The anchoring data shows averages. What was the median, and did you control for outliers? (Craig)

Cunha conducted a quick clean by removing the two extreme bands from each distribution, for example, estimates that Turkey's population was two billion or just 80 people. He noted that any remaining outliers would inflate standard error, making it harder to detect statistically significant differences. The fact that results were significant well below p = 0.01 suggests the findings were not driven by outliers.

Why did the 1.8% cash back at $50 not work as an effective decoy in the credit card experiment? (Praveen)

Cunha attributed it to insufficient calibration. He suggested that raising the decoy's price further, for example to $90, might have created enough perceptual distance to shift preferences. He also noted that the utility curve for percentages is concave, so the perceived difference between 1% and 2% is greater than between 2% and 3%, meaning small percentage differences near the higher end carry less weight.

Have you seen the Dunning-Kruger effect interact with any of these decision-making methodologies? (Brian)

Cunha said he had not directly studied this interaction. Because participants were randomly assigned to conditions, any differences in self-assessed expertise would be distributed as noise across groups rather than systematically favoring one condition. With a large enough sample and a measure of expertise, the interaction could be tested, but the current data do not speak to it.

How do you foresee loss aversion and the conceptualization of 'free' changing in a post-COVID-19 world? (Kevin)

Cunha suggested that the pandemic has heightened perceived risk and is pushing more decisions into System 2 processing. Behaviors that were previously automatic, such as touching grocery store bags, now require deliberate evaluation. He expected this to be at least a temporary shift, noting that the constant processing of negative risk information is cognitively depleting and likely contributes to fatigue and health issues.

What is your take on the ethics of manipulating consumer behavior based on irrational decision-making? (Molly)

Cunha acknowledged the tension directly. Marketers will use these tools to serve their own interests, which is not always aligned with consumer welfare. He argued for consumer education and for applying behavioral insights at the policy level to improve welfare outcomes, such as better energy use or healthier eating, while recognizing that the same tools can be used in ways that do not benefit consumers.

How often are companies using available metrics to shape attainment of KPIs rather than measuring whether they are actually achieving their objectives? (Kevin)

Cunha framed this as an incentive alignment problem. Brand managers rewarded on short-term sales will invest in promotions over long-term brand advertising. Similarly, clients who enter research with a preferred outcome will seek data that confirms it and question data that does not, a pattern he described as confirmatory hypothesis bias. Aligning incentives and building in checks against confirmation bias are the structural solutions, though neither is easy to implement.

Behavioral Economics for Marketing and Consumer Research

Marcus Cunha, Director of the MMR Program at the University of Georgia, presents behavioral economics as a practical lens for understanding how consumers actually make decisions, contrasting it with classical rational-choice theory. Drawing on live survey experiments run with conference attendees, he demonstrates how anchoring, decoy effects, range-frequency perception, loss aversion, and hedonic editing can systematically shift responses, and argues that market research design must account for these effects to avoid misleading results. The session closes with a Q&A covering outlier handling, decoy calibration, COVID-19 effects on decision processing, and the ethics of applying behavioral nudges.

Key Takeaways

Most consumer decisions are driven by System 1 (fast, automatic), yet most survey and research designs evoke System 2 (deliberate, effortful). This mismatch can produce responses that do not reflect real purchase behavior.
Anchors shift estimates even when they are clearly irrelevant. In live experiments, a high anchor doubled population estimates for Turkey (41M vs. 85M) and raised expected software pricing by 65% ($75 vs. $124). Research teams should audit question order and numeric context for unintended anchoring.
Decoy options, even dominated or unavailable ones, reshape perceived value and shift choice proportions. Marketers can use this deliberately; researchers must control for it when presenting product concepts or pricing options.
Loss aversion is asymmetric. Participants consistently preferred risk-free gains but switched to risk-seeking behavior when facing equivalent losses. Framing the same product as '1% fat' versus '99% fat-free' exploits the same mechanism.
The distribution of contextual stimuli, not just their absolute values, determines perceived price and quality. Shifting the skew of surrounding price points changes how a fixed price ($15.99) is judged, with downstream effects on perceived generosity and quality.
Confirmatory bias and short-term incentive structures push brand managers and research clients toward data that supports pre-formed conclusions. Awareness of this tendency is the first step toward more objective research design.

