An application of item response theory for agricultural sustainability measurement

The concept of agricultural sustainability has been evolving since the mid-twentieth century. However, there is still not a universally accepted method for its measurement. Strong data requirements are a major obstacle to developing a useful farm-level sustainability index. We propose using item response theory models to generate a farm-level agricultural sustainability index. Item response theory models have several advantages over existing methods, the most important of which is that our index is independent of the variables used in the model. As such, farm-level sustainability scores can be estimated with readily available data and compared across different sets of variables from multiple regions. We use data from the Farm Accountancy Data Network and other secondary sources to estimate a farm-level index in Germany. In line with the literature, the results of our estimations indicate a positive relationship between farm size and sustainability, higher levels of sustainability for crop and mixed farming systems, and below-average performance for livestock farms and vineyards. We further test the sensitivity of the index against randomly missing data and simulate a scale linking procedure that tests the flexibility in measuring multiple regions with different data sets, finding that the index is generally robust in both analyses. Supplementary materials accompanying this paper appear online.