Spencer Greenberg: Effective Behavior Change

If we want to im­prove the world (and our­selves), we need to start by chang­ing the way we live — our habits and be­hav­iors. In this talk, Spencer Green­berg, founder and CEO of Clear­erThink­ing.org, dis­cusses why be­hav­ior change mat­ters and tech­niques we can use to get bet­ter at it.

Below is Spencer’s talk, which we’ve lightly ed­ited for clar­ity. You may also watch it on YouTube or read it on effec­tivealtru­ism.org.

The Talk

Thank you so much for hav­ing me. I’m re­ally happy to be here.

If you con­sider hu­man­ity’s prob­lems from 3,000 years ago, you’ll note that many of them have to do with the in­abil­ity to un­der­stand and con­trol the phys­i­cal world.



They in­clude things like how to get your crops to grow, treat in­fec­tion, and stop the spread of dis­ease.


But as sci­ence ad­vanced, our prob­lems be­came in­creas­ingly of a differ­ent kind. They in­volve a sub­stan­tial amount of be­hav­ior change that makes them differ­ent from many of the prob­lems we had be­fore.


Heart dis­ease is the No. 1 cause of death wor­ld­wide, ac­cord­ing to the World Health Or­ga­ni­za­tion’s 2016 figures. De­pres­sion is the No. 1 cause of dis­abil­ity wor­ld­wide, ac­cord­ing to the World Health Or­ga­ni­za­tion’s 2018 figures. Both of these, of course, have sig­nifi­cant ge­netic and en­vi­ron­men­tal com­po­nents, yet they also can be sub­stan­tially im­proved via be­hav­ior change.


For heart dis­ease, ex­er­cise, diet, and med­i­cal com­pli­ance can help. For de­pres­sion, cog­ni­tive be­hav­ioral ther­apy and be­hav­ioral ac­ti­va­tion are both ev­i­dence-based tech­niques [that can help].

But with sci­en­tific progress, be­hav­ior change has ac­tu­ally cre­ated a whole new class of threats.


Nu­clear war, global warm­ing, and bioter­ror­ism from new viruses are, fun­da­men­tally, re­ally about be­hav­ior. With that in mind, I want to talk about what I’m go­ing to call “pos­i­tive be­hav­ior change,” which is be­hav­ior change de­signed to im­prove your­self or the world.

If you think about the clas­sic three cause ar­eas in effec­tive al­tru­ism (EA) — global health, an­i­mal welfare, and tech­nol­ogy risk — one in­ter­est­ing thing we can do is look at ca­pac­i­ties that cut across all three ar­eas.


One of these ca­pac­i­ties is ev­i­dence eval­u­a­tion: how we [in­ter­pret] ev­i­dence and de­cide how strong it is. I think we can agree that this cuts across all EA work. For global health, we ask ques­tions about the ex­tent to which bed­nets re­duce malaria. For an­i­mal welfare, [we con­sider whether] cage-free cam­paigns are cost-effec­tive for re­duc­ing [the num­ber of an­i­mals in cages]. For tech­nol­ogy risk, AI re­searchers fore­cast [the po­ten­tial effects of] de­vel­op­ments in AI. So, [ev­i­dence eval­u­a­tion] is one ca­pac­ity that we need in a com­mu­nity that’s try­ing to do good.

Another ca­pac­ity is an­a­lyt­i­cal or philo­soph­i­cal anal­y­sis. In global health, we have ques­tions like “When forced to choose, should we save chil­dren or save adults?” In an­i­mal welfare, we ask, “Are fish, chick­ens, pigs, and hu­mans equally ca­pa­ble of suffer­ing?” With tech­nol­ogy risk, we have ques­tions like “How difficult would it be to con­trol a su­per­in­tel­li­gent AI?”

What I’m propos­ing to­day is that there’s a third ca­pac­ity that we talk about a lot less, but is crit­i­cally im­por­tant to many of the things we’re try­ing to do: the ca­pac­ity to change be­hav­ior. Get­ting peo­ple to use bed­nets in high-malaria ar­eas is fun­da­men­tally a new be­hav­ior we’re try­ing to cre­ate, as is get­ting con­sumers to ac­cept plant-based and clean meat prod­ucts in the an­i­mal welfare space. A be­hav­ior change in tech­nol­ogy risk is get­ting AI re­searchers to deeply con­sider pos­si­ble risks of their work.

