Humansys: Logic Smogic
Human designs, especially those which use people as an active part -- the law and courts, law enforcement and police, and even huge corporations -- are susceptible to the effects of illogical human decision making
The democratic election system is one such machine; elaborate mechanisms of laws and procedures are in place in order to ensure that the sensors (voters) can accurately determine the suitability of an official for office. Corruption of the machinery is often blamed for bad political choices and poor governing. But new findings indicate that it may be the sensors, and not the machinery which is flawed.
A study, published yesterday in Science (full text here, requires sub.) found that a single arbitrary visual trait -- "Competence" -- was predictive of the outcomes of 71.6 percent of U.S. Senate races in 2000, 2002 and 2004. Participants in the study were given pairs of one winner and one runner up from each race. They were asked to rate various traits, like attractiveness, age, and competence. Interestingly, attractiveness, which is usually blamed for swaying votes, had a very weak correlation (.07 out of a possible 1) compared to competence (.94 out of 1). Even more surprising was that the number of times a candidate was rated more competent was linearly related to the margin of victory in the actual election.
The implications of this are huge politically, but also from a design standpoint. The study did not pursue a description of what competence was, or try to check to see if people's visually inferred competence actually matched some real life competence. So even though we are apparently basing a large part of our initial visual assessment on competence, we could be right or we could be wrong. Does that mean graphic designers should include candidates pictures in ballots? Or should pictures be banned from being used in all political communication?
Another recent finding regarding the events of September 11th 2001 points to similar illogical behavior. A study found that it took, on average, 6 minutes for workers in the buildings to begin their evacuation, even after they knew the building had been hit by a plane and was burning. Scientists call this behavior "negative panic" and are unsure of its cause. In this case, if a plane crashing into a building isn't alarm enough, and can apparently trigger a state of complacency, might it be better to stay away from hard-core, high volume alarms for emergency situations?
In both these cases, and in others where human decision making is important, it's dangerous to base your designs on pure logic. Instead, a strategy like ethnography, or serious user testing can help identify problem areas. Just because people base a large part of their decisions on inferences, rather than facts, doesn't mean that facts, presented in the correct way can alter people's perceptions. In the above faces study, participants were also asked to pick who they would vote for out of the two candidates, based solely on visual assessment. The winning candidate in this simulated vote correlated even more strongly with the competence ratings than the actual winner did. This indicates that other information available to voters in the real world elections tended to sway people's votes, or at least introduce more variability into their choices.
The truth is, humans are driven by forces that we are largely unaware of, even after the fact; We still have democratic elections, and still believe that we are choosing the "best" candidate, even though we can arrive at very similar conclusions in seconds, using much less information. Designers need to know that this is what they are up against. You think you are giving all the information, or direction, or recommendation that a person needs to act. But it's important to check to see if that information and structure is being used, or if we are relying on something much more efficient and base.
Whether we can or cannot eventually influence or limit these inferential processes is unclear. But we definitely should know as much about them as we can. Otherwise, we're reading instructions to the passenger of the car, instead of the driver.
Copyright 2004-2006 Dominic Muren and IDFuel Team