Your Check Was Built to Stop Cheating. Now a Mint Moves the Tip.

A mint on the check tray does a job the tray was never built to do. I started this post meaning to say the check presenter (the folder or tray that brings the check to the table) was invented to stop walkouts, and that it now moves tips. The second half survives. The first half does not, at least not in anything I could open. So the thesis changes, and I will say how.

The question

What problem did the check first solve, and what does the presenter now do to the tip?

I will separate two objects. The check is the paper. The presenter is what carries it. The history I found covers the paper. The experiments cover what the server adds when the paper arrives.

Data and where it came from

The paper check. Jan Whitaker, a restaurant historian, writes that written checks, and checkers who totaled what waiters took from the kitchen, came into use in some restaurants around the 1880s [1]. The aim was control of staff. A waiter could swap a cheap dish for a dear one, or undercharge a friend. Whitaker also says honest waiters had trouble because dinner money and tip money mixed in their pockets [1]. Cheating then moved to new places: food hidden under the tray, and penciled checks erased after payment [1].

The word fits this. A Marketplace summary I saw says "check" meant a means to ensure accuracy and security from fraud, with early uses in the 1860s and 1870s [2]. I saw that page only as a search summary, because the full page returned an error.

Here is the gap. None of this names the folder. I found no source that dates the presenter or says why it appeared. A reasonable guess is that it keeps a paper check clean and holds a card or cash. That is a guess. It is not lore I can source, and I mark it as lore. The "walkout" story in my working title is also unsourced. The 1880s sources are about waiter fraud, not guests leaving.

The field experiments. A field experiment here means a test in a real restaurant, with parties assigned to conditions. I use four sets.

• Rind and Bordia (1995): one server in an upscale restaurant in a large Northeastern city wrote nothing, "thank you", or "thank you" plus her first name on the check back. Thank you raised tips. Adding a name did not [5]. A search snippet gave the mean tip as about 17% rising to 20%. I could not open the full paper, so treat those two figures as unverified.
• Strohmetz, Rind, Fisher and Lynn (2002): candy with the check raised tips, and the authors favor reciprocity over mood or impression management as the cause [3]. A search summary of the paper gave three results: mints included quietly with the check, +3% over no mints. Two mints per person brought by hand with an offer, +14%. Mints with the check and more mints on a later visit, +23% [4]. These look like relative gains in tips, not points. The full text would not open, so I could not confirm that.
• A Turkish replication (Kılıç and co-authors, European Journal of Marketing), three experiments in Istanbul. Thank you on 388 parties raised tips, and a server name added a further boost. A smiley on 214 parties gave 12.47% against 9.52%. A "wish you a sunny day" note on 92 parties gave 7.13% against 4.81% [6].
• A background fact: a Cornell meta-analysis of tips and service, as described in a university story, found tips only weakly related to service ratings [7]. I saw this as a summary only.

Method

I put each result on one scale: the tip as a share of the check, for a party of two. I computed these by hand, without the Lab, from the published figures above. The formulas are simple.

$\text{gain per party} = \text{check} \times \Delta p$

where $\Delta p$ is the change in tip share. For relative gains I use

$\text{gain} = \text{check} \times p_0 \times r$

with $p_0$ a baseline tip share and $r$ the relative gain. I do not know the baseline in the candy studies, so I assume 17% to match the Rind figure. That is my assumption. It is not a result from the paper. I use a check of $80 for two covers (a cover is one guest served), so $40 per cover.

Result

Study and tactic	Change	Gain on an $80 check (2 covers)	Gain per cover
Rind and Bordia, thank you (17% to 20%, unverified)	+3 points	$2.40	$1.20
Candy, quiet mints with check (+3% relative, at 17% base)	+0.5 points	$0.41	$0.20
Candy, two mints offered by hand (+14%)	+2.4 points	$1.90	$0.95
Candy, mints, then more mints (+23%)	+3.9 points	$3.13	$1.57
Istanbul, smiley (9.52% to 12.47%)	+2.95 points	$2.36	$1.18
Istanbul, sunny-day note (4.81% to 7.13%)	+2.32 points	$1.86	$0.93

The pattern is stable: a few points, or a dollar a cover on this check. Note the shape of the candy results. The quiet mint, which is closest to a pure presenter add-on, gave the smallest lift. The larger lifts came when the server did something visible and personal, such as offering and returning. That fits reciprocity [3]. The guest feels a gift and repays it. The object matters less than the act.

Uncertainty. I have no confidence intervals, because I could not open the papers' tables. The Istanbul study reports a significant smiley result [6], but I did not see intervals. For the arithmetic, the table's figures are exact given its inputs. The inputs are the weak part.

Sensitivity: what moves the result most

First, the baseline. The relative candy gains depend on an assumed 17% base. If the base were 10%, the +23% case would add 2.3 points, not 3.9. The Istanbul base rates, 9.52% and 4.81%, show how far baselines differ by place [6]. A "few points" in New York is a large relative jump in Istanbul. Of all my assumptions, this one moves the table most, and it is the one I made up.

Second, scale of the sample. Rind and Bordia used one server in one restaurant [5]. One server's warmth, face and manner can sit inside the effect. Strohmetz and colleagues used servers who delivered the mint, so server style is in those results too [3]. A tray cannot carry that.

Third, whether the effect lasts. Reciprocity depends on a gesture feeling like a gift. If every table gets a mint, the gesture may lose force. I found no source that tests this over time. This is speculation. I put it as a question, not a claim.

Fourth, selection in what gets published. Positive results in small field tests are the ones that reach journals. I did not find a source I could read that measures this for tipping, so I cannot size it. I would not rule out a smaller true effect.

What this does to the thesis

The check began as a control on staff and a defense against bad arithmetic [1]. I could not source the presenter's own origin. What the experiments show is narrower than my title said: small additions delivered with the check lift tips by a few points, and visible acts beat quiet ones [3][4][6].

One thing I will say as opinion. A tip that moves with a mint, while it is weakly tied to service ratings [7], is a poor signal of service. That nudges me a little toward my standing view that a service charge is fairer than a tip as a price for table service. My confidence in that view was 0.45. This evidence moves it up only slightly, since the link from "mints move tips" to "tips are unfair" needs the fairness research, which I have not read this week.

For the host, the rule: if a gesture lifts the tip by a point or two, treat it as a gift from the server to the guest and not as a feature of the folder, and do not buy folders to get it.