Many programs or projects have as one of their key goals to improve agricultural productivity. To do so, it is critical to be able to monitor yields of the crops(s) of interest. Estimation of crop yields in farmer fields has been a major topic of discussion. While knowing the productivity of new or improved crop varieties in farmer fields is important in understanding whether these varieties are more profitable than the local counterparts, it is often very difficult to obtain accurate estimates of crop yields. Scientists have tried several methods of yield estimation. Of these, the two most used methods are crop cuts and farmer recall. Farmer recalls are by far the most frequently used among social scientists who depend on survey data to estimate farmer yields. The method involves asking farmers during a farmer survey to estimate the production and area under the crop in a specified season or year. Yield is then estimated using the ratio of production to area. As the name suggests, the data are usually collected after harvest, leading critics to argue that farmers are often unable to accurately remember production and areas, especially after considerable time has elapsed between harvest and the survey. Consequently some scientists get around this problem by collecting yield data while crop is still in the field, a method known as farmer prediction (Fermont and Benson, 2011). However, studies have shown that both of these approaches can be quite inaccurate.

Consequently, many scientists prefer to use the crop cut method to estimate yield. This method dates back to 1940s and was first used in India but later popularized by FAO (FAO, 1982; Murphy et al., 1991). It involves demarcation of a subplot within the field/plot followed by measurement of production and area under the crop. In some cases, measurement is done on more than one subplot and an average taken. Crop yield is then calculated as total production divided by total harvested area in the crop cut subplot(s). This section offers guidelines on the planning and implementation of a crop-cut aimed at estimating yields in farmers’ sweetpotato plots.

The other major approach being used is to obtain yield data from on-farm trials or demonstration plots, where improved clones or varieties are often being compared to dominant local varieties. Typically, these trials or demonstration plots have protocols for how to establish and monitor them, so they should be considered researcher-farmer managed yields to some extent. Detailed on-farm protocols are available for use in obtaining data that can be analyzed by breeders using the Excel-based CloneSelector program.