Biased pre-decisional processing of leading and non-leading alternatives
United States academic research, April 2014
We often assume that we can evaluate information objectively – but is that really true? Do we fairly assess alternatives before making a decision? Recent American research suggests from the moment we make our first choice between alternatives we start to pre-determine the outcome by increasingly preferring the “leader” choice and increasingly distancing the “trailer”. By Juliet Bourke - Consulting, Partner.
Cognitive psychology has long told us that people make decisions and then unconsciously become more confident about the “correctness” of their decision over time, but how does it happen? Is it only a matter of selectively attending to positive information which supports our preliminary decision (confirmation bias)? What happens to the information which supports the alternative? Is this merely neglected (selective attention bias)? Recent research by Professors Blanchard and Carlson (Georgetown University), together with Professor Meloy (Pennsylvania State University) suggests that decision-making (and in particularly choosing between alternatives) is a multi-dimension process in which we both amplify the option we have started to lean towards through the incorporation of additional information and we increasingly distance ourselves from the option we have decided to leave behind. However, when additional information is evaluated independently, and not by reference to our initial choice, we see a very different picture.
The aim of these two experiments was to understand how the evaluation of information becomes distorted. In particular, the researchers sought to understand how the evaluation of information becomes distorted once a preliminary selection has been made between two (or more) alternative choices. Why does our thinking process become biased?
Professors Blanchard and Carlson together with Professor Meloy conducted two experiments to test preliminary decision-making and the evaluation/incorporation of subsequent information. In the first experiment 180 participants from Amazon’s Mechanical Turk were randomly assigned to one of two conditions. In the first condition the participants were asked to make a preliminary choice between either two back packs or two drycleaners, they were then provided with information about six attributes, one attribute at a time. For example, after being provided with information about Backpack V and Backpack M, they were provided with a follow up piece of information about the material it was made from and then another piece of information such as its dimensions. In relation to the drycleaners the information included location and turnaround time, for example. Each time the participants were provided with a new piece of information they were asked to evaluate how appealing they now found the two alternatives (from “very unappealing” to “very appealing”, rank the two alternatives (e.g. Backpack V and Backpack M) and asked to their confidence that the leading alternative would be their final selection. In the second condition, just like for the first condition, participants were asked to make a preliminary choice between the two items (i.e. from either the backpack pair or dry-cleaner pair), however when it came to evaluating subsequent information, the control group was asked to evaluate subsequent information in isolation and not told which item it related to.
In the second experiment the researchers asked 121 university students to make a preliminary selection between six alternative restaurants (rather than just two) and assess six linked attributes (e.g. dining area and menu). In the control condition, 83 students made a preliminary selection and were, as for experiment 1, provided with information about the six attributes independently. Once again participants were asked to evaluate the appeal of their initial choices, rank and express a confidence level.
The researchers found that participants distorted their evaluation of linked information in both Experiment 1 and 2 after making a preliminary decision about an alternative. In particular, after participants in Experiment 1 made an initial selection between two alternatives, they were significantly more likely to have a positive bias for information which appeared to support the item they had selected and a negative bias for information relating to the item they had rejected. In contrast, control participants showed no distortions in the way they evaluated subsequent information. These findings mean that the bias of information evaluation is not just one of simple favouritism and cherry picking, but also negativism and distancing. They also mean that the process is accumulative, namely that over time we build on our initial preference, and gain increasing confidence with the incorporation of new information which we have evaluated in light of our initial choice.
In Experiment 2 the researchers found similar outcomes, with some slight differences. In particular, as for Experiment 1, the distance between the initial choice in favour of the “leader” item, and against the five alternatives grew over time with the addition of new information, meaning that new information “pushed up” the leader and “pushed down” the alternatives. What was different however was that not every “alternative” was the same, meaning that the one that seemed closest to the leader was not pushed down so hard, whilst the one that was seen as most different was pushed down harder.
All of these findings support the idea that an initial preference has a significant impact on how we evaluate subsequent information which is linked to that preference. In particular, we don’t evaluate information objectively and on its own merits, but rather see information which supports our initial choice in a positive light, and, conversely, see information which conflicts with our choice, negatively. These findings have significant implications for the way we present information, for example in relation to decisions about strategy, operations or talent. Clearly the more we are able to present information independently, the more likely we are to give it the weighting it deserves.
To read the full article, see Blanchard, S. J., Carlson, K. A., and Meloy (2014) Biased pre-decisional processing of leading and non-leading alternatives, Psychological Science Vol 25(3) pp. 812-816.ermelmelBoris Groysberg and Katherine Connolly. "Great Leaders Who Make the Mix Work." Harvard Business Review 91, no. 9 (September 2013): 68–76.