 first-year students at a Midwest public university who had moved to town in the two weeks before classes began. Each described two people met since arriving. Surveys ran in weeks 3, 6 and 9 of the semester; the hour estimates come from the first wave, about three and a half weeks after meeting.

Hours were self-reported: last week's hours plus a typical week's hours multiplied by the weeks since meeting. The Study 1 prompt asked for hours "actually in their company or communicating with them" and added: "not just you were working the same shift." [1]

Each estimate comes from a logistic regression comparing two adjacent labels, such as people who said the person was not yet a friend against those who called them a casual friend. The reported figure is the cumulative number of hours at which the predicted chance of the closer label exceeds 50%, with the other variables in the model held at their averages. At that many hours, the model gives roughly even odds of either label. [1]

Hall also describes the Study 1 data in a chart: casual friendships "emerge around 30 hr", friendships around 50, good friendships after 140 and best friendships after 300. Those describe where each label starts to appear among respondents, a looser measure than the cut-points in the table. [1]

[1]

Why the two studies disagree

Hall treats the adult figures as likely too high. The adults were recalling relationships about three months old, so hours kept accumulating after a relationship had already changed category, and recall over many weeks is less precise. The students were asked about three weeks in. [1]

He also suggests a second explanation: students living in dorms have unusual amounts of shared time, and working adults may genuinely need more hours to reach the same label. The paper leaves that open. For a reader, the practical conclusion is that the answer depends on who is making friends and at what stage, so a single number overstates what is known.

The figures are running totals

The press release says it takes "roughly 50 hours" to become a casual friend and "90 hours to go from that stage" to friend. [2] Read literally, that sounds like 50 hours followed by another 90, or 140 in total before someone becomes a friend. In the paper, every estimate counts total hours since meeting. On Hall's combined reading, the friend threshold falls somewhere around 80–100 total hours, not 140, and "200+" is a total too.

The release also describes the student follow-ups as "four and seven weeks later". The paper places them in weeks 6 and 9 of the semester, with table notes of 6.3 and 8.4 weeks after meeting. Where the two differ, use the paper.

The kind of time mattered

In Study 1, the share of time spent working or in class together was associated with less closeness. Time spent hanging out, and watching TV or films or playing games together, was associated with more. Hours predicted closeness more strongly for relationships people chose than for those formed at work or school. The share of time spent talking was not associated with closeness at all. [1]

Study 2 measured what the talk consisted of. A combined measure that Hall calls "striving" talk, made up of five items (catching up on events since you last saw each other, talking about each other's day, serious conversations, playful talk and affectionate talk), predicted increases in closeness beyond the hours spent together, in both later waves. Small talk about current events, pets, sports and TV in weeks 6 to 9 predicted a decline. [1]

Because the five items were scored together, the study cannot say that catching up alone did the work.

Your schedule against the study window

The ledger below is our own extrapolation from the paper's figures. Four rows are fictional schedules. The last two apply each study's reported average "typical week" (12.99 and 19.14 hours) as if it were sustained. Monthly events are converted at 12 per 52 weeks.

Two passages in the paper decide how to read this. Among the students, people who stayed acquaintances "rarely" spent more than 30 hours together over nine weeks. And after about three months, Hall writes, acquaintances may keep accumulating hours, "but this time does not appear to increase the chance of becoming casual friends". [1]

A weekly dinner reaches 26 hours by week 13, below the 40–60 hours Hall places in the first six weeks. The ledger therefore does not say that 25 weeks of dinners will produce a casual friend. It says that slow, steady schedules are far from the pace these studies' participants typically reported, and the paper did not test whether pace matters. Their participants built friendships during a dense early period, averaging 13 to 19 reported hours a week with one person.

The coffee row needs a flag of its own. Coffee on a break is chosen time, but it happens inside a workplace, and the Study 1 prompt excluded time spent merely working the same shift.

What the study cannot tell you

• Old friends. Both studies excluded them by design, so these hours say nothing about reconnecting. That is a different problem, covered in why the message to an old friend stays in drafts.
• Mutual choice. Hall notes that spending more time together is "unlikely to be a decision that can be made unilaterally". [1]
• Online friendships. Only 5.1% of Study 1 partners were met online, and the hours question combined time together with time communicating.
• Cause. The design is correlational, the hours are self-reported, and Study 2 drew on one university.
• Replication. We screened the titles of the 172 works that OpenAlex lists as citing the paper and found none that re-estimated the hour thresholds. [3] A title screen can miss a study, so treat the figures as one paper's estimates.

What to take from it

The practical advice Hall gave in the release is modest: "You can't make people spend time with you, but you can invite them," and if you already work together, "switch up the context" by going to lunch or for a drink. [2] The findings point toward concentrating time early, spending it in each other's company outside work or class, and using it for the conversations the student study measured: catching up, serious talk and play.

Catching up on what changed since you last spoke depends on remembering what was said last time. The keep-in-touch guide suggests leaving yourself a short note after a conversation, and Kluro's Catch-up brings earlier conversations together so you can revisit a shared plan or a change someone mentioned before you speak again.