Be­hav­ior change is not just im­por­tant at the level of these causes; it’s also re­ally im­por­tant for you.


If you want to have more of an effect on the world, it’s helpful to learn how to change your own be­hav­iors to do so. Plus, be­ing able to change un­de­sir­able be­hav­iors can make you sig­nifi­cantly hap­pier.

Un­for­tu­nately, be­hav­ior change doesn’t hap­pen au­to­mat­i­cally.


Imag­ine tel­ling your­self, “I should re­ally go to the gym ev­ery day.” Well, I think we all know that just hav­ing that thought, and even be­liev­ing that you want to do it, is a far cry from ac­tu­ally go­ing to the gym ev­ery day. It’s just not go­ing to hap­pen au­to­mat­i­cally.

It’s also in­ter­est­ing to think about this in a global-health con­text.


How would you go about get­ting peo­ple in ru­ral com­mu­ni­ties to add chlo­rine to un­safe drink­ing wa­ter? You just give them chlo­rine, right?
If you ask Ev­i­dence for Ac­tion, they’ll tell you that’s not even close to what you have to do.


Ev­i­dence for Ac­tion cre­ates ed­u­ca­tional cam­paigns to try to get peo­ple to use chlo­rine to make their wa­ter safer. They went into com­mu­ni­ties and ed­u­cated them, as well as giv­ing them the chlo­rine. They de­ter­mined that if you put dis­pensers of free chlo­rine where peo­ple are pul­ling their drink­ing wa­ter, it is in­cred­ibly con­ve­nient to use. The chlo­rine is right there when you need it. It re­minds you to use it.

You might think that [by mak­ing the chlo­rine free and con­ve­nient], peo­ple would en­gage in this be­hav­ior. But it’s not enough. If the dis­pensers were emp­tied, not only were peo­ple un­able to use the chlo­rine [in that mo­ment], but they stopped us­ing it al­to­gether [even af­ter the dis­pensers were re­filled]. Maybe this is be­cause peo­ple fell out of the habit of us­ing the dis­pensers. Maybe it’s be­cause peo­ple stopped hav­ing faith in the re­li­a­bil­ity of the chlo­rine dis­pensers.

In any case, Ev­i­dence for Ac­tion then went through a lot of effort to make sure that the chlo­rine dis­pensers would always be kept full. Even that wasn’t enough. The or­ga­ni­za­tion ended up giv­ing peo­ple who lived in the com­mu­nity a t-shirt and pay­ing them to pro­mote chlo­rine use. They stud­ied mul­ti­ple vari­ables as­so­ci­ated with which pro­mot­ers did the best job [to help en­sure they se­lected effec­tive pro­mot­ers]. But even with all of these be­hav­ior change strate­gies, they found that while they could get peo­ple to use the chlo­rine, their use tended to trail off over time. Be­hav­ior change can be a re­ally sig­nifi­cant challenge.

I want to sum­ma­rize what I’ve cov­ered so far:


1. Pos­i­tive be­hav­ior change is an im­por­tant part of most at­tempts to im­prove the world to­day.
2. It can be very hard to do it well.
3. I be­lieve this is sig­nifi­cantly un­der-dis­cussed. [Those of us in the EA com­mu­nity] don’t usu­ally con­cep­tu­al­ize our­selves as try­ing to change be­hav­iors, even though I think that is a sig­nifi­cant amount of [what’s in­volved in try­ing to help the world].

But does it make sense to talk about im­prov­ing ca­pac­ity at be­hav­ior change? Can some­one be­come an ex­pert in be­hav­ior change? I see it as analo­gous to ev­i­dence eval­u­a­tion. We can all agree that or­ga­ni­za­tions work­ing effec­tively to im­prove the world are of­ten good at ev­i­dence eval­u­a­tion. I think we can also agree that they could still benefit from out­side ex­per­tise — a statis­ti­cian, data sci­en­tist, epi­demiol­o­gist, or some­one who is an ex­pert in ex­per­i­men­tal de­sign.