Questions & Answers

The anchoring data shows averages. What was the median, and did you control for outliers? (Craig): Cunha conducted a quick clean by removing the two extreme bands from each distribution, for example, estimates that Turkey's population was two billion or just 80 people. He noted that any remaining outliers would inflate standard error, making it harder to detect statistically significant differences. The fact that results were significant well below p = 0.01 suggests the findings were not driven by outliers.
Why did the 1.8% cash back at $50 not work as an effective decoy in the credit card experiment? (Praveen): Cunha attributed it to insufficient calibration. He suggested that raising the decoy's price further, for example to $90, might have created enough perceptual distance to shift preferences. He also noted that the utility curve for percentages is concave, so the perceived difference between 1% and 2% is greater than between 2% and 3%, meaning small percentage differences near the higher end carry less weight.
Have you seen the Dunning-Kruger effect interact with any of these decision-making methodologies? (Brian): Cunha said he had not directly studied this interaction. Because participants were randomly assigned to conditions, any differences in self-assessed expertise would be distributed as noise across groups rather than systematically favoring one condition. With a large enough sample and a measure of expertise, the interaction could be tested, but the current data do not speak to it.
How do you foresee loss aversion and the conceptualization of 'free' changing in a post-COVID-19 world? (Kevin): Cunha suggested that the pandemic has heightened perceived risk and is pushing more decisions into System 2 processing. Behaviors that were previously automatic, such as touching grocery store bags, now require deliberate evaluation. He expected this to be at least a temporary shift, noting that the constant processing of negative risk information is cognitively depleting and likely contributes to fatigue and health issues.
What is your take on the ethics of manipulating consumer behavior based on irrational decision-making? (Molly): Cunha acknowledged the tension directly. Marketers will use these tools to serve their own interests, which is not always aligned with consumer welfare. He argued for consumer education and for applying behavioral insights at the policy level to improve welfare outcomes, such as better energy use or healthier eating, while recognizing that the same tools can be used in ways that do not benefit consumers.
How often are companies using available metrics to shape attainment of KPIs rather than measuring whether they are actually achieving their objectives? (Kevin): Cunha framed this as an incentive alignment problem. Brand managers rewarded on short-term sales will invest in promotions over long-term brand advertising. Similarly, clients who enter research with a preferred outcome will seek data that confirms it and question data that does not, a pattern he described as confirmatory hypothesis bias. Aligning incentives and building in checks against confirmation bias are the structural solutions, though neither is easy to implement.

Session Notes

Context and Framing

Marcus Cunha opened by situating behavioral economics relative to classical economics. Classical theory assumes unbounded rationality: people maximize utility, calibrate risk accurately, hold consistent preferences across time, and treat goods and money as perfectly substitutable. Behavioral economics does not reject this model but reframes it as a normative benchmark, a standard showing how far actual behavior deviates from the rational ideal.

"Behavioral economics theories are positive. They describe how the world works, not how it should work."

Cunha traced behavioral economics from Herbert Simon's bounded rationality concept (1955) through its move into policy, notably the UK Behavioural Insights Team and the Obama administration's health and energy programs. The field's growth, he argued, came less from new ideas and more from economists and practitioners successfully translating academic psychology into applied settings.

System 1 and System 2 and the Research Design Problem

Cunha used a dual-process framework, noting that System 1 (fast, habitual, non-conscious) and System 2 (slow, controlled, resource-intensive) are one instance of a broader class of dual-processing theories in psychology, including the elaboration likelihood model. Because System 1 handles most low-involvement decisions, research methods that prompt deliberate reflection, for example, asking respondents to walk through a laundry detergent purchase, surface System 2 responses that do not represent actual behavior. Prior purchase behavior, not reasoned preference, is often the best predictor of the next purchase.

Live Experiments with Conference Attendees

Cunha ran a pre-conference survey with 58 participants, randomly assigning them to experimental conditions across several studies. He used these results to illustrate core behavioral effects.

Anchoring and Adjustment

Participants were asked to estimate the population of Turkey after first being exposed to either a low anchor (is it larger or smaller than 300,000?) or a high anchor (is it larger or smaller than 300 million?). Results were statistically significant despite the small sample.

