Mark ở vũ trụ khácVIMark học AI
Mark ở vũ trụ khác · Another Universe Famous experiments · Social psychology

The Copy Machine Line

may I go first "because…"

The Copy Machine Study · Ellen Langer, Arthur Blank and Benzion Chanowitz, 1978

WhoEllen Langer and colleagues · Graduate Center library, CUNY
Whenpublished 1978 · re-run and criticised in 1985 and 2024
Size120 people at the copier, only 15–25 per group (Langer et al., 1978)
Verdictgood idea, thin evidence · robustness 5 / 10
Reread in 1 minute

Essence

Cutting in at the copierA stranger asks to use the machine first, phrased in 3 different ways (Langer et al., 1978).
"Because I have to make copies" still worksWith 5 pages: 93% said yes, close to the real reason's 94%, above no reason at 60%.
20 pages is differentThe empty reason got 24%, exactly the same as no reason. Only the real reason rose, to 42%.
Small favor: autopilotLarge favor: people wake up and check whether the reason makes sense.
Small samples, few re-runs15–16 people per "small favor" group. A 1985 re-run did not reproduce the result.
Part 1

The question behind the experiment

When someone asks us for something, do we actually listen to the reason, or does hearing "because" make us nod? Langer's answer: for small favors, many people nod out of habit; for big ones, people start to listen. But the evidence for this answer is thinner than its fame.

1978The experimentLanger sets out to show that much behavior that looks thoughtful actually runs on scripts.
→
1984 onwardsThe "magic word"Persuasion and sales books retell it: add "because" and people agree.
→
1985 and 2024RecheckingFolkes re-runs it and does not get the old result. Gelman recalculates: the samples are too small to be sure.
Part 2

How the experiment worked

A field experiment: participants did not know they were taking part.

In a library, someone is using the copy machine. A stranger asks to go first, with 5 or 20 pages, in one of three ways: just asking; asking with an empty reason, "because I have to make copies"; asking with a real reason, "because I'm in a rush". The researchers recorded whether the person at the copier let them in. copy machine person copying person asking 5 or 20 pages 1 · "May I use the xerox machine?" 2 · … "because I have to make copies" 3 · … "because I'm in a rush" reason 2 adds nothing: everyone in line makes copies
The original wording: "Excuse me, I have 5 (20) pages. May I use the xerox machine?", with or without "because I have to make copies" or "because I'm in a rush". 120 copier users at the library of the Graduate Center, City University of New York (Langer et al., 1978).
Episode 1

The experiment and "because"

3 ways to ask5 pages, 20 pageswhy it got famous
Episode 2

Autopilot mode

scriptsthe office memoasking at work
Episode 3

Rechecking

15 people per groupFolkes 1985recalculated 2024lessons
Part 3

Key concepts

The "empty" reason

placebic informationLanger et al., 1978
The three requests, judged on two things: does it have the form of a reason, and does it carry new information. No reason: neither. Empty reason: the form "because…", no new information. Real reason: both. "because…" form? new information? no reason nono empty reason yesno "I have to make copies" real reason yesyes "I'm in a rush"
Placebic, like a placebo pill. It has the shell of a reason but no reason inside. If listeners react to it exactly as they do to a real reason, they are only hearing the shell.

Small favors on autopilot, large ones wake us up

effort · small vs large favorTable 1, Langer et al., 1978
Share who let the person go first. 5 pages: no reason 60% (15 people), empty reason 93% (15), real reason 94% (16). 20 pages: no reason 24% (25), empty reason 24% (25), real reason 42% (24). 050%100% 60%none 93%empty 94%real 5 pages · 15–16 people per group 24%none 24%empty 42%real 20 pages · 24–25 people per group
This is the full result. Few retellings include the right half: when the favor costs more, the empty reason stops working. The authors themselves saw this as the main point.

Scripts and "mindlessness"

mindlessness · scriptsLanger, 1978 and 1989
Autopilot: hearing the familiar form "may I… because…" runs the script "a reason means yes". Alert mode: weighing what the reason says. We switch to alert mode when the favor takes effort, or when the message has an unfamiliar form. AUTOPILOT hears "may I… because…" → "a reason, so yes" fast, saves effort ALERT MODE weighs the reason "in a rush? with 20 pages?" slow, takes effort switches on when: the favor takes effort · the message looks unfamiliar
"Mindless" is not an insult here. Langer uses the word for following a familiar pattern without weighing the content. Most of the time that works, because familiar patterns are usually right.

