Evidence for behavioural interventions looks increasingly shaky | Hacker News Reader

Evidence for behavioural interventions looks increasingly shaky

economist.com

This is a poor headline (from The Economist) that does not reflect the referenced work. The correct interpretation is that nudges need to be tested in the field to find out if they work. This is common knowledge in behavioural science.

Experts are not good at predicting which nudges are going to work in practice[1]. Trials are always needed.

This is no different to the requirement for drug trials. Drugs are not released just on the say so of pharmacologists, empirical evidence is needed. And yet nobody says pharmacology is looking ‘shaky’.

[1] https://www.pnas.org/doi/10.1073/pnas.2115126119

I wonder if the problem with "nudge" is that for publicity reasons, people only call it a nudge if it sounds like a tiny thing that doesn't sound like it would work. If it's an obvious intervention, such as a default option in a pension form, or higher prices on tobacco, or condom vending machines in hotel restrooms, they don't call it a nudge anymore.

And I also have the feeling some of these may be babies that the Economist's editors would like to throw out with the bathwater.

Nudges are basically the opposite of dark patterns.

If a headline seems dubious when you swap the two then it's probably got an agenda.

"Dark Patterns don't work anyway, so we should carry on using them" vs "Nudges don't work anyway, we should stop using them"

https://www.simpleusability.com/inspiration/2019/03/dark-pat...

Nudge does seem to imply some degree subtlety. In a colloquial sense, >100% tax rates on cigarettes is more of a shove than a nudge.

Pretty much every policy stick short of a bullet seems to be sailing under the flag of a nudge these days.

Absolutely: the definition of a nudge (Thaler and Sunstein’s “libertarian paternalism”) was meant to include that it is cheap and easy to avoid, and therefore makes no difference to the (mythical) fully rational consumer. Taxes or other price interventions absolutely do not qualify.

I don't follow. The Economist's editors aren't the ones deciding what is or isn't counted as a nudge in the academic literature.

The headline doesn't say nudges. It says behavioural interventions.

(and as far as I can tell, the academic literature isn't consistent on what counts as a nudge or not)

Nudges are behavioural interventions. It's just a headline.

Default options in forms are the standard example of a nudge, a tax however is definitely not a nudge, the article actually explicitly says so.

In that big meta study that questioned the efficiency of nudges, it was said in the discussion on HN that they excluded default options from the definition of nudges.

Either way, the line seems arbitrary to me, and it's part of why I think they're moving the goalposts.

I don't know what people on HN said, but it's the first example of nudges of you look here (and pretty much anywhere else):

https://en.wikipedia.org/wiki/Nudge_theory#Types_of_nudges

The article in question was "No evidence for nudging after adjusting for publication bias". That study apparently counted default choices on forms as "structural interventions" - a term wide enough to usually include immunisation programs and taxes.

Germans speak German; wow people store information and take up the routines their understanding of that information implies. Field test complete.

Social science is based on observing the same world as “hard” science. We already have plenty of models and theory about information transmission.

Human biology retains information it’s structure encounters; social science solved.

Wrapping hard science in cultural semantics does not exclude the hard science explanation from being the right explanation. The social science explanation is superfluous and almost intentionally manipulative; professors will give extra fake points for jumping higher. But if the fake points are of little real world value, people don’t care. It’s a parlor trick that works on the ignorant, not a fundamental cognitive function of everyone.

Following the same erroneous line of thought the next strategies might now situate themselves between nudging and strong enforcement. Something to really look forward to.

Asking people or setting an example are of course ideas that have to be discarded, applied statistical analysis is far more convenient and you have to write some kind of paper. They don't strictly have to but are certainly decisively nudged to do that. In the interest of their career of course, you got to get that numbers up.

At least nudging worked for pissoirs though. I like the subtlety of the critic saving the theory. It would be good to do something against publication bias as suggested in the article.