Low anchor group average estimate: 41 million
High anchor group average estimate: 85 million
Actual population: approximately 82 million

A second experiment asked participants to estimate the age at which Mahatma Gandhi died (actual: 78), after anchoring to either 9 years old or 140 years old, both obviously implausible values.

Low anchor (age 9) group estimate: 68.3 years
High anchor (age 140) group estimate: 82 years

A marketing application tested Snickers bar purchase quantities. Participants shown a sign saying 'buy 18 bars for your freezer' bought an average of 5.5 bars versus 2.4 bars for those shown no quantity anchor. A software pricing experiment showed that adding a high-end enterprise tier ($299) raised the estimated fair price for a small business version from $75 to $124.

Decoy Effect and Contextual Comparisons

Using a wine choice scenario, Cunha demonstrated that adding a dominated option (higher price, lower quality than one of the existing options) shifts preference toward the option it most closely resembles. The effect can occur even when the decoy is marked as unavailable or sold out. He cited the Economist subscription case, where a print-only option priced identically to a combined print-and-web subscription served as a decoy, boosting selection of the combined subscription from roughly 16% to 84%.

In his own conference experiment using credit card cash-back offers, a decoy at 0.8% cash back and $30 annual fee shifted about 40% of participants toward the 1% / $20 option. A second decoy at 1.8% and $50 produced no meaningful shift, which Cunha attributed to insufficient perceptual distance at that point on the concave utility curve for percentages.

Range-Frequency Effects on Price Perception

Participants were told a friend gave them a bottle of Brazilian wine priced at $15.99. One group saw a store shelf where most prices were below $15.99 (left-skewed); another saw most prices above $15.99 (right-skewed). The fixed price of $15.99 was identical in both conditions, yet participants in the left-skewed context rated the wine as more expensive, higher quality, and the gift as more generous. Cunha connected this to range-frequency theory: both the extremes of the distribution and the density of values around the target influence perception.

The Power of Free

When Lindt truffles were priced at 4 cents and 1 cent, 81% of participants chose the premium truffle. When prices dropped by 1 cent each (to 3 cents and free), the proportion choosing the premium truffle fell by 22 percentage points. From a classical standpoint the relative difference is identical; behaviorally, 'free' creates a disproportionate pull toward the cheaper item.

Loss Aversion and Prospect Theory

Participants chose between a 50% chance of winning $1,000 (expected value $500) and a guaranteed $450. Most chose the sure $450, consistent with risk aversion in the domain of gains. When the scenario was reframed as losses (50% chance of losing $1,000 vs. a certain loss of $450), most switched to the risky option, consistent with risk seeking in the domain of losses. Cunha illustrated the asymmetry using the value function from prospect theory: the pain of a $10 loss exceeds the pleasure of a $10 gain of equal magnitude.

A pizza-pricing scenario showed the same logic. Two restaurants charged $30 total for a pizza and wings. One discounted the pizza; the other discounted the wings by the same amount. Participants preferred the pizza discount, because discounting the higher-reference-point item (pizza) produces a larger perceived gain than discounting the wings.

Hedonic Editing

Participants were asked how they would mentally account for drinking a bottle of wine originally purchased for $25 that now costs $75 to replace. Responses varied: approximately one-third answered $75 (the replacement cost, the objectively rational answer); others reported $0 (account closed), $25 (original cost), $50 (framed as profit), or $25 plus interest. The variety illustrates that people actively reframe past transactions to manage regret and satisfaction.

Implications for Marketing and Research Practice

Marketing Applications

Use anchors: marked-down reference prices, per-customer quantity limits, and high-end product tiers all set the range within which buyers evaluate value.
Design decoys to make the most profitable option appear relatively superior.
Skew the surrounding price distribution to influence how a target price is perceived.
Bundle products so that one item in the bundle is free; this disproportionately increases overall perceived value.
Frame discounts against the item for which a loss feels largest to the buyer.

