ECON30019 Chap.10 Availability, Correlation Neglect and Overconfidence
Availability, Correlation Neglect and Overconfidence
Under the availability heuristic we assess the probability of an event by the ease with which it comes to mind, treating one event as more likely than another if it comes to mind more easily. As a shortcut it is not foolish, since frequency and memorability are correlated, and it fails because of everything other than frequency that makes an event memorable.
Correlation neglect is the failure to allow for information that shares an origin, so several reports tracing back to one source get counted as separate confirmations, which inflates confidence rather than merely shifting it.
The chapter then covers the three-way overconfidence taxonomy, which separates claims about an absolute level, a relative rank and an interval width, along with the planning fallacy and a statistical warning about better-than-average surveys that the subject issues about its own literature.
What this chapter covers
- 01
The availability heuristic, stated with its direction
- 02
What makes an event memorable other than how often it happens
- 03
Three domains where the distortion shows, and where availability meets base rates
- 04
Correlation neglect, and why double counting inflates confidence
- 05
A naivety parameter, and the bimodal result that an average would hide
- 06
Overestimation, overplacement and overprecision as three different quantities
- 07
Interval estimation, and intervals that are too narrow
- 08
The planning fallacy, with a case that ran ten years late
- 09
The hard-easy effect, and where the direction of error reverses
- 10
Better than average against better than median, and why only one is evidence
Can most people be above average?
- 1Compute the mean: zero times one plus ten times nine, divided by ten, which is nine.
- 1Count above the mean: nine of the ten score ten, which is above nine, so a clear majority are above the mean.
- 1Compute the median, the point splitting a group into upper and lower halves, which is ten, and note that nobody scores above it.
- 1Draw the conclusion: a skewed distribution allows almost everyone to be above the mean, so a majority claiming to be above average is consistent with perfect accuracy. At most half a population can be above the median, so only rank-based surveys can evidence overplacement.
Key terms
- Availability heuristic
- A rule that reads the chance of an event off how readily it springs to mind, so whatever is easier to recall is taken to be more likely.
- Correlation neglect
- The failure to allow for information that shares a common origin, so that signals tracing to one source are treated as independent evidence.
- Naivety parameter
- A measure of how much of the correlation correction is skipped: zero indicates fully rational behaviour and one indicates full correlation neglect.
- Overestimation
- Overestimating one's own actual performance. The quantity in question is an absolute level, with no comparison to anyone else.
- Overplacement
- Overplacing one's performance relative to others. The quantity is a rank, which is why the median rather than the mean is the correct benchmark.
- Overprecision
- Excessive certainty about the accuracy of one's own beliefs, showing up as stated confidence intervals that are too narrow.
Availability, Correlation Neglect and Overconfidence FAQ
Why does correlation neglect matter more than a slightly wrong posterior?
Because confidence grows with the number of independent observations. An agent counting five correlated reports as five independent ones ends up far more certain than the evidence warrants, and in the direction the shared source pointed. It also compounds with availability, since heavy coverage makes an event easy to recall and supplies many apparently separate confirmations of one story.
Why is the experimental result described as bimodal rather than as an average?
Because almost nobody sits at the average. Where the signals share no origin most participants get it right, and when signals are correlated the distribution splits, with similar proportions behaving fully rationally and exhibiting near-full correlation neglect. Reporting a population average would misdescribe every individual, and an intervention should ask which group a person is in.
Which of the three overconfidences do trading results speak to?
Mainly overprecision. Investors overrate how sharp their information about particular firms is, which raises the perceived value of acting on it, so trading volume rises and costs reduce net returns. The published comparison found men traded forty-five per cent more than women, with returns reduced by 2.65 points a year, against 1.72 for women.
Exam move
Every question in this chapter is answerable by naming the quantity the person was wrong about: an absolute level, a rank, or an interval width. Write that phrase into your first sentence, before the label, since it forces the classification to be right rather than guessed. Practise the mean and median counter-example until you can produce it in two lines, because it is the fastest way to dismantle a badly framed survey claim.
For correlation neglect, rehearse the remedy as a procedure that makes provenance visible rather than as an exhortation to be careful.
Working through Availability, Correlation Neglect and Overconfidence in ECON30019? Sia is AskSia’s AI Economics tutor — ask any ECON30019 Availability, Correlation Neglect and Overconfidence question and get a clear, step-by-step explanation grounded in how ECON30019 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.