Familiar-looking documents get obeyed unread

structural cues · interoffice memoExperiment 3, Langer et al., 1978

Learning the format from the bin

The authors collected 83 interoffice memos from secretaries; 68% shared one format: a request, unsigned.

Sending a pointless memo

40 secretaries received a memo asking them to return the memo itself to "Room 238". There was no reason to do so.

Familiar format: 90% returned it

Other formats: 60% on average. Only 10 people per group, so these numbers swing a lot.

The point: what looks familiar gets handled by habit. At work, an email in the usual template is more likely to be acted on, and also more likely to be skimmed without reading.

Re-run and recalculated

Folkes 1985 · Gelman 20241985 and 2024
Pooling 5 and 20 pages, each type of request had 40 people: no reason 37.5%, empty reason 50%, real reason 62.5%. Andrew Gelman's 2024 calculation from Table 1 of the original paper. no reason 37.5% empty 50% real 62.5% 40 people per request type · 5 and 20 pages pooled · standard error of each gap about 11 points
Pooled, the empty reason is not "as good as" the real one. Gelman (2024) worked out that each 12.5-point gap between neighbouring groups is only about one standard error (about 11 points), so the sample is too small to be sure. Folkes (1985) re-ran the same copier situation in 4 studies and did not reproduce the original result.
Part 4

Evidence: how far can we trust it

10 common claims about the experiment, checked against the original paper and later checks. Tap each row to see why.

SOLIDPeople agree to small favors far more readily than large ones

In the original paper this is the strongest difference (5 pages vs 20 pages, p < .001), and it fits a great deal of later research on the cost of helping.

SOLIDWhat the reason says matters, especially when the favor takes effort

Even Folkes (1985), who disputed Langer, found that the type of reason changed compliance: reasons within the asker's control got less help than reasons outside it. So listeners do weigh reasons.

SOLIDWe underestimate how often a direct request gets a yes

Not part of Langer's experiment, but the lesson that goes with it. Flynn and Lake (2008): to get strangers to fill in a questionnaire, participants guessed they would need to ask about 20 people; in fact about 10 was enough. Bohns's (2016) review, across more than 14,000 strangers: we underestimate how many will say yes by about 48%.

CAUTIONFor small favors, an empty reason works as well as a real one (93% vs 94%)

Each group had only 15–16 people: 93% is 14 of 15, 94% is 15 of 16. The direction is clear, but one or two people changing their mind would change the numbers a lot.

CAUTIONFor large favors, people wake up and only real reasons work

42% vs 24%, about 25 people per group. The original paper itself reports this comparison as only marginal (p between .05 and .10).

CAUTIONThe word "because" by itself triggers a yes-script

This is Langer's interpretation, not something measured directly. Folkes (1985) argued her results fit people thinking rather than running a script. Langer and colleagues wrote a reply standing by their view.

CAUTIONDocuments in a familiar format get followed without close reading

The office memo experiment: 90% vs 60%, only 10 secretaries per group, 40 in total. Plausible and interesting, but one small experiment.

CAUTIONMost everyday human behavior is "mindless"

The big idea Langer developed later (the book Mindfulness, 1989). Three small experiments in 1978 can suggest it, not prove a claim that broad.

CRITICISEDAdd "because" and people will agree

The popular version in sales books and "magic word" articles. It skips the second half of the experiment: with 20 pages, the empty reason got exactly 24%, the same as no reason. Gelman (2024) shows that figures like "more than 50% more likely" misread the table.

CRITICISEDThe result has been replicated many times

Mark has not found a direct replication that reproduced the original result. The most cited re-run, Folkes (1985), with 4 studies, did not reproduce it. So treat this as an idea worth trying, not a law.

Part 5

Verdict

The "running on scripts" idea
3.5/5
Original study design
2.5/5
The "magic because" claim
1/5
Value of the lesson
4/5
Strengths
  • Run in a real setting, people did not know they were observed
  • Includes both small and large favors, so the result has nuance
  • Easy to understand, easy to try in daily life
Weaknesses
  • 15–25 people per group, one library, one moment in time
  • The key comparison for large favors is only marginal
  • The 1985 re-run did not reproduce it; few direct replications

Robustness 5/10. The experiment raises a good idea: we nod to small favors by habit and only listen for big ones. But the 93% figure is asked to carry more than 15 people can bear.

More: why such a small experiment became so famous

In 1984, Robert Cialdini retold the experiment in Influence, one of the best-selling books on persuasion. From there, the "magic because" spread into sales courses, copywriting and business articles, usually with only the first half of the result.

The story has everything it needs to spread: one simple word, a very high number (93%), and a trick anyone can try right away. The second half (20 pages, empty reason does nothing) is less catchy, so it is told less, even though it is what makes the experiment valuable.

