The Restaurant Mint Study: The Real Numbers Behind the Reciprocity Effect, Read From the Paper

We sell a $35 logo mug gift box, printed, packed and shipped to each person's address, so we have a stake in how this study is read. This page takes its numbers from the paper itself.

Key facts

• The paper: David B. Strohmetz, Bruce Rind, Reed Fisher and Michael Lynn, "Sweetening the Till: The Use of Candy to Increase Restaurant Tipping", Journal of Applied Social Psychology 32(2):300–309, 2002 (Cornell eCommons).
• Experiment 1 (92 dining parties): mean tip 17.8% when each diner got "a fancy, foil wrapped piece of chocolate" with the check, against 15.1% with none.
• Experiment 2 (80 parties, 20 per condition): no candy 18.95%; one piece 19.59%; two pieces 21.62%; one piece plus a second offered as the server left 22.99%.
• 21%, not 23%: the authors estimate the one-plus-one approach would have raised the server's tip total by 21% ("a 21% increase"). The 22.99% is a tip rate, not an increase.
• No mints: Experiment 2's server offered "Hersey Assorted Miniature Chocolates" (the manuscript's spelling) from a wicker basket.

It Was Chocolate, Not Mints: The Paper Behind the Story

The popular retelling calls it a mint study (one version is quoted below). The paper describes candy, and the candy was chocolate.

The authors ran two field experiments in real restaurants, during real dinner shifts, with servers acting as their accomplices. Each experiment tested one question:

• Experiment 1: does a small piece of candy with the check raise tips at all?
• Experiment 2: does it matter how much candy, and how it's offered?

They set out to test three explanations for why candy might work. The first was a good impression: the server looks friendly. The second was a good mood: "unexpected food treats have a positive impact on a people's moods and subsequent likelihood of helping others", citing Isen and Levin's 1972 cookie study. The third was reciprocity: "People often feel obligated to reciprocate acts of generosity even if those acts were not requested or anticipated." For that they cited Regan (1971), in which "research participants who unexpectedly received a bottle of Coca-Cola from a fellow research participant were more likely to reciprocate this 'favor' by buying raffle tickets from the fellow participant."

All quotes here come from the authors' manuscript on Cornell eCommons, read in full on Oct 29, 2026.

Experiment 1: One Chocolate With the Check, 92 Tables

Where and when: "Ninety-two dining parties at a restaurant in downtown Ithica, NY" (the manuscript's spelling of Ithaca), over two weekday and two weekend dinner shifts in the fall of 1992. The restaurant seated about 66 people.

How it was randomized: just before delivering a check, each of two "seasoned waiters" drew a playing card from a shuffled deck. A red card meant each diner got "a fancy, foil wrapped piece of chocolate". A black card meant no chocolate. The fancy chocolate was chosen "to differentiate this token gift from the inexpensive wrapped mints that are often provided by restaurants."

What happened:

The difference was statistically significant (t(90) = 5.25, p < .0001, effect size r = .48). By our arithmetic, 17.8% is 2.7 points, or about 18%, above 15.1%. Paying by credit card also raised tips (17.1% against 15.3%), and neither the payer's sex nor the payment method significantly changed the candy's effect.

Experiment 2: The Same Two Candies, Given Two Ways

Where and when: 80 dining parties (293 customers, 2 to 12 per party) at an upscale Italian-American restaurant in central New Jersey, over several weeks in the spring of 1998. One female server with about 5 years' experience ran every table.

How it was randomized: the server drew the top card from a shuffled stack of index cards. Each card told her which of four things to do when she delivered the check:

• No candy: the check as usual.
• One piece: each diner chose one candy from the basket.
• Two pieces: each diner chose two.
• One plus one: each diner chose one; then, "as the server was leaving the table, the server stopped and offered her customers the choice of an additional piece of candy."

In every condition she was told to thank the party and leave the table right after delivering the check. Here is the paper's Table 1, with our arithmetic in the last two columns:

Three planned comparisons all came out in the predicted direction. Candy beat no candy (effect size r = .41). Two pieces beat one (r = .46). And the one-plus-one gesture beat two pieces chosen together (t(76) = 2.06, p = .02, one-tailed, r = .22). The amount of candy was identical in the last two rows. Only the way it was given changed.

Is It 23%? The Paper Says 21%

One popular version is on Influence at Work's page "Dr. Robert Cialdini's Seven Principles of Persuasion." It says a single mint "typically increased tips by around 3%", two mints gave "a 14% increase in tips", and when the waiter "pauses, turns back and says, 'For you nice people, here's an extra mint,' tips go through the roof. A 23% increase" (Influence at Work, read Oct 29, 2026).

Set against the paper:

The 3% and 14% match the paper. The 23% doesn't. The mean tip in the one-plus-one condition was 22.99%, which rounds to 23%, but that is the tip rate, not the increase.

