Digging Deeper: Evaluating the Impact of Customized Fertilizer Recommendations on Smallholder Farmer Investment, Yields, and Learning in Tanzania

How to cite

Michelson, Hope; Janzen, Sarah; Magomba, Christopher; Maertens, Annemie; Fairbairn, Anna. 2025. "Digging Deeper: Evaluating the Impact of Customized Fertilizer Recommendations on Smallholder Farmer Investment, Yields, and Learning in Tanzania." AEA RCT Registry. AEARCTR-0017252. https://doi.org/10.1257/rct.17252-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Hope Michelson* , Sarah Janzen , Christopher Magomba , Annemie Maertens , Anna Fairbairn

* Primary investigator

Registry ID
AEARCTR-0017252
DOI
https://doi.org/10.1257/rct.17252-1.0
First registered
Last updated
Status
In development
Trial dates
2026-01-05 – 2029-01-01
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/17252

Abstract

Soil nutrient limitations vary widely across plots and are often unobservable to farmers, yet they are crucial for the effective use of fertilizers, hybrid seed, and irrigation. This study evaluates whether rapid, affordable soil testing can improve farmer decisions and outcomes through better information. In collaboration with OCP Tanzania, we will implement a cluster randomized controlled trial across 270 villages during the 2025–2026 agricultural season to assess the effects of providing soil test results at the plot and village level. In all treatment villages, both randomly selected and lead farmers will have their plots tested. While tested farmers receive their own results, untested farmers will receive either the national fertilizer recommendations or aggregated village-level results. This design allows us to estimate the direct effects of receiving a soil test, as well as spillover effects on untested farmers. We further analyze heterogeneity to inform targeting strategies and explore whether converting individual-level soil information into village-level recommendations generates broader public value. Geo-referenced soil test data will be used to characterize local variability and improve future recommendation systems.

Experimental design

The proposed study is a clustered randomized control trial and takes place in 270 villages in the Njombe, Singida, and Songwe regions of Tanzania. Of these, 180 villages will be randomly assigned to treatment, and 90 villages will be randomly assigned to a pure control. In each treatment village, 15 lead farmers (as determined by OCP in consultation with village leaders) and 10 randomly selected farmers from the general population will receive a free soil test. For plots receiving these tests, the test results will be shared directly with individual farmers, including plot-specific input recommendations based on these soil test results.

Design details

Randomization method
randomization done in office by a computer
Randomization unit
Treatment status: village; lead and non-lead survey respondents: individual
Clusters
270 total villages
Observations
In each village, we will survey a stratified random sample of 12 farmers. These samples will be stratified by lead farmers (4), and non-lead farmers (8). In treatment villages, all lead farmers in our sample will receive a free soil test. Among the non-lead farmers in the same villages, half will be assigned to receive soil tests, while the other half (“spillover farmers”) will not receive soil tests for any of their plots. Our total sample size is 3,126 smallholder farming households.
Arms
180 villages will be randomly assigned to one of the two treatment arms (individual recommendations only and individual plus village-level aggregation recommendations), and 90 villages will be randomly assigned to a pure control
Primary outcomes
farmer beliefs, crop yields, farmer investment, farmer learning, spillovers, profitability

Cited by

No citation data available for this registration yet.

Keywords