I think the same think­ing ap­plies to be­hav­ior change. You can imag­ine peo­ple who are ex­perts in the psy­chol­ogy of be­hav­ior change and who un­der­stand the bar­ri­ers to chang­ing be­hav­iors helping or­ga­ni­za­tions that are already very good at be­hav­ior change get even bet­ter at it. Ex­perts could help or­ga­ni­za­tions figure out new strate­gies that they may not have con­sid­ered.

[There is already some ev­i­dence that this works.]


Be­hav­ior change ex­perts in or­ga­ni­za­tions like Be­havi­oural In­sights, the Cen­ter for Ad­vanced Hind­sight, and ideas42 have been work­ing with differ­ent non­prof­its and gov­ern­ment or­ga­ni­za­tions to im­prove pos­i­tive be­hav­iors more effec­tively. And a re­cent se­ries of ran­dom­ized con­trol­led tri­als shows that their work cre­ates pos­i­tive out­comes in a va­ri­ety of differ­ent ways.


In my own work at Spark Wave, I see the power of be­hav­ior change all the time. We cre­ate new com­pa­nies from scratch that are de­signed to help solve prob­lems in the world through a so­cial sci­ence lens. Then, if a product is promis­ing, we re­cruit a val­ues-al­igned, pas­sion­ate, driven CEO to work with us and ul­ti­mately spin the product out into a new com­pany.


The work we do tends to break into two cat­e­gories. One is im­prov­ing peo­ple’s lives through be­hav­ior change. We do this with prod­ucts like UpLift, which aims to provide the best self-help for de­pres­sion in the world, and Mind Ease, which aims to re­li­ably help peo­ple suffer­ing from anx­iety feel bet­ter when­ever they need to. We also cre­ated Clearer Think­ing, a web­site that has more than 20 free tools and train­ing pro­grams to help you make bet­ter de­ci­sions and bet­ter un­der­stand your­self.

[The sec­ond cat­e­gory cen­ters on] ac­cel­er­at­ing the study of be­hav­ior change and hu­mans more gen­er­ally. For this, we have prod­ucts like Positly, which is a new plat­form for re­cruit­ing peo­ple for stud­ies. Positly aims to make it easy to get the right par­ti­ci­pants, at the right time, to do the right thing. Guid­edTrack is an­other product of ours. It’s a new lan­guage for build­ing be­hav­ioral in­ter­ven­tions and com­plex stud­ies.
What can you do to use be­hav­ior change more effec­tively? First, if you work for or run an or­ga­ni­za­tion, I think it’s helpful to (1) clearly define your or­ga­ni­za­tion’s be­hav­ior change challenges, (2) be sure to view them as be­hav­ior change challenges, and (3) po­ten­tially even bring in ex­ter­nal be­hav­ior change ex­per­tise to help you meet those challenges.
Se­cond, if you want to change your own be­hav­ior to be more effec­tive or hap­pier in your life, I would ad­vo­cate that you ap­ply be­hav­ior change strate­gies sys­tem­at­i­cally.
But what does it mean to ap­ply be­hav­ior change sys­tem­at­i­cally? For the rest of this talk, I’m go­ing to make be­hav­ior change ex­tremely con­crete.

There are a num­ber of differ­ent frame­works for try­ing to cre­ate be­hav­ior change. I’m go­ing to walk you through some of them. I’ll then point you to­ward a new re­source that Spark Wave launched. It sum­ma­rizes all of these differ­ent frame­works and at­tempts to make the prin­ci­ples of be­hav­ior change very easy to un­der­stand.


The first frame­work I’d like to dis­cuss is the trans­the­o­ret­i­cal model. The main idea is that peo­ple go through differ­ent stages in a be­hav­ior change, such as:

* The pre­con­tem­pla­tion phase, dur­ing which some­one is not yet ready to change
* The con­tem­pla­tion phase, dur­ing which they’re get­ting ready but not ac­tu­ally tak­ing ac­tion
* The prepa­ra­tion phase, [dur­ing which peo­ple in­tend to take ac­tion and start tak­ing small steps to change]

There are sev­eral more stages. What’s neat about [this frame­work] is you re­al­ize that differ­ent tech­niques are ap­pro­pri­ate when peo­ple are at differ­ent stages. Also, you can con­cep­tu­al­ize [the change pro­cess] as peo­ple move from one stage to the next.