It seems weird to me that no one mentions the importance of timing and source of "nudging". I hoped govs and agencies would learn not to go overboard with nudging when trust levels in their policies was low to begin with.

Honest debate and persuasion is so 20th Century.

I kind of miss honesty, though.

Honest debate was the ultimate adventure. Sad its gone, replaced by yelling matches in feeling hive-mind choirs. Guess this how medieval times felt, with oppossing faiths.

At least for now we still have the yelling matches where people of different faiths can yell at each other without fear of burning at a stake. Getting "cancelled" as a result of yelling the unpopular thing is still less painful than torture and execution on the town's plaza.

"For now" and "still" are the words which I had to insert in the above, and their forced presence seriously scares me.

"Guess this how medieval times felt, with oppossing faiths."

And we all know that led to...crusades?

Welp.

Read the article. Retitled article should read “Not all nudges work”.

The interesting bit from the source article is that changes to the mean do not always reflect the value of the nudge. So, when judging a nudge, data should be spun various ways to determine impact on different segments of the population.

> “Not all nudges work”.

The real question being, what's the actual hit rate. If below 50%, they'd be better off flipping a coin.

What is a nudge? Do dark patterns, tactics employed by casinos or loot box games that depend on whales and children of wealthy parents count?

Fundamentally yes, except that nudge theory is about manipulating collective action (or inaction) in populations. Promoters of this theory also envision themselves as trying to help, not fleece people of their money.

I don't know a lot about this, but would an example be like when Draghi was at the ECB and he responded with "Whatever it takes" when asked about the S. Europe debt crises?

Or how exactly does nudging look like in practice?

>Or how exactly does nudging look like in practice?

If you want people to take the stairs you might make the elevator stay at the top floors so that people always have to press the call elevator button, they might choose to walk up instead of waiting.

If you want people to eat less animal based proteins you can make vegan options the default for catering/meal options with meat only available as a option you can request. Some people will take whatever is the default.

You want more organ donations you make being a organ donor the default with a option to opt-out.

The goal is to make the desired option the easiest way out so most people will avoid the cognitive work of doing what they would actually prefer if they thought about it.

Turning the option that most people want into the hardest/most tedious/most onerous option for them to select is a great way to foster contempt for your system.

I think nudges are supposed to be subtle enough that you don't notice them. And probably for every not so subtle manipulation we notice there are 5 we don't notice.

I’m sure that’s what professional nudgers like to think, but I certainly feel like I see them all over the place.

Some are meant to be noticed: mandated calorie counts on food menus (nudge to health eating), or dire warnings on cigarette packs (nudge to quit smoking). Neither of these have had any measurable impact on their goals; I welcome evidence to the contrary.

Anecdotally I was still smoking when they brought in mandated gore on British tobacco pouches, very quickly people were joking like 'I got the bloke with a hole in his side, just need the curious baby with a cigarette and I've got the lot' etc. Those who were really bothered by it just got a baccy tin or if they smoked straights, a cigarette case.

What led me to quit wasn't any 'nudging' by the government but simply that vaping is a lot more convenient than smoking and lacked the finger-wagging inherent in other nicotine replacement approaches, and once I was only vaping I could turn down the nicotine over the course of six months or so to spare the worst of the grimness of quitting.

No, Draghi's statement was signalling, not nudging. He was willing to throw billions at the problem (and eventually did), his communication was meant to reduce the need for such an intervention.

There's a difference between nudge as in the technical meaning here from behavioural psychology, and what we would commonly call a nudge in everyday english. An economist or politician might describe Draghi's statement as trying to nudge the behaviour of the banks, sure, but that doesn't conform to what behavioural psychologists mean by a nudge in the technical sense.

Which is fine of course, pretty much every specialist discipline has special meanings for some everyday words.

I thought a good example is when you get a drivers license and it asks if you'd like to become an organ donor, it "pre-checks" the "yes" box rather than leaving it blank.

