A Practical Guide to (Successfully) Collect and Process Images Through Web Surveys

Abstract

Asking web survey respondents to answer questions by sharing images is a practice that has gained notoriety during the last years. Although this new collecting strategy may offer many advantages, it requires researchers to know how to operationalize, collect, process, and analyze this type of data, which is not yet an extended expertise among survey practitioners. This paper aims to guide researchers inexperienced in image analysis by presenting the main steps involved in the process of using images as a new source of data: 1) operationalization, 2) definition of the labels, 3) choice of the most suitable classification method(s), 4) collection, 5) enhancement, and 6) classification of the images, 7) verification of the classification outcomes, and 8) data analysis. Following this eight-step process can help practitioners assess whether image collection is appropriate for their research problem and, if so, plan their image-based research, by providing them with the key considerations and decisions to address throughout their implementation.

Keywords: images, manual classification, automatic classification, visual data, web surveys