The impact of including images in a conjoint measurement task: evidence from two small-scale studies

Abstract

In the Netherlands, a large ongoing study seeks to provide insight into the process behind preferences and choices with regard to moving and housing. As part of the development of the survey instrument that was used in this study, we examined whether visual images (such as photographs) should be included, since they may clarify particular attributes (e.g., architectural style) and make the choice task more realistic for the respondents. However, images may disturb the results as respondents may evaluate accidental details on the photographs such as the color of the window frames. This paper presents the results of two subsequent pilot studies on the impact of including images in a web-based questionnaire. In the first study, eight dwelling profiles were presented in three different ways (''text only'', ''text and color photo'' and ''text and black-and-white impression'') to 28 respondents. In the second study two different instruments were used: one with written information and directly shown photo-collages; the other with written information and photo-collages on request only. Both studies showed that the inclusion of images led to a number of differences between the results obtained with the various presentation methods. These differences may be explained by accidental details on the images. Furthermore, dwelling characteristics appear to be more important to the respondents' overall evaluation of a dwelling when shown with photos than when presented in written format. However, since including images also has a substantial number of benefits, it may not be advisable to leave out images completely from the measurement task. Should images be included in a survey instrument, it is of utmost importance to make sure that all potentially disturbing details are cleared away. Furthermore, it may be better to show a collage of pictures than only one in order to decrease the specific impact of accidental details.

Understanding Residential Location Choices for Climate Change and Transportation Decision Making: Phase 2 Report

“…In example, Jansen et al (Jansen et al 2009) estimated choice-based conjoint models to examine the influence of a set of dwelling characteristics including architectural style (traditional, modern, innovative) and number of rooms to measure the impact of image inclusion in experimental surveys of residential location. Building period (new house, secondhand house) and number of rooms was similarly examined in the Katoshevski and Timmermans (Katoshevski and Timmermans 2001) study, while Senior et al (Senior et al 2006) incorporated a preference for distinctive housing character as a housing feature.…”

“…Senior et al (Senior et al 2006) proposed a three-level land use mix measure that segmented walking time to non-residential land uses as being 2-5 minutes, 6-10 minutes, or not within a reasonable distance. Aside from the measurement of accessibility to shopping destinations, past residential choice experiments have also specified multilevel attributes related to the accessibility of green spaces and local parks (Katoshevski and Timmermans 2001;Walker and Li 2007;Jansen et al 2009). …”

“…Traditionally, stated choice experiments have used text descriptions of these varying concepts; however, recent studies (Jansen et al 2009) have noted the potential for visual imagery to better illustrate residential location characteristics. Conceptually, images likely better convey what some may interpret as abstract concepts (e.g., neighborhood, accessibility) and complement the customary use of text descriptions.…”

“…Images are provided to represent some item the research is presenting, but if the visualization itself is not carefully prepared, the respondent may interpret the images in a way that the researcher does not intend. Jansen et al (Jansen et al 2009) suggested the following seven benefits of including images in choice experiments surveys which may outweigh the potential drawbacks:…”

“…At the recommendation of Jansen et al (Jansen et al 2009), such dynamic transitions between images allow researchers to prevent observers from focusing on any potentially disturbing or distracting details that may be seen as unfavorable.…”