A sec­ond frame­work that I think is in­ter­est­ing is the the­ory of planned be­hav­ior.


This frame­work iden­ti­fies three ma­jor fac­tors in­volved in be­hav­ior change:

* A per­son’s at­ti­tude to­ward the be­hav­ior change: Do they think it’s good and do they de­sire it?
* The sub­jec­tive norm: Do the peo­ple this per­son cares about ap­prove or dis­ap­prove of the be­hav­ior?
* Per­ceived be­hav­ioral con­trol: Does the per­son be­lieve that they can ac­tu­ally achieve this be­hav­ior?

To­gether, these three fac­tors in­fluence some­one’s in­ten­tion to en­gage in the be­hav­ior, and that ul­ti­mately leads to be­hav­ior change.

A third frame­work I want to tell you about is the EAST model by the Be­havi­oural In­sights team.

EAST pack­ages some use­ful be­hav­ior change tech­niques into a sim­ple-to-use for­mat. So, if you’re try­ing to change a be­hav­ior, you want to try to make it:

* “E” for easy, by do­ing things like ap­ply­ing de­faults and re­duc­ing has­sle
* “A” for at­trac­tive, by at­tract­ing at­ten­tion, pro­vid­ing in­cen­tives, etc.
* “S” for so­cial — for ex­am­ple, by show­ing that most peo­ple have already en­gaged in the be­hav­ior
* “T” for timely — for ex­am­ple, by prompt­ing peo­ple when they’re most likely to be re­cep­tive to en­gage in the new be­hav­ior

Fi­nally, I want to in­tro­duce a be­hav­ior change frame­work from Spark Wave.


It’s called “the “10 con­di­tions for change frame­work.”

We de­vel­oped this frame­work to help you pin­point ex­actly why a be­hav­ior is not oc­cur­ring ei­ther in your­self or in a pop­u­la­tion. It al­lows you to an­a­lyze what’s block­ing a be­hav­ior change, so that you can de­ter­mine which strate­gies to ap­ply to re­move those block­ers.

It’s de­signed to be a suffi­cient model for be­hav­ior change; if all 10 con­di­tions are jointly met, it’s very likely that some­one will en­gage in the be­hav­ior.

I should say that this is a work in progress. I can’t guaran­tee we won’t add an eleventh con­di­tion later.

If some­one is not en­gag­ing in a be­hav­ior, that means that one or more of these 10 con­di­tions are not be­ing met.
The frame­work al­lows you to di­ag­nose which of block­ers are oc­cur­ring. The con­di­tions are grouped into three phases:



* The de­ci­sion phase, dur­ing which some­one de­cides to do the be­hav­ior. As we’ll see, we need three differ­ent con­di­tions to be met in the de­ci­sion phase.
* The ac­tion phase, which in­volves a se­ries of ac­tions across time. For each ac­tion, six differ­ent con­di­tions must be met. For any given be­hav­ior, there are prob­a­bly many differ­ent se­quences of ac­tions that would be suffi­cient to make that be­hav­ior change. So, you can com­pare differ­ent pos­si­ble se­quences and pick a se­quence that you think is rel­a­tively easy to do, and then an­a­lyze each of the ac­tions one by one, look­ing for the six con­di­tions I’m go­ing to talk about for each ac­tion.
* The con­tinu­a­tion phase, the idea be­ing that af­ter you finish com­plet­ing an ac­tion, and as time passes, some of the con­di­tions that were pre­vi­ously met might start to fail to be met as time goes on and things change. Those con­di­tions need to con­tinue to be met.

Let’s break this down a bit more. In the de­ci­sion phase, we need three con­di­tions to be met: (1) the per­son con­sid­ers chang­ing the be­hav­ior, (2) the per­son de­sires to change the be­hav­ior, and (3) the per­son in­tends to change the be­hav­ior.

In the sec­ond phase, for each of the given ac­tions that make up the be­hav­ior, we want the per­son to (1) re­mem­ber to take the ac­tion, (2) be­lieve that at­tempt­ing the ac­tion will help them reach a goal, (3) choose to take the ac­tion, (4) know what it takes to take the ac­tion, (5) have the needed re­sources to take the ac­tion, and (6) em­body the skills and traits needed to take the ac­tion.

