|
Field
Intervention (Public)
|
Before
The primary intervention is a business directory with a description and contact information for approximately 10,000 businesses in 6 districts of the Kagera region. The directory takes two forms: a printed paper directory similar to a traditional print Yellow Pages, and a digital directory that is available via both USSD and smartphone app. Households are assigned to control, paper directory, or digital directory.
|
After
The primary intervention is a business directory with a description and contact information for approximately 10,000 businesses in 6 districts of the Kagera region. The directory takes two forms: a printed paper directory similar to a traditional print Yellow Pages, and a digital directory that is available via both USSD and smartphone app. Households are assigned to control, paper directory, or digital directory.
******* (below added in September 2026)
In the endline survey, which we conducted in 2024, we embedded a discrete choice experiment (DCE) on parental preferences over children's occupations. The DCE is separate from the trial's main intervention and outcomes, and the directory treatment arms play no role in identification. It was designed and fielded before this amendment was submitted and was not pre-registered. This entry is filed retrospectively to document the design and the planned analysis as implemented.
Design. Respondents were the heads of farming households in the trial's 120 villages across five districts of Kagera region, Tanzania (Bukoba Rural, Karagwe, Kyerwa, Missenyi and Muleba). The sample was the 3,247 endline respondents. Each respondent was asked to think of a boy or girl under 10 in their life, such as a son or daughter or a niece or nephew. They then answered three questions of the form "Would you rather that this child grow up to be [A] or [B]?" Each job profile varied on three attributes: job type (farmer; salaried worker in a physically demanding job; salaried office worker; entrepreneur with a small trading business), average annual earnings (1.2, 2.4, 3.6 or 6.0 million TSH), and earnings variability (variable year to year, or protected by insurance). This gives 32 possible profiles.
Randomization. The child's gender was randomized with equal probability by the survey software at the moment the module began. Respondents were randomly assigned to one of 16 question blocks. Block assignments were determined in advance to ensure balance across enumerators and times of day, and were concealed from both enumerator and respondent until the module began. The number of respondents per block ranged from 198 to 208 respondents.
Outcomes and analysis. The outcome is the binary choice between the two profiles in each question. Choices are analyzed with a conditional logit model, with standard errors clustered at the respondent level. The main specification includes job type indicators (office worker omitted), an insurance indicator, and log earnings. A robustness check replaces log earnings with indicators for earnings levels. We compute the implied earnings multiple required for a respondent to choose farming over each alternative occupation, holding insurance status fixed, with confidence intervals from the delta method. We also estimate heterogeneity by respondent gender, child gender, above-median acreage planted, above-median farm revenue in the most recent season, and above-median household size.
|