RSVPs

First published: ; Last modified:

You’ve probably gone to a few events in your life, and declined or ignored invitations to others. Maybe you occasionally cancel last-minute (or even no-show without notice). What is it that informs these choices?

Your schedule and expectations about the event contribute, of course, but there are other cognitive factors at work too. One of my goals over the last year has been to organise more gatherings and get-togethers, so I’ve spent a fair bit of time mulling over what these further factors might be, and how they all interact. I think my tentative answers contain an interesting mix of economics and psychology, though there is a lot of unfounded speculation which could be shored up by some probability theory and/or survey data (so get in touch if you could help!).

I want to start by setting up a distinction between two different types of events. Although there’s always a person or team of people in the background, sometimes it feels like an event is being run by an entity separate from any of the individuals organising it. Conferences, concerts, and charity fundraisers are all examples of what I call institutional events. On the other hand, some events are individual-centric, like birthdays, weddings, and baptisms. In between these two extremes we have events which are perceived as institutional by some and individual by others – for instance, a dinner party that’s open to all, or the Matchmaking Society mixers I’ve put on. I think that invitees respond differently depending on which category they view an event as being in, for reasons we’ll look at further down when I come on to talk about personal parties. First, let’s consider the dynamics of attendee decision-making at institutional events.

Pricing

Imagine you’re an event organiser whose aim is to to maximise attendance. One approach you might adopt is be to keep ticket prices as low as possible, or even pay people to come (perhaps indirectly, with free alcohol or suchlike). This is what a lot of student societies at Oxford do when hosting external speakers for talks, but I don’t think it’s always the most effective strategy, because people’s choices in this area are not classically economically rational.

The most obviously relevant cognitive bias is the sunk cost fallacy, when people continue pursuing a course of action they know is suboptimal simply because they’ve invested in it already. If you as an organiser care a lot about eliminating no-shows, because they put all your plans into disarray, then charging a small registration fee would be a sensible way to make it more likely that RSVPed guests actually turn up.1 But raising ticket prices might put people off signing up in the first place, so would it really increase attendance overall? Clearly in a situation where potential attendees are extremely averse to « wasting » sunk costs and sufficiently insensitive to price when deciding whether to buy a ticket, charging a fee would boost turnout, but there are more interesting scenarios we can explore.

One such set-up draws on the possibility that going to events is not an “ordinary” good for which demand decreases as the price rises. Because it’s hard to work out your valuation of an experience you’ve never had before (like going to a certain talk), you might look for external information to decide how much you’re willing to pay for it. If you noticed lots of other people buying tickets for an event, that could lead you to conclude it’s worth more than you initially thought, since it appears very popular.2 The price itself can also be a signal of quality: when looking to hire a lawyer, for example, you might be suspicious about somebody whose quote is 10x less than the next-cheapest option, for example.3 It’s therefore possible that raising ticket prices might lead to more people wanting to attend the event overall, in addition to reducing the number of no-shows.

Often the organiser isn’t single-mindedly trying to maximise attendance, though: larger events are naturally less intimate than small, exclusive ones, so there’s a trade-off to be made. I really like the comparison to plant species arriving at the optimal seed:pollen ratio as a way of thinking about this – if you’re trying to use an event for marketing, you’ve got to decide whether it’s better to aim for a high quality experience or a large quantity of attendees, and sometimes you’ll want to lean towards the former.

Charging a fee can help achieve this aim of increasing the average engagement of attendees. Since someone registering must attach a value to the event at least as high as the ticket price, individuals who weren’t that enthusiastic will be filtered out. I think the selection effect of even a nominal ticket price is more pronounced than you might expect, because psychologically there’s a discontinuity between an event being free and it costing some amount of money. Partly there’s the trivial inconvenience of having to get out your debit card and pay, but also the purchase of a ticket is a prompt to think about whether you actually want to go or not. There are opportunity costs from going to an event, but those are very easy to forget; parting with real money is harder to ignore.4

Another potential consequence of charging a nominal fee is an increase in demand for your event. If the value one attendee gets out of the experience is a function of the average enthusiasm of other attendees (e.g. as might be the case at networking drinks, speed dating, and live concerts), raising the price really does make the event more valuable – you’d rather be in the company of a more exclusive group of highly-engaged peers than in a crowd of people who’re only lukewarm about being there. Price isn’t merely a signal of quality, it’s a contributor to it.

