Beliefs of Childcare Center Managers about Parental Characteristics and Admission Policies – Experimental Evidence from Germany

We send out an online questionnaire comprised of two parts. First, we evaluate the context of admission decisions of childcare center managers (e.g., demand relative to available slots at the childcare institution, stakeholders in the allocation process etc.). Second, we investigate the underlying beliefs of childcare center managers that might lead to discriminatory behavior in the admission process using a survey experiment.
The questionnaire will be sent out to childcare centers all over Germany. Each targeted childcare institution will receive an invitation email with a personalized link to the survey. The email asks childcare center managers to participate in the survey. In the survey, respondents will first be asked general questions about the childcare center (e.g., how many children they cater, how they communicate with parents, etc.). Then, respondents will be shown two fictitious email inquiries by parents who are interested in a childcare slot in the center. The emails differ in whether the fictitious parent has a name indicating native vs. migration status, and whether the stated occupations of both parents indicate a higher vs. lower occupational status. Our study design is a factorial 2x2 design (four treatment cells) in which we independently vary (i) migration background and (ii) lower/higher occupational status. Upon reading each email, respondents will be asked to answer a set of questions to elicit their beliefs about the fictitious parents and their child.
The sample is drawn from a dataset comprising contact details for nearly the universe of childcare centers in Germany. We will send out invitation emails to a randomly selected subset of these childcare centers, targeting childcare centers hosting children below the age of three, and explicitly ask the managers of the centers to participate in our (incentivized) survey. Data collection will be operationalized in two waves. First, we will send out three different types of invitation emails (1. no incentives, 2. conditional incentives of EUR 5, and 3. conditional incentive of EUR 10) to a randomly chosen subsets of childcare centers (2000 (gross) recipients per type of incentives, stratified on the federal-state-level). Upon learning what incentive scheme realizes the most advantageous response rate, we will send out additional email invitations using the preferred incentive scheme until the envisaged effective sample size of 1,020 (net) responding childcare center managers is reached.
As a follow-up study to this survey experiment, we plan to conduct an audit study testing variations in individual background characteristics for successful applications for childcare slots. This follow-up study will be informed by the results of the survey experiment at hand.

2020-09-01

2020-10-01

• Childcare center managers’ beliefs about parental characteristics (measured on a five-point Likert scale)
• Childcare center managers’ beliefs about child characteristics (measured on a five-point Likert scale)
• Stated probability that the manager would offer the family a childcare slot (measured as stated probability)

Heterogeneity by location, size, (excess) demand for slots, and organizational structure of the childcare centers

We implement a questionnaire which is structured in two parts. The first part is a standard survey eliciting details about the childcare center (e.g., the number of available slots, communication with parents). The second part is a survey experiment that follows a factorial survey design:
• Respondents are asked to imagine that they receive an email request from parents.
• Then the childcare center managers are shown two randomly drawn fictitious email inquiries by parents. The emails contain information about the migration background of the parents by including either a German or Turkish/Arabic sounding name (e.g., Christian Müller and Haydar Kaya) and by providing information about the occupation of the parents (e.g., jobs indicating higher education background, such as doctor, and jobs indicating lower education background, such as bus driver).
• Each of the fictitious emails is followed by a set of questions to elicit respondents’ beliefs about the child (e.g., agreement to the statement “I think that the child has high cognitive abilities”) and about the parents (e.g., agreement to the statement “I think that the parents are reliable”). Furthermore, we elicit the stated probability with which the manager would offer the family a childcare slot.

• Stratified randomization (childcare centers are randomly assigned to treatment and control within federal states in Germany)
• Randomization is carried out by the survey program “Qualtrics”

at the individual level

No

1,020 childcare center managers

255 childcare center managers per treatment cell; 1,020 in total

255 childcare center managers per treatment cell; 1,020 in total

According to the powerBBK Stata package from Bellemare et al. (2016), a sample size of 1020 participants (255 per treatment cell) allows us to detect treatment effects of 0.13 standard deviations with 80% power at the 5% significance level.