with No Comments

Post No.: 0144anchoring


Furrywisepuppy says:


Any number that you are asked or primed (either manipulatively by another party or unwittingly by yourself or anyone, and either consciously or unconsciously) to consider as a possible solution to an estimation problem will induce an ‘anchoring effect’; where, unless you explicitly know the correct answer, your estimate will gravitate towards that number, and any adjustments from that number will be typically insufficient. For example, if you don’t know the correct answer and I asked you how many moons have been confirmed to currently and reasonably stably orbit around the planet Jupiter, and I guessed 45 – your answer will likely be around 25 to 65. Even a number concerning a totally different query will tend to produce this effect i.e. people do not rely on anchors because they believe they’re informative of the correct or best answer yet they’ll be influenced by them nonetheless. (The correct official answer as of writing is 79.)


It plays on the fact that we tend to rely on mental shortcuts based on information we receive from the environment – but sometimes an anchor is random and sometimes it is purposely planted (e.g. when the other side asks for an outrageous opening price in a bartering situation).


So anchoring sets a reference point, which can be totally arbitrary, from which all values are then relatively compared from. As another example, to someone who doesn’t know much about the market prices for a particular wine – if a bottle of this wine was originally priced at €20, then spending €10 for it would seem cheap while spending €30 for it would seem expensive – but if that exact same bottle of wine was originally sold at €40 (a relatively higher anchor) then spending €30 for it would suddenly seem cheap. So if you’re considering how much to pay for a property or product and you have no specific expertise on the matter (or there is no single right answer) then you’ll be influenced by the initial asking price – the same item will seem more valuable if its listing or recommended retail price is high rather than low, and vice-versa, and this perception will be hard to resist even though one’s biases may think that an anchor hasn’t influenced one’s willingness-to-pay value.


Due to anchoring, you could, if there’s a sales promotion going on for a product, put up a sign that says ‘limit of 8 units per person’, and customers will buy more units on average than if the sign said ‘no limit per person’ or if there was no sign at all! This would not only be due to the anchoring effect though – this fiendish trick also plays on the implication and mental shortcut that, if something is restricted or rationed then it ‘must’ be due to it being popular and scarce, and therefore valuable and desirable (i.e. people follow each other and tend to think that something must be desirable because lots of other people are (or appear to be) buying or doing it), and the urgency of getting something that’s selling rapidly before it’s gone (scarcity value and time pressure tactics). Of course, as in this above example, the setup could’ve been artificially fabricated.


The anchoring effect is so commonly employed in our daily lives yet we’re typically blind to it. Anchors affect selling prices, buying prices, purchase quantities, donation amounts, salaries, fines, damages, sentencing durations and so on. Anchoring is a very common haggling tactic. ‘Discounts’ play on the anchoring effect all the time – by making customers think that something has suddenly become an irresistible bargain based on the price drop from a (often artificially set) high anchor price. But it can work the other way too i.e. customers using the discounted price as the new anchor and so paying anything more would seem like a rip-off.


Anchoring can occur with ideas too e.g. we can anchor our predictions and expectations for next year’s turnover on a ‘best-case scenario’ as set out by a business plan (for which business plans are typically biased towards presenting best-case scenarios). But they’re only estimates, thus when trying to forecast actual outcomes, one should think of ways that plans can go wrong too. Anchoring affects political policies e.g. an extreme anti-immigration stance or party can make an only-slightly less extreme stance or party thereafter seem moderate and acceptable in comparison.


Anchoring results in a particularly strong bias when estimates are stated in the form of a confidence interval e.g. a reliable finding, when attempting to predict the value of a stock market index on a particular day by defining an upper and lower bound so that people are ‘98% confident’ that the true value will fall within this range, is that people will anchor their upper and lower bounds too close to their best estimate (their own generated anchor), thus leading to an overconfidence effect. When people are ‘98% certain’ that a number is within a particular range, they are often wrong about 30-40% of the time(!) Woof!