Fi­nally, in the con­tinu­a­tion phase, the per­son main­tains the at­tributes re­quired to take fu­ture ac­tions.

That’s a lot of in­for­ma­tion. I’m go­ing to break this down even fur­ther to de­ter­mine whether I can use the frame­work re­cur­sively.


Imag­ine some­one named Bob. There are three new be­hav­iors Bob wants to en­gage in: (1) to be­come a ve­gan, (2) to go to a ther­a­pist to treat his anx­iety, and (3) make sure he takes so­cietal risk into ac­count when he pub­lishes re­search about viruses, which is what Bob does for a ca­reer.

For each of these be­hav­iors, Bob will be likely to suc­ceed if the 10 con­di­tions for change are met.
Let’s go through them.

Dur­ing the de­ci­sion phase, Bob con­sid­ers chang­ing the be­hav­ior. If he never even con­sid­ers chang­ing this be­hav­ior, it’s very un­likely Bob will ac­tu­ally change it. So, what might this look like?


For go­ing ve­gan, it might look like hav­ing a con­ver­sa­tion with a friend that gets him to con­sider it. For go­ing to a ther­a­pist for his anx­iety, maybe he sees an ad­ver­tise­ment for ther­apy. For tak­ing so­cietal risks into ac­count in Bob’s re­search, maybe he reads an ar­ti­cle about the im­por­tance of do­ing that.


Next, Bob must de­sire to change his be­hav­ior.


“De­sire” can mean in­tu­it­ing a pos­i­tive effect of chang­ing it — a very “Sys­tem 1” type of de­sire us­ing [Daniel Kah­ne­man’s frame­work from Think­ing, Fast and Slow, in which Sys­tem 1 refers to fast, in­tu­itive think­ing] — or it could be a more cog­ni­tive, “Sys­tem 2” de­sire [a more log­i­cal, de­liber­ate re­al­iza­tion]. It’s best if you ex­pe­rience both. But you can imag­ine that if Bob doesn’t ex­pe­rience ei­ther, it’s very un­likely for him to en­gage in these be­hav­iors.

So, how might we help Bob de­sire these be­hav­iors?


We might ap­peal to the moral­ity of eat­ing ve­gan and the po­ten­tial benefits of go­ing to a ther­a­pist for anx­iety. [For the third change], maybe ap­peal­ing to con­sis­tency would work; if Bob went into biolog­i­cal re­search to help peo­ple, why would he pub­lish re­search that might harm peo­ple?

Next, we need for Bob to in­tend to change his be­hav­ior.


This means that Bob an­ti­ci­pates do­ing the be­hav­ior at a par­tic­u­lar time, or in a par­tic­u­lar place, or over a par­tic­u­lar pe­riod. If you made a bet with Bob, would he bet against or for him­self in terms of chang­ing that be­hav­ior? We can all re­late to that feel­ing (e.g., “I know I should go to the gym and I want to go to the gym, but I’m not go­ing to do it right now be­cause I’m too busy in life”). You don’t ac­tu­ally in­tend to do the be­hav­ior.


For Bob to in­tend to change these be­hav­iors, we might help him set a fre­quency in­ten­tion (“I’m go­ing to do this be­hav­ior ev­ery day”), a cal­en­dar in­ten­tion (“I’m go­ing to do this be­hav­ior at a par­tic­u­lar point in time”), or an im­ple­men­ta­tion in­ten­tion (“When­ever X oc­curs, I’m go­ing to do this be­hav­ior”).

Then, we reach the ac­tion phase. Re­mem­ber, a be­hav­ior change usu­ally in­volves a num­ber of differ­ent ac­tions, and for each one, six con­di­tions must be met. First, Bob must re­mem­ber to take the ac­tion.


If he for­gets to take the ac­tion, it prob­a­bly won’t oc­cur.


To help him re­mem­ber, we might cre­ate cal­en­dar or alarm re­minders for Bob’s phone, or maybe so­cial re­minders — Bob’s friends or fam­ily could re­mind him.

Bob must choose to take the ac­tion.


