Abstract
There are many methods that can be used to test questionnaires, each with its own strengths and weaknesses. The best approaches to questionnaire testing combine different methods to both broaden and strengthen the results. The US Census of Agriculture (COA) is conducted every five years and collects detailed information on agricultural production, inventories, practices, and operator demographics from agricultural establishments. Preceding each COA, evaluation and testing is done to test new items in the questionnaire and improve data quality for the subsequent COA. This article will describe how a multi-method approach, which we call Bento Box Testing, was applied to establishment questionnaire testing leading up to the 2017 COA. Testing included solicitation of expert opinion, historical data review, cognitive testing, a large scale field test, and qualitative follow-up interviews. The benefits of these testing methods, considerations for establishment survey testing, and how their results in combination provide a stronger evaluation are discussed.