So try your best to identify the presence of an anchoring effect and try to ignore these anchors when valuing something (such as ‘sale prices’ or relative discounts from over-inflated ‘recommended retail price’ anchors). Compare to other prices in the wider market (entire sub-markets could be initially based upon arbitrary anchors, hence e.g. the price of one piece of art, anchored to an ambitiously inflated estimate, can be apparently equivalent in value as the cost to build several fluffy schools! Such a sub-market has arguably lost touch with the entire market of goods and services.)


Our ‘system one’ anchors because of an unconscious priming effect – a case of the power of suggestion – system one tries its best to construct a world in which the anchor is the correct or true answer (in the same way that if I mention a stripy elephant, your mind has just automatically imagined a stripy elephant as if one exists or is true). The truth bias plays a role too. Suggestibility occurs when we hear something so strange that we try to make sense of it. But in trying to make sense of it, we become prone to believing it and it becomes plausible. We can imagine seeing, hearing or feeling something by merely bringing it to mind (e.g. if asked the question ‘do you feel a slight numbness in your left leg?’ You may report that your left leg does indeed feel a little stranger than usual, as suggested). Also, high and low anchors activate different sets of associatively coherent ideas in memory (e.g. the assumption that a person is ‘innocent and energetic’ when presented with a ‘young age’ anchor) and one’s guess draws from these biased samples of ideas and will therefore be biased too.


Our ‘system two’ anchors due to a deliberate process of adjustment, for which one usually doesn’t adjust sufficiently enough, just like slowing down to drive in the city after coming off the motorway/freeway, or someone turning the music volume down after being told by someone else that it’s too loud (where the first person will say it’s turned down enough because they’re anchored on the high volume and the other person will say it’s not turned down enough because they’re anchored on silence, thus making the first person think that genuine attempts at compromise are being unappreciated!) The perceived correct or appropriate answer has a range of uncertainty in our minds, but we tend to stop adjusting from the anchor towards that range when we only just about reach the nearest boundary of that range (e.g. if the anchor is 70, but we loosely think the correct answer is somewhere between 40 and 60, we’ll likely declare an answer nearer to 60 than 40).


People tend to adjust up or down only about half of the amount they should. Finding reasons to adjust further away from an anchor is an effortful system two operation, thus people tend to adjust less if their mental resources are depleted (e.g. when overloaded with trying to hold many digits in one’s working memory, when drunk or when overly stressed i.e. we don’t tend to make great decisions that impact the long-term when we’re overly stressed). In an experiment, people who were instructed to shake their heads whilst listening to an anchor moved farther away from that anchor, and vice-versa when instructed to nod their heads!


Anchoring proves that most of us are more suggestible than we think we are, and there are many situations where people are willing and able to exploit our suggestibility. The perfect con is one where the mark (target) never ever consciously knows he/she’s been conned – and the best way to do this is via tricks that work on one’s unconscious, subconscious and/or inattentive conscious. Since system two works on data and memories passed on by system one, system two is also susceptible to the biasing influences of anchors that make some estimates and associated ideas easier to retrieve or conceive than others. Furthermore, system two will likely have no knowledge of the effect, never mind any control of it, such that people influenced by even an absurd, useless or random anchor will tend to confidently claim that it had nothing to do with their judgements! (Experiments with control groups reveal otherwise.)


In summary – in estimation tasks, different anchors will yield different estimates because our estimates will be unwittingly biased towards those anchors and any adjustments away from such anchors will likely be insufficient. Incentives for accuracy do not reduce the anchoring effect. In many contexts though, without anchors we’d have no clue or even ballpark where to place a value (e.g. how much compensation should be awarded to a person who sustained an unusual accident?) But not having a clue will force us to rely on system two and an active, attentive and effortful search for more information; or at least we will withdraw from expressing an (emphatic) opinion about something we know insufficiently about.


Woof! I hope you will now be more alert to when arbitrary numbers are trying to anchor you up or down, particularly in commercial or negotiation contexts.


Comment on this post by replying to this tweet:




Share this post