Preference Stability and Choice Consistency in Discrete Choice Experiments

Abstract

We investigate preference stability and respondents' levels of choice consistency within discrete choice experiments. These are investigated via a discrete choice experiment featuring four information treatments and a retest survey completed 6 months after the first. Three information treatments concern a novel, stem cell, food technology with the fourth information treatment featuring a non-novel technology. We find stable welfare estimates over the three information treatments; the use of emotive keywords does not systematically change WTP to avoid the stem cell food technology. We find a significant WTP to avoid the non-novel technology. We find high levels of intertemporal preference stability and choice consistency. The determinants of choice consistency are examined. Two factors are significant: a measure of the choice task's complexity (entropy) and a novel measure of respondents' cognitive capability, derived within the survey. We find a significant selection issue, not previously identified in discrete choice experiment test-retest studies, with those opting to take the retest exhibiting greater choice consistency in the initial survey. This suggests that studies which report high levels of test-retest choice consistency, without accounting for the selection issue, are overstating population choice consistency.

Choice certainty, consistency, and monotonicity in discrete choice experiments

“…However, we follow the definition by Carlsson, Mørkbak, and Olsen (2012), and define choice consistency as 'making the same choice in two equal choice tasks'. An identical definition of choice consistency is used in the DCE literature on the determinants of consistency in a test-retest setting (Brouwer, Logar, and Sheremet 2016;Mørkbak and Olsen 2015;Rigby, Burton, and Pluske 2016;Schaafsma et al 2014). Finally, 'choice monotonicity' is assumed to hold if a respondent chooses a non-dominated alternative in a choice task that contains a dominated alternative, which is a hypothetical alternative that is worse than at least one other alternative in a choice task with respect to all attributes.…”

“…However, even if evidence for their existence is found, this does not necessarily imply choice irrationality. Preferences are, for example, allowed to shift between choice tasks (Rigby, Burton, and Pluske 2016). In addition, a key assumption in random utility theory (McFadden 1974), which underlies the DCE method, is the existence of an informational asymmetry between the decision maker and the analyst.…”

“…The occurrence of uncertainty, inconsistency, and non-monotonicity is especially likely in DCEs in an environmental context, since respondents are often confronted with unfamiliar goods and/or attributes. Even if the characteristics of the good are familiar to a respondent, it may still be cognitively challenging to conceptualise specific attribute levels (Rigby, Burton, and Pluske 2016).…”

Transfers of landowner willingness to accept: A convergent validity and reliability test using choice experiments in two Canadian watersheds

“…The two measures differ in their reference levels of utility (Freeman, Herriges, & Kling, 2014;Whittington, Adamowicz, & Lloyd-Smith, 2017). 1 The temporal reliability and validity of transfers have also been assessed by examining the stability of models or values over time (e.g., Brouwer & Logar (2014); Rigby et al (2016); and Price et al (2017)). 2 Several stated preference valuation studies have elicited landowner WTA concurrently at multiple sites or regions, though none test transfers.…”

“…The temporal reliability and validity of transfers have also been assessed by examining the stability of models or values over time (e.g., Brouwer & Logar (2014); Rigby et al. (); and Price et al. ()).…”