The Reimagined Study — Summary

Dr. Kavitha Subramanian studies maternal nutrition in Telangana’s tribal districts. She has run three studies over seven years. Each was well-designed by conventional standards. Each produced clean findings. Each finding, when implemented, worked less well than the study predicted.

Her third study, a cluster-randomized trial of micronutrient supplementation for pregnant women, improved hemoglobin levels by a statistically significant margin. The intervention was recommended for scale-up. Kavitha wrote the recommendation herself. She also knew what the study could not find: adherence dropped 40% during planting season because the women were the primary agricultural laborers. The hemoglobin improvement did not reduce preterm births because the mechanisms producing preterm births in this population are compound, not primarily nutritional. The study found what it was designed to find. The mechanisms that determined whether the finding actually mattered operated at a stratum the study was not designed to reach.

Retroduction approaches the same question from the other direction. Start from the outcome, not the intervention. Map the compound: nutrition interacting with labor pattern interacting with water access interacting with household structure. Identify the stratum gap: where does the compound produce outcomes the published studies cannot explain? Reason backward from the gap to the mechanism. Apply the skeptic operations before the study begins: Is “pregnant woman” the right unit of analysis, or is it the household whose labor arrangement determines what her body endures?

The essay extends the method to education in Madhya Pradesh and agriculture in Maharashtra, showing in each case that the conventional study measures what it was designed to measure while the mechanisms that determine whether the finding matters operate elsewhere.

The honest position on RCTs: not wrong, not replaceable, not sufficient. The right tool for isolable mechanisms. The wrong tool for compound causation. The research enterprise has one tool for every job. Some jobs need a different tool.

Kavitha has started designing her fourth study. The unit of analysis is the household’s compound load across the agricultural cycle. The outcome is not hemoglobin but whether the pregnancy unfolds inside conditions the body can sustain. She does not know if the journal will accept it. She is submitting it anyway. The old maps on her shelf are beautiful and wrong. She would rather draw a new one, imperfect but pointed at reality, than keep refining the old one until its precision is flawless and its relationship to reality is nil.