This is about choos­ing to take the ac­tion over other op­por­tu­ni­ties available at that mo­ment. Does Bob do the ac­tion or does he play video games in­stead? If Bob doesn’t choose to take the ac­tion, he’s prob­a­bly not go­ing to en­gage in the be­hav­ior. For this, we might do things like re­move temp­ta­tion.


We could put Bob in a place where there aren’t video games available at the mo­ment he’s sup­posed to take the ac­tion. Or, we might struc­ture the ac­tion around Bob’s pro­cras­ti­na­tion. Maybe he can work on this ac­tion when he’s avoid­ing do­ing his taxes. Or maybe at the mo­ment [where he has a chance to take] the ac­tion, Bob can vi­su­al­ize the harm that will oc­cur if he doesn’t take the ac­tion.

Bob must know how to take the ac­tion.


If he doesn’t, he’s prob­a­bly not go­ing to do it.


We could help him mem­o­rize what to do, have him con­duct his own re­search on how to do it, or en­gage him in a dis­cus­sion about how to do it.

Bob must have the nec­es­sary re­sources to take the ac­tion. If he doesn’t, it’s go­ing to be re­ally hard for him.


We could help Bob by stock­piling ve­gan foods in his home. Maybe the gov­ern­ment offers free ther­apy in cer­tain cases. Maybe Bob could get grants for con­sid­er­ing the risk of his re­search.


Bob must be­lieve that tak­ing the ac­tion will help him achieve his ul­ti­mate goals around the be­hav­ior. If Bob doesn’t have a sense of self-effi­cacy and thinks that he’s not go­ing to be able to suc­ceed at tak­ing the ac­tion, or if he doesn’t be­lieve that the ac­tion will re­ally get him where he’s try­ing to go, then it’s un­likely he’s go­ing to take the ac­tion.


We might, for ex­am­ple, find peo­ple similar to Bob who have suc­ceeded at achiev­ing the be­hav­ior and de­scribe those peo­ple to Bob. Maybe we can provide quan­ti­ta­tive ev­i­dence show­ing the effec­tive­ness of tak­ing these ac­tions. Maybe we can provide Bob with a role model to coach him.


Bob must em­body the phys­i­cal and men­tal char­ac­ter­is­tics or skills needed to take the ac­tion. Of course, if Bob doesn’t have those at­tributes, it’s very un­likely he’ll be able to take the ac­tion.


We can have Bob prac­tice. We can put Bob through an as­sess­ment to make sure that he has the nec­es­sary skills. We can give Bob real-time as­sis­tance at the mo­ment when he’s try­ing to take the ac­tion.


Fi­nally, in the con­tinu­a­tion phase, Bob must main­tain all of the con­di­tions we talked about. If any of them fall out of place as time passes, or af­ter Bob com­pletes cer­tain ac­tions, then Bob prob­a­bly can­not ul­ti­mately suc­ceed at chang­ing the be­hav­ior.


To help him, we can do things like offer a big re­ward if Bob suc­ceeds, help him iden­tify which vari­ables are [slip­ping], or have him make a pub­lic com­mit­ment de­signed to keep those vari­ables in­tact over time.


Quick sum­mary: There are three phases.


The de­ci­sion phase re­quires that three con­di­tions be met. In the ac­tion phase, the be­hav­ior is bro­ken into mul­ti­ple ac­tions, and for each of those ac­tions, six con­di­tions must be met. Fi­nally, there is the con­tinu­a­tion phase; if we want the per­son to con­tinue the be­hav­ior, cer­tain con­di­tions must be main­tained over time.

I don’t ex­pect you to re­mem­ber all of this. The frame­work is to­tally free for you to use.


You can find it at spark­wave.tech/​con­di­tions-for-change, a web­page that walks you through these phases in great de­tail.

One thing that I didn’t go into much is that for each of these con­di­tions, there are a num­ber of differ­ent strate­gies you can use. So, if you’re think­ing about cre­at­ing a new be­hav­ior in your own life or in some­one else’s life, you can walk through the differ­ent con­di­tions, figure out which ones may not be met, and look at the pos­si­ble strate­gies you can use (or brain­storm oth­ers). That can help you di­ag­nose what’s go­ing on.

