Advice taking and decision-making: An integrative literature review, and implications for the organizational sciences

The question

When someone gets advice before making a decision, how much do they actually use it, and what changes how much?

The method

An integrative review of about two decades of “Judge-Advisor System” lab experiments [1] — studies pairing a decision-maker (the “judge”) [1] with a person or source giving a recommendation (the “advisor”) [1].

The findings

The single most consistent result across the literature is “egocentric advice discounting” [1] — judges systematically underweight advice relative to their own opinion, typically shifting only about 20–30% of the way toward the advisor’s estimate, per Harvey and Fischer’s (1997) commonly-cited finding [1]. Discounting is smaller — but doesn’t disappear [1] — when the advisor has more relevant expertise than the judge [1], when the advice has a track record of being accurate [1], and when money is riding on getting the decision right [1]. Using advice generally does improve decision accuracy overall [1], and averaging several independent advisors helps more than following any single one, because it cancels out individual error rather than just averaging it in [1]. Advisor confidence sways judges more than advisor accuracy does [1] — but the two aren’t reliably linked, so a confident advisor isn’t necessarily a right one [1].

The limits

A review of many separate lab studies, most using artificial judgment tasks (estimating quantities) [1], with advisors who are strangers to the judge rather than partners in an ongoing relationship [1]. The authors flag the effect of a prior relationship between judge and advisor — trust, familiarity — as an area needing more research [1].

• When you ask a report or a peer for advice and then quietly go with your own first instinct anyway: that’s not a one-off lapse, it’s the modal finding in this literature — egocentric discounting happens even though it’s somewhat smaller when the advisor is demonstrably more expert than you are [1].
• When you’re deciding how many people to consult before a hard call, versus leaning on one trusted advisor (see The Calibration Call): the research favors polling a handful of independent sources, because averaging uncorrelated advice cancels out error in a way a single advisor’s opinion — however good — can’t [1].
• When an advisor’s confidence is what’s swaying you: check their track record instead. The paper’s finding is that confidence and accuracy aren’t reliably linked [1] — judges follow confident advisors more, but confidence often isn’t a valid signal that the advice is right.