Norms and Behaviors around Son Preference in India

The study is cross-sectional and does not primarily aim to evaluate a certain intervention. However, we do randomize certain components of the questionnaire in order to contain rigorous evidence on certain behaviors and beliefs. The following randomizations are applied and further specified below:

(1) Double list experiment to measure the extent of preferential treatment of sons vs. daughters by parents (school choice)
(2) Randomization of the order of the vignettes to measure preferential treatment of sons vs. daughters by parents in a behavioral game
(3) Randomization of whether asking about the son or the daughter in question about a hypothetical child behaving socially undesirable
(4) Randomization of the order of the vignettes to measure perceptions of certain social norms
(5) Randomization of the order when asking about perceived beliefs of the women vs. the men in the community

2024-08-28

2024-11-30

The outcomes are intended to measure son preference of mothers and fathers in our samples, elicited in several different ways by employing a combination of incentivized choices, vignettes and simple survey questions.

Outcome 1: Decisions in behavioral game (Likelihood of choosing the boy or the girl child in several consecutive hypothetical scenarios). Scale from 1-4

Outcome 2: Proportion of true responses in double list experiment (son preference in educational investments). Scale: Proportion in control vs. treatment group

Outcome 3: Community-conditional son preference (reaction to hypothetical son's/daughter's socially undesirable behavior conditional on community knowing). Scale: 1-4

Outcome 4: Health and educational status of children (birthorder, number of siblings, school choice and fees)

We will employ the different outcome variables to obtain a comprehensive understanding of the nature (degree and type) of son preference within the sample population. We will utilize principal component and/or factor analyses in order to learn about whether there are different types of son preference (e.g., before/after birth, education, health) and how they are related to each other. This approach allows us to construct the outcome variables correspondingly, reflecting the relevance and nature of the underlying types of son preference.

The study is cross-sectional and primarily aims to rigorously quantify, on the one hand, perceived and actual norms regarding son preference (constructed from individuals' own attitudes and their beliefs about others' attitudes towards norms) and, on the other hand, the degree of son preference in the sample population. The study therefore does not evaluate any intervention, but selectively uses randomization methods to obtain rigorous measures of these concepts, for instance through list experiments. Primarily, we aim to (i) quantify the prevalence of different types of son preference in the sample population, (ii) quantify the degree of misperception regarding norms about son preference and general gender norms at the household- and community-level, and (iii) analyze the relationship between norm perceptions and son preference.

Perceptions (and the potential degree of misperception) of norms about son preference are measured as follows. As part of the survey, participants are presented with five different norms about son preference in particular and about patriarchal norms in general (Wife visiting/caring for natal family, Moving because of the wife's job, Desired fertility, Sons' vs daughters' right/resources to health and education, Dowry payments). These norms are either presented as vignettes or statements. Participants are first asked about what they themselves, without anyone else knowing, think about the behavior in the statement or vignette. Subsequently, they are asked to guess what their spouse might think about the behavior. Both of these questions are measured with a 5-point likert scale (1=Do not agree at all; 5=Fully agree). Finally, in a separate section of the questionnaire, we also ask participants for each of these norms, how many out of 10 women/men (order of asking for women/men first is randomly assigned) in their community would agree with the behavior. Participants’ answers to the question about their perceptions of the community's agreement with certain behaviors are incentivized: After the study, 10% among the participants, who correctly guessed the proportion of women/men in the sample, who would agree with a certain behavior, will receive an in-kind gift worth 125 Rps.

By asking about participants' own agreement with/attitudes of a certain behavior or statements and about their guess of their spouse's and the community's agreement with that same behavior or statement, we can quantify whether and how certain norms are misperceived in our sample population, both within the household and within the community.

Most randomizations in the study are done by a computer (survey CTO software or Stata). The randomization of the order of the scenarios in the game to measure son preference in different hypothetical situations is done through letting participants choose game cards. The surveyor shuffles the cards, then holds them in front of the participant (with the participant only seeing the back of the card) and asks her/him to choose a card. The surveyor repeats this procedure until all cards (scenarios) have been randomly chosen. This randomization is done by randomly drawing cards without putting them back in the pile after the scenario has already been played.

Individual-level (double-list experiment)

No

300 households (interviews will be conducted with husband and wife of each household)

600 (interviews with husbands and wives)

Double list experiment: All 600 participants receive both a treatment and a control list, but it is randomized at the individual level which one is which (i.e., to which of the two lists the target statement is added). Thus, in 300 participants the first list is the treatment list and in 300 participants in the second list is the treatment list.

N.A.