The other thing I’m ex­cited about is the bot­tom of the web­page, where we’ve sum­ma­rized seven other be­hav­ior change frame­works. We want you to think about be­hav­ior change and do­ing it more effec­tively in your work and life. We en­courage you to use our frame­work, but also check out oth­ers. They can help you think about how to do it more suc­cess­fully.

With that, I ask you to go out and make pos­i­tive be­hav­ior change in your own life or the lives of oth­ers. The world needs it.

Moder­a­tor: Thanks very much, Spencer. My first ques­tion is: How did you come to these 10 steps? What was your method­ol­ogy?

Spencer: Great ques­tion. I like to think of what we did as a suffi­ciency anal­y­sis: if A, B, C, and a range of other con­di­tions are all met, here is what will oc­cur. We tried to sys­tem­at­i­cally figure out what A, B, C were — and that’s why we might end up adding more con­di­tions in the fu­ture.

We also made a database of over 250 be­hav­ior change strate­gies. So, as we work to link those differ­ent strate­gies to the differ­ent con­di­tions we iden­ti­fied, we may even­tu­ally see that we’re miss­ing some­thing — that there’s a strat­egy that doesn’t fit with any of these 10. That will help us iden­tify more con­di­tions.

The ba­sic idea is to try to cre­ate as strong a set of suffi­cient con­di­tions as you can. Of course, I’m not go­ing to say that it’s ab­solutely guaran­teed that some­one who changes a be­hav­ior will have met all 10. The idea is that it’ll be very likely.

Moder­a­tor: Did you test this out on any­one?

Spencer: I’ve been ap­ply­ing it to differ­ent be­hav­ior change challenges that I see at Spark Wave. This is a re­ally new model, so it’s still early days.

But re­ally, the idea of the model is to use it to sys­tem­atize your think­ing about be­hav­ior change and iden­tify what’s go­ing wrong. So, you can think of it like this: There are hun­dreds of pos­si­ble be­hav­ior change strate­gies and there are sev­eral pos­si­ble be­hav­ior change block­ers. We want you to be able to sys­tem­at­i­cally think through them to figure out which ones are most ap­pro­pri­ate for your challenge.

Moder­a­tor: I have a ques­tion from the au­di­ence fo­cused on helping cre­ate change in other peo­ple — for ex­am­ple, pre­ven­ta­tive mea­sures like helping a child eat healthily to pre­vent is­sues later on in life. When the pos­i­tive out­comes come later, how does this frame­work help?

Spencer: Let’s say you’re try­ing to help a child eat more health­fully. You can just go through these 10 con­di­tions and for each of them ask, “Are they met?” Has the child con­sid­ered eat­ing health­fully — yes or no? You prob­a­bly will know that. Does the child de­sire to eat health­fully — yes or no? As you go through the con­di­tions, you will iden­tify sev­eral block­ers.

Some­one was talk­ing to me yes­ter­day about helping their friend cre­ate a pos­i­tive be­hav­ior and we were go­ing through this model to­gether. We no­ticed that about five of these vari­ables were out of whack. They were like, “Wow. Okay. Now we’re re­ally clear on why that per­son is not en­gag­ing in the be­hav­ior.” It doesn’t mean you’re go­ing to work on all five vari­ables, but you can pin­point the op­tions with the high­est lev­er­age — i.e., where you think that they’re go­ing to get the most im­prove­ment for the amount of effort.

Moder­a­tor: How long have you been work­ing on this pro­ject?

Spencer: I’ve been think­ing about be­hav­ior change now for quite a long time, be­cause our com­pany, to a large ex­tent, is a be­hav­ior change com­pany and we’ve been de­vel­op­ing this database of be­hav­ior change tech­niques. That’s where this all be­gan. That was over a year ago.

More re­cently, we’ve been try­ing to sys­tem­atize the pro­cess, look at ex­ist­ing frame­works, and say, “Where do we think we can add value be­yond what the ex­ist­ing frame­works do?”

Moder­a­tor: Is this in­tended to be just a re­source for peo­ple or is there a fur­ther aim?

Spencer: It’s in­tended to be a re­source so that or­ga­ni­za­tions try­ing to help the world can use it, and so that peo­ple can use it in their own lives. Also, we use it in our own work.

