Highlights
- In markets characterised by preference mismatch and low communication costs, a “less is more” approach can improve matching outcomes by preventing users from overly focusing on mismatched attributes and prematurely excluding suitable candidates.
- When the preferences of the two sides were aligned, high attractiveness scores predicted a higher likelihood of reply. Conversely, preference mismatch could lead to situations where users who were not the top choices on the other side were more likely to reply, revealing the mismatch’s impact.
- Users with lower attractiveness scores still received replies, indicating that mutual preference was not solely based on observable attributes.
The study investigated the influence of information on two-sided matching platforms where preference mismatch can be prevalent. Two-sided platforms or markets like Airbnb, Tinder, and others are characterised by different preferences held by each party involved, and a match occurs when both sides express mutual preference for each other. In practice, however, preference mismatch is common, complicating the matching process. The study uses a large dataset from an online dating platform in China to empirically examine how preference mismatches shape platform’s optimal information design and affect matching outcomes. The online dating platform lets users contact others with potential matches immediately after reviewing their short profiles, or send messages after browsing their longer profiles. Users can also choose to ignore a candidate, contact them immediately, or visit the long profile for more information. The researchers used this feature to analyse how the amount of information available about the other side influences matching outcomes.
Therefore, the data offers an ideal setting to observe natural user behaviour in response to different levels of information.
The platform does not provide personalised recommendations and displays the same candidate list to all users. Therefore, the data offers an ideal setting to observe natural user behaviour in response to different levels of information.
In total, the dataset included 1,198,943 clicks from 33,504 focal users, providing a comprehensive view of user interactions. To measure the degree of mutual interest, the researchers used the number of exchanged messages as a proxy for successful matching. They also constructed attractiveness scores for candidates based on attributes such as height, age, income, and education level. These scores were standardised to facilitate comparisons across candidates for each user.
The core concept that was examined was preference mismatch, exemplified by attribute differences such as height. For example, a typical man preferred a woman approximately 10 cm shorter, while a typical woman preferred a man who was about 20 cm taller. These differences illustrate the existence of preference mismatch, which influences proposal (i.e. contact) and reply behaviours. When preferences are mismatched, a user might propose to a candidate they perceive as ideal, but that candidate might not reciprocate interest, indicating a misalignment.
The study found that when the preferences of the two sides were aligned, high attractiveness scores predicted a higher likelihood of reply. Conversely, preference mismatch could lead to situations where users who were not the top choices on the other side were more likely to reply, revealing the mismatch’s impact. Notably, the data showed that users with lower attractiveness scores still received replies, indicating that mutual preference was not solely based on observable attributes.
This attribute-level preference mismatch suggests that more information about the other side can often exacerbate the mismatch.
A significant finding was that users tended to have different opinions about ideal match attributes, such as height, age, income, and education. This attribute-level preference mismatch suggests that more information about the other side can often exacerbate the mismatch.
The data demonstrated that the degree of preference mismatch was greater in complete information scenarios than in partial information scenarios. Specifically, more information about the other side tended to strengthen the preference mismatch, as it allowed users to focus on attributes that may not align with their true preferences or to overlook more suitable candidates.
Preference mismatch affected the matching process by causing users to select candidates who were less likely to accept their proposals. When users had complete information, they tended to rule out potential matches early, based solely on observable attributes, before considering unobservable qualities such as personality or hobbies.
As a result, more information could inadvertently reduce the likelihood of successful matches. The findings confirmed that providing more information does not necessarily improve matching outcomes.
In markets where preferences are mismatched and communication costs are low, providing less information may lead to better outcomes.
The research emphasises that the success of two-sided matching platforms critically depends on the role of information. In markets where preferences are mismatched and communication costs are low, providing less information may lead to better outcomes. The phenomenon dubbed the “less information is more” effect suggests that restricting available information can mitigate preference mismatch, allowing users to focus on a broader set of potential matches without prematurely dismissing candidates based on superficial attributes.
The data provided a controlled environment to observe these dynamics. Users could initiate contact based on short profiles containing basic information, such as nickname, age, education, and city, or visit the long profiles for greater details. Since user activity outside the platform was unobserved, the researchers used the intensity of mutual communication as a proxy for match success, measured by the number of mutual messages that were exchanged.
The researchers concluded that in markets characterised by preference mismatch, users are generally better off when obtaining less information about the other side at the initial stage.
The researchers concluded that in markets characterised by preference mismatch, users are generally better off when obtaining less information about the other side at the initial stage. This effect is driven by a preference mismatch, which causes users to prematurely exclude candidates based on superficial attribute differences.
Finally, this study suggests that there exists an optimal amount of information that one side should know about the other before initiating contact. Excessive information, especially in contexts where preferences do not align perfectly, can hinder the matching process by amplifying mismatch effects.
The insights from this study are particularly relevant for decentralised matching mechanisms with low communication costs, such as dating platforms, where user preferences significantly influence search and matching behaviours.
Preference mismatch, Matching platforms, Two-sided markets, Online dating, Information disclosure, Information design
* Learn more from the full research article here:
