Impacts of redesigned Soil Health Cards on farmer comprehension of fertiliser-related information

Indian agriculture faces a major problem of fertiliser misapplication. In many regions of the country, overall fertiliser application is extremely low, hurting agricultural productivity. In other regions, where overall fertiliser application is comparable to global trends, application is heavily skewed towards nitrogen-based fertilisers like urea, of which India is the second largest consumer in the world.
Fertiliser misapplication has led to reduction in long-term soil fertility and agricultural productivity for farmers. To solve this, the government of India launched the Soil Health Card Scheme, under which every farmer in India receives a customised set of crop-specific fertiliser recommendations that, if applied, would ensure the long-term productivity of their soil. These recommendations are printed onto cards, and distributed to farmers by agricultural extension workers, who are expected to explain the general purpose and recommendations of the card to farmers.
However, in a qualitative study conducted in 2018, we found that the card’s existing design is inscrutable to most farmers, inevitably leading to low adoption of the recommendations mentioned in it.
In this paper, we evaluate the impact of distributing redesigned, simplified versions of the card and supplementing in-person extension with mobile-based messages on farmers’ comprehension of fertiliser recommendations.

2019-04-29

2019-05-31

1. Comprehension of communication materials: I.e, the rate at which farmers' are able to understand what is written in their soil health cards.
2. Stated intention to adopt recommendations: i.e, the rate at which farmers' report their intention to adopt the fertiliser recommendations written on their Soil health Cards.

Our key outcome is farmers’ comprehension of fertiliser recommendations.

Two major components of the comprehension will be tested:

1. The ability to calculate the quantity of fertiliser required in their field.

We will assess whether farmers can accurately calculate the recommended quantity of fertiliser that is to be applied in their plot. A single crop-specific recommendation will be selected, and the farmer will be asked to calculate how much of that specific fertiliser they are expected to apply in their field.

1. The ability to distinguish between different columns in the recommendations table.

In their Soil Health Cards, farmers are given two recommendation options, each with a different combination of fertilisers. Farmers can apply the more easily available of these two sets.
Thus, we will assess whether farmers can correctly interpret what the multiple options on the card represent through a categorical variable. They will be asked why multiple quantities of the same fertiliser are mentioned for the same crop. If their stated reason is that the different quantities are due to the two different options that have been provided (with either of the two quantities being accurate), we will consider that they have understood the purpose of the multiple columns. If they provide any other reason, their response will be recorded in text, and recoded after data collection is complete.

The evaluation is a three-arm RCT. The first arm (C) is the control arm, where farmers will receive Soil Health Cards with the pre-existing card design. The second arm (T1) is the card redesign arm, where farmers will receive redesigned Soil Health Cards. The third arm (T2) is the combined arm, where farmers will receive both the redesigned card and ICT messages. Our experimental design, thus, allows us to isolate different elements of the intervention.
We are first randomly assigning villages into treatment and control, with all sampled farmers in control villages being assigned to the C group. In the treatment villages, we randomly assign 50% of the farmers into the T1 group and 50% of the farmers into the T2 group.

The evaluation is a three-arm RCT. The first arm (C) is the control arm, where farmers will receive Soil Health Cards with the pre-existing card design. The second arm (T1) is the card redesign arm, where farmers will receive redesigned Soil Health Cards. The third arm (T2) is the combined arm, where farmers will receive both the redesigned card and ICT messages. Our experimental design, thus, allows us to isolate different elements of the intervention.
We are first randomly assigning villages into treatment and control, with all sampled farmers in control villages being assigned to the C group. In the treatment villages, we randomly assign 50% of the farmers into the T1 group and 50% of the farmers into the T2 group.
C and T1 are randomised at the village level as this is granular enough to retain technical precision while also avoiding spill overs. Additionally, farmer comprehension may be impacted by peer effects (e.g. farmers discussing their soil health cards with each other), which can be mimicked by clustering the intervention.
T2 is randomised at the individual level as this provides the most power, while the treatment itself is considered unlikely to spill over to other arms. As households in the T2 arm are drawn from the same villages as those in the T1 arm, the likelihood of these farmers interacting with the control group farmers (who will be drawn from different villages) is low, given the geographic spread of the villages in the sample and the quick turnaround between the implementation and evaluation of the treatment.

Randomisation was done in office by a computer (using STATA code).

For the comparison between C (existing card design) and T1 (redesigned card): Village-level
For the comparison between T1 (redesigned card) and T2 (redesigned card and mobile-based messaging): Individual-level

Yes

68 villages

1530 households

34 villages control, 34 villages treatment

Within the 34 treatment villages, 50% of households in T1 and 50% in T2

10 percentage points