Automatic Gratuity for Large Parties: Who Counts, Who Pays

Do large groups actually tip less?

The folk answer is yes, and it traces to one 1975 study with a memorable title. Psychologists Stephen Freeman, Marcus Walker, Richard Borden, and Bibb Latané analyzed the tips left by 396 groups of restaurant diners — observed, not assigned, which is worth holding onto — and reported that the percentage tipped fell as the table grew — a pattern they called “cheaper by the bunch.” Most tipping clustered around the 15% norm, but the percentage left was, in their analysis, an inverse function of the number of people at the table: the more diners, the smaller the slice.

Their proposed explanation — an interpretation of the pattern, not something the data measured directly — was diffusion of responsibility — the same force Latané had documented in bystander studies, where the more people present, the less any one person feels on the hook. At a table, the tip is everyone’s job and therefore no one’s. Each diner quietly assumes the others will cover it, rounds their own contribution down, and the rounding compounds.

“To the extent that many people contribute to a check, the responsibility of each to the waiter may be psychologically divided among the people present.” — Freeman, Walker, Borden & Latané (1975)

Source: Freeman, Walker, Borden & Latané, “Diffusion of Responsibility and Restaurant Tipping: Cheaper by the Bunch,” Personality and Social Psychology Bulletin (1975); diffusion quote at p. 584, as reproduced in Lynn & Latané (1984).

Is “cheaper by the bunch” actually settled science?

No — and this is the part the explainer pages skip. The group-size effect is one of the most-challenged findings in the tipping literature. Within a year, Donald Elman (1976) published a direct rebuttal titled “Diffusion or Just Deserts?,” questioning whether diffusion of responsibility explained the pattern at all and proposing that larger parties simply face larger bills, so the higher dollar cost of tipping nudges the percentage down. Cornell tipping researcher Michael Lynn later noted the catch in that account: neither the original study nor its follow-ups found a reliable link between per-person bill size and the percentage tipped, which is what a pure cost explanation would require.

The deeper problem is replication. When Lynn and Andrea Grassman re-examined the question in 1990, they found no relationship between group size and tipping in their data, and listed several earlier studies that had also failed to reproduce the effect. Worse for the original claim, Lynn and Bond had shown in 1988 that at least one — possibly two — of the literature’s reported inverse relationships were statistical artifacts of using “percent tip” as a ratio variable, not a real behavioral pattern.

Sources: Elman, “Why Is Tipping ‘Cheaper by the Bunch’: Diffusion or Just Deserts?” (1976); Lynn & Grassman, “Restaurant Tipping: An Examination of Three ‘Rational’ Explanations,” Journal of Economic Psychology (1990); Lynn & Latané (1984).

If the science is shaky, why do restaurants still impose it?

Because a restaurant isn’t running a regression — it’s managing risk on one table at a time. The question that matters behind the counter isn’t “do large parties tip less on average?” It’s “what happens to this server if this particular eight-top tips badly?” And the answer is: a lot, because two things stack up at once on a big table — high cost to serve it, and high variance in what it pays.

Start with cost. A large party isn’t four small tables’ worth of work; it’s one table absorbing a large share of one server’s attention — coordinating courses, refills, and the inevitable separate-check requests. The measured part is narrow but real: Lynn’s own research found that the amount of service a table receives keeps rising with group size, just at a slowing rate, so the server really is pouring more labor into the big top, not less. The rest is operator arithmetic rather than a research finding — attention spent on one table is attention not spent on the tables it crowds out.

Now add variance. On a smaller check, the difference between a great tip and a stiff is a few dollars — survivable. On a $400 table, the same percentage swing is tens of dollars, and a single round-number guess can quietly land the server well below the norm on the biggest check of their shift. The automatic gratuity is a floor under that downside. It doesn’t need the psychology to be real; it only needs the occasional disaster to be real and expensive — which, on a table that size, it is.

Service-by-group-size finding: Scarlett, Lynn & Latané (1982), as discussed in Lynn & Latané, “The Psychology of Restaurant Tipping,” Journal of Applied Social Psychology (1984).

Why did Olive Garden and other chains drop automatic gratuity?

Because a 2014 enforcement date made it expensive to keep. The classification itself wasn’t new. Revenue Ruling 2012-18 says its purpose is to “clarify and update guidelines first presented in Rev. Rul. 95-7,” and the criteria it applies — that a payment isn’t a tip unless the customer freely chooses it — come from a ruling issued in 1959. An automatic gratuity was already a service charge, and service charges paid out to staff already counted as wages. What changed was the deadline. The IRS applied the ruling to gratuities collected on or after January 1, 2014, and that turned a simple line on a check into a payroll problem. The headline taxes were mostly not the difference — tips already carried essentially the same income and payroll taxes, with the employer matching its share either way, though one real gap remains: a tip credit can offset the employer’s payroll-tax share on tips, and a service charge doesn’t qualify. What changed hands was everything else: withholding now runs through the employer’s payroll instead of the server’s self-reporting, the charge becomes restaurant revenue before it can be paid back out — and, redistributed, it enters the regular rate used for overtime calculations — and the money waits for the next paycheck instead of going home in a server’s pocket that night.

