Do families suffer from discrimination in the sharing economy?

We want to contact rental hosts with the following text in the control group (couples without kids):

"Hello,
I am looking for a place for my husband and me. Do you think your place is appropriate for us?"

Additionally, we need a control group for couples with one, two, or three friends. Thus, we will change the
text as follows for three, four, or five adults:

"Hello,
I am looking for a place for a (two, three) friend(s), my husband, and me. Do you think your
place is appropriate for us?"

Families will contact the hosts as follows:

"Hello,
I am looking for a place for our (two, three) child(ren), my husband and me. Do you think your
place is appropriate for us?"

Hosts have almost no information available about the people who will stay in their place. We are using
accounts that have not received any reviews. To signal playful but well-behaved kids that are engaging
in quiet activities, we want to modify the sentence as follows:

"Hello,
I am looking for a place for our (two, three) child(ren), my husband and me. All of us (“We” for
the couple) are really looking forward to visiting the Louvre (or the appropriate local main
museum). Do you think your place is appropriate for us?"

2023-07-04

2023-07-25

The response from the host is the main outcome variable. We expect that we will get either a rejection or an invitation.

The secondary outcome focuses on the fact if the additional information that signals well-behaved children changes the response rate of the hosts.

We want to contact vacation rental hosts asking if it is possible to rent an apartment. The control groups are a couple, a couple with one, two, or three friends. In the treatment group a couple has one, two, or
three kids. Additionally, we want to analyze if signaling the children’s ability increases the chances of receiving a response for
families.

We want to contact hosts in the five largest cities in the European Union.

We will use exclusively native-sounding names, as previous research shows that people with foreign-sounding names suffer from discrimination (see, e.g., Gomez-Gonzalez et al., 2021). Thus, we will construct five native-sounding female and five native-sounding male names for each country. We will confirm in local surveys that the names sound indeed native.

We will use the computer program Stata to randomize the selection of hosts. We will set a seed before the start of the experiment to make the randomization method reproducible.

We have five groups - each group is one city. In each city, a unit of randomization is one host.

No

10,000 rentals. 2,000 in each city.

10,000

We have around 1250 observations in the control group and 750 observations in the treatment group. 2000 observations for each city. The control group consists of couples (500 obs.), with one adult friend (250 obs.), two adult friends (250 obs.), or three adult friends (250 obs.). The treatment group consists of families with one child (250 obs.), two children (250 obs.), or three children (250 obs.).

With a probability of failing to reject the null hypothesis β = 0.1, 750 observations in each group (control group without the couples), a power analysis provides a minimum detectable difference of 0.085 (with a binary response rate). With a probability of failing to reject the null hypothesis β = 0.1, 500 observations in the control group and 750 observations in the treatment group (control group only with couples), a power analysis provides a minimum detectable difference of 0.095 (with a binary response rate).