*2.1. Conceptual Framework*

Irrigation practices can be explained by the context within which the farm is exposed. The changing context (price and climate variability) often leads to changes in practices. Price and climate variability are external factors that constitute the main driving forces. However, within the same context, farmers can have different practices. The adoption of practices can also be explained by internal factors, i.e., factors directly linked to the farming system and the farmer (e.g., structure of soil, age of the farmer). The conceptual framework (Figure 1) is based on the association of material, structural and psychological factors previously identified as potential factors that explained the heterogeneity of practices. It assumes that both observable and non-observable factors contribute to the heterogeneity of practices.

**Figure 1.** The conceptual framework used to analyze the observable and non-observable factors that a farmer uses to choose the date of first irrigation of an irrigated grain maize crop.

Observable factors (i.e., structural and material; [27]) are categorized into four groups: farm characteristics, agronomic practices, irrigation practices and farmers' personal characteristics (i.e., age, education level, experience as a maize grower). For the non-observable factors, we used the conceptual framework developed by Martin-Clouaire (2017), which considers risk preferences and decision-making process factors. Daydé's (2017) three decision-making process factors are defined as follows:


#### *2.2. Implementation of the Conceptual Framework*

#### 2.2.1. Survey Design

One challenge of the survey design (Figure 2) was to identify ways to obtain subjective data related to non-observable factors. To do this, we used a variety of elicitation methods in the survey questionnaire [28]: a lottery game to assess the level of risk aversion, scenarios to assess the level of reactivity, and a mind map and role-playing to assess the level of deliberation. The level of assistance was assessed using direct elicitation of information by asking a variety of questions. Observable factors were also assessed using direct elicitation of information. We asked about financial data at the end of the interview, when the farmer was more comfortable and more inclined to provide important and confidential data.

**Figure 2.** Survey structure and elicitation methods to obtain structural data (dark gray), farming practices data (light gray), and psychological data (white).
