- Sample size calculation: The sample size is determined using a formula that takes into account the total population size and the desired confidence level (typically 95%) and margin of error (commonly set at 5%). This ensures that the sample is large enough to produce statistically significant and generalisable results. For example, a larger population requires a larger sample to maintain the same level of confidence and precision in the results.
- Population frame: To determine the appropriate sample size, the M&E local team needs a comprehensive list of all the farmers involved in the project. This list serves as the foundation for defining the population frame—the group of individuals from which the sample will be drawn. The population frame should be as complete and accurate as possible, encompassing all relevant farmers within the scope of the project. If there are multiple project locations or distinct groups within the farmer population (e.g., different farming practices, crops, or regions), the sampling approach may need to be stratified to ensure that these variations are adequately represented.
- Adjustments for non-response: When calculating the sample size, it’s also important to account for potential non-responses or incomplete data. Typically, a buffer (e.g., increasing the sample size by 10-20%) is included to compensate for this possibility and maintain the desired level of statistical power.

- Snowballing technique: This method is useful in reaching populations that are difficult to sample when a list is not available. It starts with a small group of known respondents (seeds) who then refer to other potential respondents within their network, growing the sample size akin to a rolling snowball.
- Referral method: Similar to snowballing, this technique relies on initial respondents to refer subsequent participants. This method is particularly valuable when targeting specific characteristics within a population that are known only to insiders (e.g., certain types of farmers or farming practices).