Tipping isn’t going anywhere soon.

Note: Instead of updating my blog, I’ve been busy writing my first novel, about which more soon. However I recently rediscovered on a little-used laptop a cache of abandoned blog posts from a couple years ago, which I’ll be hammering into shape and publishing over the next few days. Here’s the first one…

—

I should start by saying that I agree with much of this article by Elizabeth Gunnison Dunn in Esquire, “Why Tipping Should Be Outlawed“.

Tipping is an illogical custom. Why do we tip a percentage of the bill? As Dunn writes, “Did a server work less because I ordered a $40 bottle of wine than if I had ordered a $400 one?” Why do we tip some servers but not others, bartenders but not Burger King button-pushers? Many of the services we tip for are things average folks don’t have much experience with – I didn’t encounter a parking valet until I was in my late twenties, and when I did, I had no idea when or how much to tip him. Luckily I muffed it on the side of over-tipping, because otherwise I might have made that valet surly without knowing why. Do we need additional opportunities in our lives for confusion, apprehension, and resentment?

Still, I think Dunn too airily waves away the many strong reasons for tipping to persist. For instance, consider her assertion that “Better service doesn’t actually beget better tips.” To back this up, she claims that “perceived service quality only accounts for two percent of the variation between tips” and links to a study by Michael Lynn that includes that figure.

But what does “two percent of the variation” actually mean? You need to go to an earlier paper by the same author, “Restaurant Tips and Service Quality: A Tenuous Relationship“, to see the original statistical analysis, which is beyond me. But here Lynn provides some concrete examples, like:

Increasing service ratings from 3 to 5 (on a 5 point scale) raises the median tip by less than 3 percent of the bill in all four of the studies where sample sizes make this comparison meaningful and by less than 1 percent of the bill in two of those studies.

So according to these four studies, on a $100 bill, the server who gives 5-star service will be tipped roughly $1-3 more than the one who gives 3-star service. That’s not much, but it adds up over the length of a shift. The question is, how easy is it for a server to improve her service ratings from 3-star to 5-star?

Lynn establishes that there’s little relationship between the size of the tip and the perceived quality of the service:

Consumers who rated the service as excellent sometimes left tips of 0 to 5 percent, so small tips do not always mean that the tipper was dissatisfied with service. Furthermore, consumers who rated the service as poor sometimes left tips of 20-25 percent, so not everyone who is dissatisfied with the service leaves small tips. In general, the weak relationship between tips and service evaluations means that tips are a poor indicator of customer satisfaction or dissatisfaction.

That’s true. But he doesn’t address whether some servers are consistently able to receive 5-star service ratings and therefore consistently claim that 1-3 percentage point 5-star service premium.

I know from my own behaviour that even in cases of exceptionally bad service I’m loath to tip below 15%, and that unless extremely inebriated I’m unlikely to tip as high as 25%. Most of the variation in my tipping comes from two factors. One, I’m not that good at doing math in my head, so every once in a while I accidentally give a $3.50 tip on a $35 bill. Two, when I pay with cash, the tip may depend on what bills I happen to have in my wallet. When splitting a bill with friends, I’d be annoyed to think that our server’s performance was being evaluated on the basis that when it came time to settle the bill, none of us had a five.

That doesn’t mean our interactions with the server have no impact on whether or not we ask her to break a twenty. I think Dunn is confused by Lynn’s custom of differentiating between what he calls “service quality” and “server behaviours”. The latter are the little tricks servers use to elicit more generous tips, like introducing yourself to the customer by name, making physical contact, and drawing a smiley face on the cheque (which apparently works for female servers only). I think I understand what Lynn intends by separating these two categories – “service quality” would include the server’s ability to accurately describe menu items and give directions to the bathroom, while “server behaviours” would be stuff like smiling, joking, flirting, and being nice. In other words, “service quality”, as defined by Lynn, excludes a whole lot of things that you or I would consider fundamental to good service.

Now as we’ve seen, “service quality” doesn’t make much difference to the size of the tip. “Server behaviours“, however, make a huge difference. For instance, introducing yourself by name adds 8 percentage points to a tip. Touching a customer’s arm adds 5 percentage points. So when Dunn says “Better service doesn’t actually beget better tips”, all she means is that memorizing the specials won’t help a server pay her student loans. Flirting, smiling, and joking will.

And it seems that – at least if you’re white – “service quality” does make a significant difference. In another study co-authored by Lynn, we learn that at one southern restaurant over one lunch hour, white servers who received a perfect 5/5 rating on “service quality” enjoyed tips 6.6 percentage points higher than those who received a less-than-perfect rating. Black servers, meanwhile, were tipped a few fractions of a percentage point below the less-than-perfect white servers even when they had perfect ratings. Lynn et al interpret this to mean that the black servers were tipped more poorly than their perfect service warranted. (They don’t address the possibility that it was the low tips, and not the high ratings, which accurately measured the customers’ true opinion of the black servers’ performance.)

Despite the small sample size of this study, the authors consider the results robust enough to draw the conclusion that the “disparate impact” of tipping on black employees could constitute a violation of the Civil Rights Act. If so, perhaps the results are robust enough for us to observe that, for a white server, the difference between 5-star and non-5-star service could be almost $7 on a $100 bill.

What would happen if restaurants got rid of tips – say if they followed Dunn’s suggestion that a service charge be “rolled into the cost of the meal”?

Obviously the restaurant that makes this change needs its service charge to be high enough for its servers to earn as much as they were making before in tips, but not so high that customers wind up paying more than the 15-20 percent they would otherwise have tipped. In other words, the service charge needs to be set at about the level of the average tip.

One thing you’d expect to happen over time is that the more successful servers, the ones who were able to reliably attract bigger than average tips – the ones who combine proficiency with a personable manner, the good-looking ones, the ones who know how to flirt, and, disproportionately, the white ones – will all migrate to restaurants where tipping is still allowed, where they can use their tip-eliciting expertise to make more money. The ones left behind will have little motivation to strive to provide 5-star service, let alone to offer those little extras, like calling us by our first name or drawing smiley faces on our bill, that we evidently find so tickling. The restaurant with its enforced average tip will wind up with a below-average staff.

This will tend to mitigate any spontaneous shift in the restaurant industry toward service charges. It would require a really committed social movement to get restaurant managers to overcome their sensible fear that service charges alienate customers and lead to indifferent employees. Frankly I don’t think anyone’s passionate enough to get such a movement started.

So what about Dunn’s idea that tipping should just be banned by law? Well, as she admits,

Between 1909 and 1915 six states passed anti-tipping laws, all of which were repealed by the mid-1920’s as unenforceable or potentially unconstitutional.

There’s no reason to suppose a similar law would be any less unenforceable (or any more constitutional) now than it was then. Tipping may just be one of those pernicious practices we have to live with. Sorry, servers, you must continue at the mercy of our terrible math skills.

PS. I previously discussed Orwell’s comments about tipping in Homage to Catalonia.