In the Data repositories section, you will find links to a wide variety of publicly available data sources related to agriculture, food systems, nutrition, and health. When choosing the right dataset for your interests and research questions, you first need to understand the attributes of the data that are available.

• Unit of measurement/observation: Some datasets provide information about respondents at the individual or household level. For example, the Living Standards Measurement Study (LSMS) conducts household surveys in multiple countries, so each observation from these datasets represents a household or individual within the household. Other datasets provide national averages. For example, the World Bank World Development Indicators are a collection of country-level indicators related to development. Each observation is a national average of a development indicator, such as average cereal yield in kilograms per hectare in a given year.
• Data collection method: Many of the datasets referenced in the Data repositories section provide data observed through surveys, environmental or biometric samples, or geospatial data collection. These data are from either observational studies – in which researchers observe subjects without assigning any treatment – or experimental studies – in which researchers assign subjects to experimental conditions and then observe outcomes. Some datasets provide data generated through modeling studies – in which researchers combine observed data with a set of assumptions to generate modeled estimates. For example, the data provided in the Global Dietary Database (GDD) are modeled estimates of global dietary intake.
• Quantitative versus qualitative data: Quantitative data is data that is collected by measuring things can be expressed using numbers. Qualitative data is collected by observing and interviewing subjects, and often answers of how and why that are difficult to answer using quantitative data. Most of the datasets referenced in the Data repositories section provide quantitative data, though some repositories such as the Harvard Dataverse do contain some qualitative datasets. Qualitative data is less commonly found in publicly accessible repositories because it is more difficult to anonymize data to protect the privacy of respondents.

SCANR Tip: Some data repositories and other resources indicate the unit of measurement, data collection methods, and type of data. On the Data4Diets website, you can explore food security indicators by data collection level (household, individual, national, or market), data source or method (e.g., dietary recall, food frequency questionnaire, etc.), and dimension of food security.