A Rose is a Rose is a Preference Signal


A rose by any other name is just as sweet, isn’t it? Even virtual roses used in Korean online dating experiments. In a new working paper by main author Soohyung Lee of the University of Maryland, economists studied the impact on preference signaling – signals sent to a select few.

In the study, a major online dating company in Korea organized dating events with 613 participants, half men and half women. Everyone was given two free “virtual roses” that they could attach to an e-mail to a fellow participant, and a few were given 8 virtual roses. Although these roses cost nothing, attaching a rose to an e-mail drastically increased rates of acceptance, even among different “desirability” groups.

Here’s the abstract:

The large literature on costly signaling and the somewhat scant literature on preference signaling had varying success in showing the effectiveness of signals. We use a field experiment to show that even when everyone can send a signal, signals are free and the only costs are opportunity costs, sending a signal increases the chances of success. In an online dating experiment, participants can attach “virtual roses” to a proposal to signal special interest in another participant. We find that attaching a rose to an offer substantially increases the chance of acceptance. This effect is driven by an increase in the acceptance rate when the offer is made to a participant who is less desirable than the proposer. Furthermore, participants endowed with more roses have more of their offers accepted than their counterparts.

So how big was the virtual rose effect?

We find that, overall, sending a proposal with a rose increased the probability that a recipient would accept the proposal by 3.3 percentage points, which corresponds to a twenty percent increase in the acceptance rate. This effect is similar in magnitude to the increase in the acceptance rate by recipients when the dating offer came from a middle rather than bottom desirable sender. This implies that, by sending a rose, a bottom group sender will be almost equally attractive as his or her counterpart belonging to the middle group.



[E]ven when everyone can send a signal, signals are free and the only costs are opportunity costs, sending a signal increases the chances of success.

“Only costs are opportunity costs”? What other kinds of costs are there?

Maybe I need a primer on the “cost signaling”/”preference signaling” distinction.


An example of a non-opportunity cost would be if it cost $1 per rose but you could send as many as you wanted or could afford. The opportunity cost is that if I send the rose to Mary and Anne, I can't send one to Karen. And there is no insentive or value in not sending both out.
Not to say that a non-opportunity cost can't also be an opportunity cost, in that you could only have $2.

The rose does symbolize a special interest in this person, as in most on-line dating there is nothing stopping from e-mailing to all 315.5? women. And most women would assume I am writing to many, but she who gets the rose is one of my top 2 choices.


How is spending $1 for a rose not an opportunity cost? Every time I would spend a dollar for a rose, I would sacrifice my opportunity to spend that dollar on something else.

I guess the idea is that the study isolated the rose economy from the dollar economy. And, yeah, I guess that makes a difference.


"In the study, a major online dating company in Korea organized dating events with 613 participants, half men and half women."

Did anybody else do a double take on that one?


Yeah, if you don't here in Korea then you might not be aware of how active the half-man, half-woman community is. But just check out the Facebook Group page, and you'll find all 613 of us.

And we're always looking for new members. ??? ?? ?!


In this situation, though, the opportunity costs are the most important. "He sent the rose to ME when he could have sent it to all these other people!" If the roses cost money, but you could send as many as you wanted, I think the signaling effect would be smaller.


This makes sense. If they are only given 2 virtual roses, they will use them for the people they are most interested in. That means that if someone gets a message with a rose on it, they can assume that the sender is very interested, giving them a better chance of getting a date or whatever they are looking for. It eliminates the risk for the person receiving the message. A message with rose attached basically signals that the sender would be a 'sure thing.' The limited supply of roses for each person makes roses valuable on the site, even though they don't exist and aren't actually worth anything.


no costs.... except oppertunity costs? That still seems like a cost, but okay.

I would be more interested in the answer to the question: do people who have more roses show less of an effect when they are given a rose? I'd expect yes.


So what? I wonder why economists spend so many hours and our expensive tuitions for this kind of silly issue. This study is just a analysis for analysis, not critical matters in our real life (survival and taking care of our family) Economists have to study for real economic issues which can improve social welfare.


Economics studies the human behaviour, using "sent roses" or using "collateral offers" reflects the same phenomena, signaling.
In economic theory economists use many assumptions, lots of them are unobservables (and not tested), using this kind of experiments the idea is try to estimate the unobservable's effect as a complement for the theory economists use to take decision for those things that "really matters"... (and are also much more funny than IS-LM)