Gelman (2024) uses this very experiment as an example of "mindlessness in the interpretation of a study on mindlessness": retellers look at the nice number, not at the sample size and the error.

Part 6

Lessons to apply: ask with a reason

A work skill: asking colleagues, bosses and partners in a way that gets a yes, and staying alert when you are the one being asked.

Four steps to asking: say exactly what; give a real reason; say how long it takes and by when; the bigger the favor, the more specific the reason and why this person. 1 · what"check slide 3"specific, one task 2 · why"it goes to the client at 5"a real reason, not an empty one 3 · how much"about 10 minutes"and when: "before 4" 4 · big asks"you know this part best"why this person Small favor: a polite "because" is enough. Big favor: people will listen closely, so the reason must be real.
Don't use empty reasons for big asks. In the original study, with 20 pages, an empty reason did no better than none. At work, an empty reason also costs trust once people notice.
SituationEasy to refuseWith a real reason
Email to a colleague"Can you take a look at this file?""Could you check the pricing section before 4? It goes to the client at 5. About 10 minutes."
Asking the boss for staff"The project needs more people because it needs more people.""I need 1 more person for 2 weeks, because testing is 5 days behind schedule."
Asking for overtime"Can you work this weekend?""The client moved the meeting to Monday and we need the demo. You know this part best. Take Tuesday morning off in return."
Always add a "because"Small favors: a short reason makes the ask easier to hear.
Big asks, real reasonsWhy, why you, how long, what's in it for you.
Just askPeople say yes more often than you guess (Flynn and Lake, 2008).
When asked for something bigPause and ask yourself: does this reason actually tell me anything?
People don't just need a reason. When the favor is big enough, they need a reason that means something.Paraphrasing the conclusion of Langer, Blank and Chanowitz (1978), not a direct quote.
Part 7

Read next

Langer, Blank and Chanowitz, "The mindlessness of ostensibly thoughtful action" (1978)The original paper, short. Read Table 1 first, then the discussion.
Robert Cialdini, Influence (1984)Where the "because" story became famous. Read it alongside Part 4 here.
Ellen Langer, Mindfulness (1989)The mindless/mindful idea developed into a book.
Daniel Kahneman, Thinking, Fast and SlowSystem 1 runs on autopilot, System 2 checks: close relatives of this experiment. Volume 3 of the book corner.
Part 8

Sources for the figures

Every number in this page, the Facebook carousel and the captions comes from the sources below.

Langer, Blank and Chanowitz (1978)“The mindlessness of ostensibly thoughtful action: The role of ‘placebic’ information in interpersonal interaction”, Journal of Personality and Social Psychology 36(6), 635–642 · Numbers used: 120 people (68 men, 52 women) at the copier in the Graduate Center library, CUNY; 5 pages: 60% (15 people), 93% (15), 94% (16); 20 pages: 24% (25), 24% (25), 42% (24); the 20-page comparison only marginal (p between .05 and .10); office memo experiment: 83 sample memos, 68% in one format, 40 secretaries, 10 per group, 90% vs 60%
Folkes (1985)“Mindlessness or mindfulness: A partial replication and extension of Langer, Blank, and Chanowitz”, Journal of Personality and Social Psychology 48(3), 600–604 · Numbers used: 4 studies (82, 42, 96 and 69 people); did not reproduce the original result; reasons within the asker's control got less help than reasons outside it
Langer, Chanowitz and Blank (1985)“Mindlessness/mindfulness in perspective: A reply to Valerie Folkes”, Journal of Personality and Social Psychology 48(3) · Numbers used: the authors stand by their view (no new data)
Andrew Gelman (2024)“Mindlessness in the interpretation of a study on mindlessness”, blog Statistical Modeling, Causal Inference, and Social Science, Columbia University, 6 March 2024 · Numbers used: pooled 40 people per request type: 37.5%, 50%, 62.5%; standard error of each gap about 11 points
Flynn and Lake (2008)“‘If you need help, just ask’: Underestimating compliance with direct requests for help”, Journal of Personality and Social Psychology 95(1), 128–143 · Numbers used: guessed about 20 people would need asking, in fact about 10, to get questionnaires filled in
Bohns (2016)“(Mis)Understanding our influence over others: A review of the underestimation-of-compliance effect”, Current Directions in Psychological Science 25(2), 119–123 · Numbers used: more than 14,000 strangers asked; number who agree underestimated by about 48%
Robert Cialdini (1984)book Influence: The Psychology of Persuasion · Numbers used: no numbers; the source that made the "because" story famous