Moder­a­tor: Great. A ques­tion from the au­di­ence: In the chlo­rine ex­am­ple, are there any ob­vi­ous failure points us­ing your frame­work?

Spencer: I haven’t ap­plied the frame­work to that spe­cific ex­am­ple. But if you think about the de­sire con­di­tion, that has to do with ed­u­ca­tion. You need to go and ed­u­cate peo­ple about the benefits of chlo­rine to hit that de­sire con­di­tion.

The choice con­di­tion in­volves the per­son get­ting to the well and see­ing the chlo­rine dis­penser. Are they ac­tu­ally go­ing to choose to do it at that mo­ment? That re­lates to the chlo­rine dis­pensers some­times emp­ty­ing, which causes peo­ple to stop us­ing them (even when they are re­filled). By keep­ing them full all the time, you’re helping to pro­mote that choice, so that in that mo­ment they don’t look at the dis­penser and think, “Ah, it’s prob­a­bly empty. I’m not go­ing to bother check­ing.” So I think if you break it down, you will find that the bar­ri­ers can be linked to the differ­ent con­di­tions [in our model].

Moder­a­tor: In terms of busi­nesses, do you see this be­ing ap­plied in ex­actly the same way, where you have a prob­lem work­ing as a team and you go through cer­tain steps?

Spencer: You mean in­ter­nally?

Moder­a­tor: Yeah.

Spencer: I’m glad you brought that up, be­cause you could think about some be­hav­ior change as get­ting an or­ga­ni­za­tion or a gov­ern­ment to change its be­hav­ior. There are cer­tainly unique dy­nam­ics to a group, com­pany, or gov­ern­ment. But at the end of the day, it’s re­ally peo­ple mak­ing de­ci­sions.

So, you can ap­ply the model at the level of the in­di­vi­d­ual peo­ple in the group. You may not know all of those in­di­vi­d­u­als, but you may know char­ac­ter­is­tics about them. So, if you’re try­ing to cre­ate a gov­ern­men­tal be­hav­ior change, who are the peo­ple with power? You can ap­ply this frame­work to that per­son (or maybe it’s those three peo­ple). Same with a com­pany. Are you an­a­lyz­ing the be­hav­ior of man­agers? You can ap­ply the model to those in­di­vi­d­ual man­agers.

Moder­a­tor: That seems quite la­bor-in­ten­sive — to walk through the model for ev­ery sin­gle per­son on the team.

Spencer: No — the idea is not that you have to do it for ev­ery sin­gle per­son, nec­es­sar­ily, but for ev­ery sin­gle *type* of per­son.

Let’s say you have a com­pany and you’re try­ing to change the be­hav­ior of all of the man­agers. You can think of them as a class of peo­ple. When you’re eval­u­at­ing the model’s con­di­tions for an in­di­vi­d­ual, you ask a ques­tion like “Is this con­di­tion met?” or “To what ex­tent is this con­di­tion met for this in­di­vi­d­ual?” When you’re do­ing it for a group of peo­ple, you ask some­thing like, “What per­centage of peo­ple will strongly meet this con­di­tion — are 90% of peo­ple already meet­ing it, or only 10%?”

Moder­a­tor: How suc­cess­ful do you find that these kinds of in­ter­ven­tions are? I mean, we prob­a­bly all wish we went to the gym more and we’ve tried these var­i­ous things. What kind of suc­cess rate do you see among peo­ple who ac­tu­ally try to im­ple­ment these strate­gies?

Spencer: That is ex­actly the point. Be­hav­ior change is re­ally hard. So many of us have things we want to do with our be­hav­ior that we don’t do. It’s re­ally, re­ally challeng­ing. That’s why we de­signed the sys­tem. It’s usu­ally not enough to just go with your gut and say, “Oh, I’ll try this and maybe it will work.” We want to provide a way to be re­ally sys­tem­atic.

I think the re­al­ity is you’re still go­ing to fail a lot, but we’re try­ing to make you fail sig­nifi­cantly less. That’s the goal. If the frame­work works, that’s our goal for it.

Moder­a­tor: Awe­some. Thanks so much, Spencer.

No comments.