Selection effects also explain why charging for tickets reduces no-shows beyond the psychology of sunk costs alone: it’s less likely that a small shock to the costs of attending (e.g. on-the-night transport delays) or their expected utility from the event (e.g. they just get cold feet) will change somebody’s mind about coming. Introducing a fee means that the average valuation of the event from those who have signed up is higher, so the probability that some random disruption tips the event from being net-positive to net-negative for an attendee is smaller.

Parties

I always appreciate it when someone else puts on a dinner where there are nice, interesting people to get to know, and I like running events, so one of my goals over the last year has been to host more things. I organised a birthday party for myself back in January, and a summer get-together for friends more recently. Although I wanted people to RSVP yes, be excited about the get-together, and not no-show, obviously I wasn’t going to charge people to come.

Underlying this is the fact that there’s a fundamentally different relationship between the host and guests for a party than for an institutional event. When you’re attending a conference or lecture, generally the reason for going is that you think the event will be of direct benefit to you. Your decision about whether to attend won’t usually have any bearing on how close you are with the organisers, since they probably don’t even know you (unless you’re a top-flight conference speaker, or something). On the other hand, when an individual is organising an event, part of what motivates your attendance is your relationship with them – it’s a way of showing that you care. If you’re not convinced, then consider the following scenario:

Two friends you’re equally close to invite you at exactly the same time to a party they’re hosting on the same day. (You can assume they’re in separate social circles and you’re the only person invited to both.) One party is just for the sake of getting people together, whilst the other is to celebrate a milestone birthday. You expect that both events will be equally enjoyable for you, and there’s no difference in the amount of travel time required. It’s not possible to attend both, so you need to decide how to RSVP.

I’m almost certain that you’d choose the birthday party, and probably even tell the other friend about the conflicting engagement when declining their invitation, in the expectation that they’ll understand why it takes precedence. But this is very clear evidence that your decision to attend a personal party is influenced by considerations about the organiser. The reason why you’d be willing to travel a long way for a wedding when you might for an equally lavish birthday is similar – you’re partly attending for the host’s benefit.

For me, this made organising parties more stressful. When I invited people and they weren’t able to come, I viewed that as a partial reflection of my relationship with them: « If we were closer friends then maybe they would’ve managed to make it ». On the other hand, for those who’d replied yes, I worried that when they saw the final guest list near the time, they might judge that they wouldn’t enjoy the party (because it’s too small, or not enough of their other friends are coming, or whatever) yet decide to still come along as a “favour” to me. But I didn’t want that at all, because (as you can tell) I have a tendency to feel very responsible for the enjoyment of others when I’ve organised or suggested an event they’re at, and so would much prefer that people simply don't come if they'd rather not. I’m still learning how to host well, so I’m hoping that with time I’ll become more confident that I can make an event good no matter how many people are there, but also just get better at not fretting excessively about other people’s entertainment. Then I can have more fun myself!

Appendix: a probability puzzle

I don’t know the answer, or even if it’s a well-constructed problem, but it’s my attempt to formalise some of the speculation in the “pricing” section.

There’s a group of people considering buying a ticket to an event. Each individual $i$ has excitement $X_i \gt 0$ about the event, sampled iid from some distribution $X \sim \mathcal D$ where the specifics of $\mathcal D$ are not known.

An individual will pay $p$ for a ticket for the event iff $X_i \ge p$, but buying a ticket doesn’t necessarily mean that they’ll attend – people could cancel instead (without a refund). Because people are “irrational” and take into account the price they’ve paid for the ticket when deciding whether to cancel, the chance of cancelling $c(p)$ is some decreasing function of $p$.

(We could assume for simplicity that $c(p)$ is independent of an individual’s excitement, though a more realistic model might do something like including randomly-generated unexpected inconvenience $u_i$ for each individual, and then saying that people cancel iff $u_i - s(p) \gt X_i$ where $s(p)$ represents the sunk cost aversion and is increasing in $p$.)

