What causes systematic under- or over-estimation of general knowledge quantities?

Question

What psychological theories predict whether individuals will underestimate or overestimate when responding to particular kinds of general knowledge quantity questions?

For instance, I have seen an unpublished study in which people were asked to guess the weight of the fuel in a certain space shuttle, and almost everyone wildly underestimated the weight. Then in another question they were asked to guess the weight of the horse which won the last Kentucky Derby and the responses were distributed fairly evenly around the truth. Why is this?

I think that the heuristics and biases literature must be relevant here, but I haven't found any citations that allow me to make predictions which are sufficiently specific.

Answer 1

One theory that may explain this is is Kahneman and Tversky's anchoring heuristic.

If you ask how much fuel the space shuttle needs, most people don't have the proper background knowledge to answer this accurately. Instead, they'll rely on a piece of information they do know--e.g., how much fuel a car needs--and adjust from there. Responses will be biased towards the anchor. Since a car requires very little fuel in comparison (~12 gallons), people will tend to underestimate the true response (~500,000 gallons).

If you first told them: The cruise ship The Norwegian Dream holds roughly 350,000 gallons of fuel, and then asked how much fuel weight the space shuttle needs, their answers will likely be closer to the truth. (Though there is the confounding variable of size-- larger vessels need more fuel, but even small vessels will need a lot of fuel to launch into space).

When it comes to horses, people have more accurate anchors-- such as a human, or an elephant-- and will tend to make more accurate judgments.

Although there are certainly more cues that people use to make size judgments, a complete theory must take into account one's background knowledge. Thus, even the anchoring heuristic is limited in that we must know what people are likely to use as an anchor for any given question.