Anchoring and Adjustment

Method

Condition 1: The area of China is 9,396,960 square kilometers.
Condition 2: The area of Cuba is 110,860 square kilometers.

Part of the Eastern Hemisphere including China:

Part of the Western Hemisphere including Cuba and Colombia (same scale as above):

What do you think the area of Colombia is?

.

.

.

.

.

.

.

Results

The actual area of Colombia is 1,141,748 square kilometers. Estimates when the anchor was... China: Cuba: 2900000 980900 200000 400000 50000 197000 2500000 118975 3132320 567000 6000 110860 3000000 175000 5427312 910999 2500000 776020 4000000 330680 2349240 800000 5000000

TWO-SAMPLE T TESTS FOR ESTIMATE BY ANCHOR SAMPLE ANCHOR MEAN SIZE ---------- ---------- ------ China 2,450,000 13 Cuba 456,743 10 NULL HYPOTHESIS: DIFFERENCE = 0 ALTERNATIVE HYP: DIFFERENCE > 0 ASSUMPTION T DF P ----------------- ------ ------ ------ UNEQUAL VARIANCES 3.94 13.1 0.0008

Conclusions

When people make a quantitative estimate under uncertainty, they look for some quantity that they think is related and use it to help them make their estimate. They decide whether they should make an estimate greater or lesser than the related quantity, and then adjust it up or down accordingly in order to get their estimate. However, the related quantity acts as an anchor. In other words, they find it hard to adjust too far up or down, as if the anchor were holding them back. Therefore, if the anchor they start with is higher than what they think their estimate should be, they find it hard to adjust down far enough. If the anchor is lower, they find it hard to adjust up far enough. The result is that the average estimate made from a high anchor will be higher than the average estimate made from a lower anchor.