The organiser is considering raising ticket prices in the hope that this leads to more people actually attending the event, even though fewer people will buy tickets.

a) If raising ticket prices really would lead to more people attending the event, what can we infer (if anything) about the distribution of $X$ and the shape of $c$? Similarly, what would $X$ and $c$ need to look like for raising ticket prices to lead to fewer attendees?

Suppose instead of maximising attendance, the organiser wants to maximise the value $V$ produced by the event, where value produced is the sum over each attendee’s enjoyment $E_i$.

One approach would be to treat $E_i$ as simply being equal to $X_i$ for each individual. However, it might be that everyone prefers events where the other attendees are enthusiastic, and so $E_i$ is equal to the average excitement of attendees (perhaps plus some term related to $X_i$ so it isn’t the same for everybody, but I don’t think this is that important).

b) In these two new cases, can we say anything about whether the organiser ought to raise ticket prices (maybe given some facts about $X$ and $c$)?

Some thoughts:

  1. One of Eventbrite’s online guides recommends this trick, in fact. ↩︎

  2. This is known as the “bandwagon effect”. See also this paper summarising the snob and Veblen effects, two other quirks of demand. ↩︎

  3. I spent a while trying to review the literature on prices as signals of quality – it’s interesting but also seemed quite fragmented and sometimes involved very technical mathematical models I couldn’t understand.

    There were a few early papers I came across that had helpful intuitive explanations: Nelson (1970) introduces a distinction between goods where consumers can get information via search (i.e. inspecting products before purchase) and those where you must get information via experience (i.e. using products after purchase); Wolinsky (1983) discusses how (given certain conditions, including that consumers can obtain some information about the quality of goods from their shopping process) there may be a “separating equilibrium” where prices unambiguously signal quality and no producer cheats by supplying lower quality than their price suggests (because foregone revenue of losing the custom of the informed consumers would outweigh the lower production costs); Milgrom and Roberts (1986) build on Nelson’s work to look at how the possibility of repeat purchases based on a good reputation can incentivise high quality firms to spend more on advertising for experience goods and also to sell their products more cheaply than low quality competitors, because an initial purchase is more valuable to them than it is to the low quality producers; Farrel (1981) has a long PhD thesis which explores the issues of advertising, cheating, and introductory offers, though I have only read the (pretty accessibly-written) introduction; Bagwell and Riordan (1991) argue that for durable goods (where the purchase frequency is much lower), high quality goods will start out expensive to signal that, but then become cheaper as the public becomes more informed about them (meaning that the price signal can be weakened), backing this up with some data about consumer purchases.

    More recently, Rroshi and Weichselbaumer (2021) analyse German shopping data and find that price is positively correlated with quality for durable goods but negatively correlated for non-durables, in line with the predictions above. (One thing I’m confused by is that for non-durables, where there is a high-quality mass market product as described in Milgrom and Roberts, if consumers are aware of the incentives facing firms, surely they would just never buy the more expensive goods from low quality firms? Maybe the equilibria described rely on some consumers being unaware of the firm incentives? But that seems a bit strange, and I feel I’m just misunderstanding.) Mastrobuoni, Peracchi and Tetenov (2014) looked at some experimental results in the Journal of Wine Economics(!) and found that price does have a positive signalling effect, particularly when moving from the cheap (€3) to medium (€5) price point. For some contradictory data, Kalita, Jagpal and Lehmann (2004) say they found that price is positively correlated with quality with non-durable as well as durable goods (going against the idea that reputation & repeat purchases incentivise high quality producers to lower their prices for non-durables), but I’m not sure how strong their empirical or theoretical work is compared with that of the others mentioned here. ↩︎

  4. As an illustration, say I have two event options open to me in an evening, as well as simply staying at home. One event is free, the other costs £5. Valuing my time at a fairly low £20/hr, and supposing they both last two hours, the full costs would be £40 vs £45 – so, the expensive event needs to be just 12.5% better (in expectation) for me to “rationally” choose it. Things don’t really feel that way in practice. It’s cognitively difficult (and sometimes unhelpful) to take into account the value of time in decision-making, so we usually don’t, instead switching into cost-benefit analysis mode only when what we’re spending becomes explicit in the form of cash. ↩︎