The idea being people who are on the fence are more likely to just accept the "yes" than go through the effort to check "yes"

Or why the “subscribe me to all your spam and your third party’s spam” checkbox is always checked by default.

Nudging would look like this (excuse the politics, it's just the most immediate source of examples): people who are up to date on their vaccinations may get perks and be allowed to enter certain venues just by showing their card, while everyone else gets no perks and has to pay for testing to enter venues or are barred entirely.

Another example: instead of regulating testing for travel, allow third party companies to manage official testing certificates and set their own prices. This happened in the UK, where Covid tests were free from the government unless you wanted to travel - at which point you had to pay a private clinic upwards of £50 (sometimes more like £150) to get a 24 hour pass.

Nudging looks very much like incentives and restrictions in a mobile game.

I think your examples are explicit incentives, not nudges. From what I’ve read, a nudge is typically something much subtler that works on an emotional level. For example the tax office might A/B test different wording for a tax demand letter template, finding that it’s more effective to say “Get a £100 early-filing discount if you pay before $DATE” instead of “An additional £100 late-filing penalty will be charged if you fail to pay before $DATE”. A strict logician or economist might point out they amount to the same thing, but the different emotional approach nonetheless makes a measurable difference in the aggregate response. That’s what is usually meant by a nudge.

That would simply be PR or friendly copy. The implementation of nudge in your example would be to create the "discount" itself to push more people to pay taxes on time. Even then, in your example everyone already understands and agrees with the rules of "pay a fee for being late" because it's such a common practice and is rational.

Nudges don't need to be rational - at least not directly. They need to push people to do what you want them to do to attain some opaque goal.

Similar to dark patterns, nudging is not about creating reasonable guides for people to follow, it's about making them feel less contempt for being shoehorned into taking actions they'd otherwise prefer not to. Maybe people don't want to pay taxes, but the proverbial "we" decided long ago that taxes are a necessity to maintain the state and so we punish people with prison time for cheating on them. In nudging, "we" don't choose anything, the institution quietly forces our hand without any democratic process.

I've heard that whales, which might be a thing in casinos, isn't really a thing in video game markets.

I mostly hear about whales in video games

The chair of behavioral insights advisory in the WHO is one of the co-authors of Nudge Theory. So it's probably safe to say that government health programs - in any country that is supported by the WHO - also implement Nudge to some extent.

Freddie Sayers made a really, really good comment at the end of his recent UnHerd video about this:

"It's not a conspiracy. It's no a communist plot. But what it is, is taking a particular worldview, a view about how society should operate, and removing it from the normal democratic conversation, where people can battle it out, and instead hardwiring it into poerful supranational organizations."

Here is the part of the video that I am referring to: https://youtu.be/Lm-S5zlnHIk?t=528

> It's not a conspiracy ... poerful [sic] supranational organizations.

Come on. You can't claim that something isn't a conspiracy and then invoke world-government conspiracies in the same paragraph and expect to be taken seriously.

It's not a conspiracy. These powerful supranational organizations are doing it out in the open and have publicly stated their intentions. Many of us disagree with their goals, and consider their methods unethical.

http://archive.today/cuHfh

This seems like a healthier language for these types of incentives.

Without discussion, engineering a system where the elevator is on the top floor always is torture, sadism even.

But what of scholarships? Is money truly the common language spoken at the tender ages these are issued? How do you convince students, once beyond the drudgery of state-mandated public school, that the sudden tradition, idolatry, and care of academia is even a change? How do you put that fruit at eye-level without using cash to whip scholars into meeting minimums, as adjudicated by paper examination?

The problem I see increasingly is the conflation of "X as an intervention doesnt work" with "There is no evidence for X", which is two entirely different things.

To give a concrete example: "Stop-smoking interventions commonly fail" equated to "no evidence that stopping smoking improves health".

I love when Idiots like google misunderstand nudge to mean “shove”. It doesn’t work and makes me hate you. Good job. Worst of both worlds.