Darden — the parent of Olive Garden, Red Lobster, and LongHorn Steakhouse — had long added an automatic 18% to parties of eight or more. Ahead of the deadline it dropped the practice at 100 restaurants across four cities and weighed killing it everywhere. The replacement kept the math and handed back the choice: instead of an automatic charge, the check prints suggested amounts for 15%, 18%, and 20% and leaves the line blank. The math still gets done for the table; the choice goes back to the customer — which is exactly what keeps it a tip.

The ruling practically drew the map. One of its two worked examples is an automatic 18% on a large party (a service charge); the other is a restaurant that prints sample tip calculations beneath the signature line and lets the customer pick (a tip). Darden’s move ran from the first example to the second. How widely the rest of the industry followed isn’t something the reporting establishes. What the trade coverage of the period records is a direction, not a headcount: a CPA firm writing after the rule took effect opened by asking whether readers had noticed “fewer restaurants are including automatic gratuities on their checks,” and reported that “many chain restaurants have begun to rethink their automatic gratuity policy” — with calculating “suggested tip amounts on the bill” named as one of the ideas.

What didn’t happen is disappearance. Operator guidance written for restaurants in 2026 still describes automatic gratuity as a predetermined service charge added for large parties, typically six or more guests, commonly set between 15% and 22%. The charge in this guide is live practice, not a historical artifact — the printed-suggestion line is the alternative some chains moved to, not a replacement that swept the industry.

Sources: IRS Revenue Ruling 2012-18 (Examples A and B); NBC Chicago, “Restaurant Group Considers Dumping Automatic Gratuity” (2013); Baker Newman Noyes, “IRS Rule Targets Automatic Gratuity” — a CPA-firm note written the year the ruling took effect, describing a directional industry trend, not a measured share. Current practice: Rebecca Hebert, 7shifts (2026).

Who has to tell the table about the automatic gratuity?

Whoever reads it first — which at a table of eight is often the person who booked it. That job carries a cost the research has actually measured: people dislike the bearer of bad news even when the bearer plainly didn’t cause it. The restaurant sets the 18%. The organizer takes the look.

Florida shows this is a real handoff rather than a hypothetical one. The state has required food service establishments to print notice of an automatic gratuity on the menu and on the face of the bill since 1986, under a statute titled, literally, “Notification of automatic gratuity charge.” An amendment effective July 1, 2026 tightens it: the notice must now carry the amount or percentage and the purpose of the charge, in a font at least as large as the menu item descriptions, on the menu, any written contract, and the website or app where orders are placed. Every receipt needs separate lines for gratuity, the operations charge, and sales tax — the wider set of fees that same law reaches is covered in what every line on a restaurant receipt means.

Printed notice, where a state requires it, is neither new nor lax — and it still doesn’t make anyone read it. A server who has worked rooms from Delray Beach to Miami Beach described the gap to CBS12 News:

“I can think of multiple times I was yelled at for telling a guest about the gratuity. Some people do not read. And some restaurants don’t allow staff to talk about gratuities, as some people find it insulting.” — Maria Stathis, server

That is the whole mechanic in one quote. The notice is on the menu, the table doesn’t read it, and the house may bar the server from raising it out loud. So the sentence gets said at the table or it doesn’t get said at all — and when the house stays quiet, the person saying it is a diner.

What that costs the sayer is documented. Leslie John, Hayley Blunden and Heidi Liu ran eleven experiments on what they called shooting the messenger and found that people rate innocent bearers of bad news as less likeable — including in a preregistered lab study where the news came from a random drawing the messenger obviously didn’t control. Three findings matter at a dinner table — two about the wound, one about the fix.

The two fixable findings point the same way — applied to the dinner table by this guide’s inference, not by the papers’ own tests — and both are about timing rather than tone. Say it before the table sits down, so the charge is expected instead of sprung. Say why you’re raising it, so nobody has to guess at your motive. That is one message in the group chat at booking, not an announcement over the appetizers.

Sources: John, Blunden & Liu, “Shooting the Messenger,” Journal of Experimental Psychology: General 148(4) (2019) — eleven experiments; bystander specificity (Study 2A), unexpectedness (Study 4), and the motives moderation with its small-effect caveat (Studies 6A–6B). Florida law: Fla. Stat. § 509.214, as amended by s. 3, ch. 2025-113, eff. July 1, 2026. Server quote: CBS12 News.