The authors' own estimate is in dollars. In Experiment 2 the server's customers spent $5,810.15. With no candy, the tips would have totaled $1,101.02; with the one-plus-one approach at every table, "an increase of $234.73 (i.e., a 21% increase)." One detail for anyone checking: the manuscript prints that total as $1,235.75, but its own figures ($5,810.15 × 22.99%) give $1,335.75, which is the total that matches the $234.73 and the 21%. For Experiment 1 the same arithmetic gives "an increase of $75.96 (i.e., a 18% increase)."

Why It Worked: Reciprocity Beat Mood and Impressions

The authors' conclusion: "reciprocity is a stronger explanation for these findings than either impression management or the good mood effect."

Their reasoning runs on the table above:

• Not just impressions. If looking friendly were the whole story, two pieces shouldn't beat one. They did.
• Not just mood. If the treat simply lifted moods, two pieces chosen together should match one-plus-one. They didn't: the extra gesture earned more for the same candy.
• Reciprocity fits both. A bigger and more personal gesture may have created a bigger sense of obligation, the authors argue. In support, the effects of the two-piece and one-plus-one conditions grew with the size of the party.

Their summary of the practical value: "Inexpensive and personalized gestures of gratitude such as giving one's customers candy at the end of a meal could mean millions of dollars of extra income annually for the nearly two million servers in the U.S."

What the Study Can't Tell You

• How big the effect is in general. Each experiment ran in one restaurant. Experiment 1 had two waiters; Experiment 2 had one server and 20 parties per condition.
• Anything about loyalty. The outcome is one tip on one bill, not repeat visits or referrals.
• Whether a line like "For you nice people" matters. The paper reports no script for the second offer.
• Who paid the bill. In Experiment 2, "Due to an oversight, neither sex of bill payer nor method of payment was recorded."
• Whether it holds today. The data are from 1992 and 1998.
• Not blind. The servers knew each table's condition, and in Experiment 1 "two or three dining parties were assigned to one condition in order to achieve equal sample sizes."

The study is strong for what it is: a randomized test in real restaurants, with the server's choice of condition set by a shuffled deck or stack of cards. Its broader lesson rests on a wider body of work on small gifts, collected in 9 studies on tiny gifts with outsized results.

What It Suggests for a Client Gift (Our Reading)

Not a finding about client gifts, our reading of a restaurant study:

• The gesture beats the quantity. The same two candies earned more when the second came as an extra. A gift that comes unasked, outside a transaction, has some of that quality.
• A treat plus a personal touch. Experiment 2's best result paired candy with a personal moment from the server. A gift with a treat inside and a note from a named person pairs the same two things. The research on why the note carries weight: why the note matters.
• No ask attached. The server left right after the check, with no request. A thank-you gift with no pitch inside keeps that shape.

What the research as a whole says, and doesn't say, about client gifts: do client gifts actually work?

What to Know Before You Order

• Printed, not handwritten. The card is your message, printed on a textured card, with each person's first name filled in from your address list. One message per order (up to 300 characters; the order page tells you if a message is too long to print).
• The treats. LINDOR truffles contain milk, soy and barley malt (gluten); read the labels.
• Gift limits. Each box is $35, delivered. Some employers and industries cap the gifts their people may accept; the limits, quoted from their sources: client gift limits by industry.
• Where we ship. The lower 48 states and Washington, DC, except Texas for now. No Alaska, Hawaii, Puerto Rico, other US territories, APO/FPO or international. The order page flags any row we can't ship to, so you can fix it or take it out.

Where Personalized Gifts Delivered Fits: A Treat Inside, and a Separate Note on Top

A box can't turn back at the table. What it can carry is the idea behind Experiment 2's best result, by our reading: a treat, plus an extra, personal gesture. Each Personalized Gifts Delivered box holds both:

• The treat: a 1.5 oz packet of Ghirardelli premium hot cocoa and 4 Lindt LINDOR milk chocolate truffles, packed inside the mug.
• The second touch: your message on a printed card that greets each person by first name, in its own envelope closed with a gold wax-seal sticker, lying on the lid so it's read before the box opens.
• The keepsake: an 11 oz white ceramic mug with your logo printed small, in a matte-black gift box with a magnetic lid, under black tissue closed with a gold "Happy Holidays" seal.

You could assemble this yourself. By our estimate, 25 boxes take about 9 hours of hands-on work (a mug printer, about 8 supply orders, 25 labels, 25 cards, 25 boxes to pack), plus a week or more waiting for supplies. Per box it comes out about even with our $35: about $38 with USPS counter postage and about $34 with commercial postage, with $1,192.84 paid up front for the mugs, the retail supply packs and 25 postage labels, as of Oct 29, 2026 (DIY vs ordering, itemized).

With us: add your logo (or type your website), write one message, add your address list, check the preview of the mug and every card, and pay $35 per box, delivered, with no minimum. We print, pack and ship each box to its own address: a logo mug gift box with Ghirardelli cocoa, Lindt truffles and a personal card.