Field Study in a Kuttanad Panchayat: Whose Data Speaks for the Panchayat?
A Plus Two class spends two days on field study in a panchayat in the Kuttanad region. Group A surveys 40 households, all of them along the main road that runs on the outer bund. Group B walks a 3 km transect from the bund into the interior paddy fields and records land use at 300 m intervals. The results are pooled: 72% of Group A's households report non-farm income, while only 31% of group B's plots are still under paddy - the rest are coconut gardens, reclaimed house plots and fallow land. Rainfall during the visit was 180 mm in two days, and several houses along the transect stood in 30–40 cm of standing water. In the debrief, one student argues that 'Kuttanad is no longer a paddy region', while another says the class has not collected evidence that can support that sentence at all.
1. Which group's design gives the more defensible evidence about land-use change in the panchayat, and what specific error does the other group's data introduce? Use the numbers in the setup.
Read the worked answer
Group B's transect is the stronger design, but neither group can yet make a statement about the whole panchayat.
- Group A's sample is spatially biased, not random: every household lies on the bund road. Roadside dwellings in Kuttanad are systematically different - better drainage, easier transport to town, higher land value, more salaried and remittance income. So 72% non-farm income is partly a real roadside pattern and partly an artefact of where the survey stood. It cannot be read as '72% of the panchayat'.
- Group B's transect is systematic sampling along a gradient, which is exactly right for detecting the bund-to-interior zonation (paddy near the bund, fallow and house plots deeper in). But 11 points on one line cannot estimate an area share either - 31% paddy is the share of sample plots, not of the panchayat's area.
The classroom disagreement is therefore a sampling problem, not a data problem. The honest conclusion is conditional: 'along one 3 km transect, paddy occupied about one-third of the sampled plots, with conversion increasing away from the bund'. Generalising requires stratification by distance from the bund, with random points inside each stratum, and an area estimate checked against the panchayat's land-use register.
2. Explain the land-use change and the flooding as one linked chain of cause and effect rather than two separate observations.
Read the worked answer
The two findings are the same process seen at different stages.
Why paddy retreats: Kuttanad paddy is below water level and needs continuous dewatering by pumps, plus bund maintenance and hired labour. Returns per hectare from paddy are low against coconut, banana or a house plot, so each field that is filled and converted shifts the cost of pumping and bund repair onto fewer remaining fields - the incentive to convert next door grows. Conversion is thus self-reinforcing, which is why a transect shows a patchwork rather than a clean boundary.
Why the water then appears in the houses: filling a field reduces the depression storage of the paddy tract, and compacted fill and house plots infiltrate far less than wet paddy soil. The bund that keeps the river out also keeps water in, so drainage depends on pumped outlets. With 180 mm in two days, the pumped outlets are exceeded and water ponds on the lowest, most recently filled plots - 30–40 cm in the survey is consistent with houses built on filled land at the tract's lowest level.
So conversion raises both the volume and the speed of runoff, and puts people on the sites that collect it. Flooding here is a land-use decision expressing itself as a hazard, not simply heavy rain.
3. Suggest two improvements to the field design and name the one thematic map the report must carry.
Read the worked answer
Design improvements
- Stratified random sampling: divide the panchayat into three belts (bund margin, middle tract, interior) and pick households/plots at random within each, so proportions can be stated with a known error rather than borrowed from one line.
- Add a measured variable, not just a described one: record ground elevation (dumpy level or GPS spot heights) and distance from the nearest outlet for each plot, so the flooding can be tested against elevation instead of asserted.
A third, cheap addition would be secondary data - the panchayat land-use register and village office crop statistics - to check the survey against official area figures, and a short interview with a pump operator on dewatering hours.
The map: a transect cross-section from the bund to the interior, showing elevation and land use at each 300 m point, with the water level during the flood drawn as a line across it. This shows the zonation and the flooding in a single frame. A choropleth of fallow/reclaimed share by survey block should accompany it, but the transect is essential because the argument is about position on a slope, which a district-level choropleth hides.