Can an expert in the field dissuade me of my pessimism about psychology as a science? From the outside looking in, it appears to be a field populated by grifters and incompetents who churn out junk knowledge.

Power Posing (Amy Cuddy) Priming (Bargh, Kahneman, and others) Outright fraudsters (Diederik Stapel, Dirk Smeesters) Precognition (Daryl Bem) Disgust & Homophobia (Yoel Inbar) Anchoring (Kahneman and Jacowitz)

Or is it all of human sciences (the scandal of Alzheimer's research comes to mind)?

> about psychology as a science?

You can easily have the opposite point of view: The so called hard sciences deal with human concepts and not reality. An alien society would not have the same physic or math. Many concepts of our physic are a bit circular or need ad hoc parameters.

Kepler make a good model from data by Tycho Brahe. Newton made an even larger model from Kepler's laws, yet it's only a model. Einstein... you get it. Without Tycho Brahe all the rest would be simply speculations. Without Michael Faraday, would we have Maxwell, and Heaviside?

The Romans did not liked math, and the Bantu migration did not even used writing scripts as we know them. Gengis Khan was able to conquer and create the greatest empire of the world and Mongols of that time used little scripts.

The so called human sciences on contrary deal with reality, it's incredibly complex. There is no real way for us to understand the complexity of reality.

> An alien society would not have the same physic or math

They would have the same math. They might have a different number system.

I disagree.

All arguments I have seen that math are universal are in my opinion circular or using assumptions that are specific to our species or even to our western culture.

For example communication with alien, by assuming that math is universal, has been proposed by sending string of prime numbers that represent 2D pictures with this algorithm: (https://math.dartmouth.edu/~carlp/PDF/extraterrestrial.pdf)

1* Create a rectangle

=> Assumes a 2D picture has the same meaning for aliens, even for some humans it is not true.

https://theconversation.com/blind-in-the-mind-why-some-peopl...

2* Divide it into units, such that each side has the length of a certain prime number.

=> Again this is a very visual way to think, even insects would have problems with that.

=> We have an interest in prime numbers which may not be shared. Factorization of prime numbers considers couples of numbers --typically 2D mindset-- what if aliens are more interested in factorization of triplets? After all it would be more logical to send graphic information

=> We have also a way to understand numbers and operators which might not be universal, for example "zero" is a weird number with some properties different from other numbers. Even IEEE had to invent NaN (Not A Number) which is not a mathematic concept in our culture. Aliens might see zero as a null operator (instead of a number) that could be composed with multiplication and division operators.

3* Encode images into the rectangular grid by making each square black or white (or a dot or a dash).

=> Again this is a very visual way to operate

4* Take apart the image row-by-row or column-by-column and make it into one long pattern.

=> Again a very visual way to operate, in addition more symbols will be needed to mark the end-of-line and end-of-picture. This is not told as the whole idea is to transmit symbols through math, so the need for extra mathematical symbols ruins the full concept. Indeed if there are end-of-line and end-of-picture, there is no need for the factorization of prime numbers trick

5* Transmit the message.

1. Evidence based psychology is legit. Non-evidence based psychology, probably not (except for coincidence).

2. Same with other fields, which ironically may not suffer from the same credibility problem (notably medicine).

3. The Alzheimer's thing was about outright fraud. Totally different kettle of fish.

4. All sciences are human sciences.

The “human sciences” are just freaking hard - for all the reasons you are pessimistic. The “hard sciences” are much easier - less of that soft stuff to work around.

Agree that the humanities are often pants-on-head whacko.

Disclaimer: hard science engineer type.

not an expert but around age 17 a psychologist saw I carried a knife and asked if I got separated from my father when I was 9 months old. he blew my mind because it was true. till that moment I was thinking of psychology as some sort of fortune telling, now suddenly it's in the realm of science because you can make deductions

Reply on news.ycombinator.com