Consumer Research and Insights Applications

Numeric context in surveys (preceding questions, scale endpoints, example values) can anchor responses independently of the topic being measured.
The skewness of stimuli distributions in surveys affects perception of those stimuli.
Decoy alternatives included in a choice task will shift preference even if they are not the focus of the research.
Low-involvement decisions should not be studied using methods that force deliberate articulation; prior behavior data is often a stronger signal.
Confirmatory hypothesis bias, and the short-term incentive structures that reinforce it, can distort how research is commissioned, interpreted, and acted upon.

Transcript

Read the full transcript

especially my PhD from Cornell very happy to move back to my hometown of Charlotte so yeah I'm very happy to be a part of the conference today more importantly I'm learning so much if you Marcus is only shared screen but I wanted to just go over the ground rules we've discussed at the beginning - one of these pots already today patient action we're happy to take care of it but please down be patient about this opportunity to engage and work with the people with all these people that available right now to take advantage of that and reach out to people getters in the industry and can potentially future so I wanted to introduce today Marcus Cunha he is going to be talking about behavioral economics today for marketing and consumer [Music] thank you still needed here so I was just thanking for the opportunity to participate thank you the ISEE conference in this day that we are wearing team it's a great opportunity to be able to share our thoughts my talk is going to be a little different than the previous talks because I come from an academic standpoint but I have great connections with the industry because of my role as director of the MMR program anniversary of Georgia so the topic today is a topic that's very hot in the industry these days behavioral economics and behavioral theories and what can we learn from such such theories to be better researchers so I usually give this talk to you know private corporations or business associations or to students and research programs and I focus on these four areas for the time being I think what we're going to focus today's to figure out what what really is behavior economics and I'm going to go over some heuristics about him biases that I want to show you based on your own data so I see that we have a hundred and thirty three people right now participating and I have 58 people who took the survey so the large majority of you already failed of course so we so behavior economy behave economic is that to you today was developed to challenge what I call classical economics is this idea that we have just unbounded rationality we are just computing machines that can maximize our you chillin all the time is the assumption that we always pursue and achieve optimal options were excellent are calibrating chances and risk and computing expected values preferences are consistent across time in places good service and money are of equal value and perfectly substitute each other in states of wealth determine how we fare against each other so that's classical economics and the idea of unbounded rationality well this classical theory is considered as a theory that's normative it tells us how we should behave they suggest that we are we should always be maximizing our relativity and to do so would process information in a way to be able to estimate what's the expected value of every situation behave economics beauty on the other hand they're positive they describe it actually how the world works not how it should work I'm not saying that you know conventional classical economics here they are of no value they are very valuable because they provide us with a standard they let us know how much we are deviating from that rational ideal behavior so they describe how we behave individual you know show how we use these heuristics they'll reach errors in perceptions and decisions and they are based on this idea that our rationality is bounded another model as you see like with that citation 1955 although behavior phenomics has been a lot of interest recently the principles underlying behavior Konami's are not allowed people have in psychology have been studying behavioral decision theory judgment decision-making but I think economists were better at marketing it too dangerous and I think behavioral economists also did a good job at bringing the academic findings into the industry in the pine-sol behave economics got a lot of attention when the behavioural insights team of the UK was put together to help influence how people behave it also during the Obama administration it was very prevalent with a lot of behavior economies helping government influence how people choose for health care options how they choose better treatments how we save energy so for example you could tell people you're spending too much much in energy and that affects the environment and that affects your cost but a better way it might be just like well look the relative to your neighbor you're spending this much just show a bar graph to influence behavior so behave economy is not that new but I think it's application leaving the academia to the industry is what's more not oh so obviously when I presented to audience I played this game is called the huge made em game so I usually I select two players and I tell them I I'm giving one of the players $10 this player has been to make an offer to another player to keep some amount of their money so so for example if I got $10 and you guys are playing against me I would make you an offer so I let's say I offer you five dollars and it's if you accept my offer we both keep our money if you guys reject my offer with none of us keep their mind so generally speaking when I do these are the audience's they know each other that people are going to make offers that are around five dollars four dollars or six dollars if they're being exerted from a classic economic standpoint the rational decisions in terms of max magnitude utility would be I would offer you $1 I would maximize negativity and you should be happy that you got a dog because if you turn down their offer you're gonna get zero dollars but what we see in reality is that people the from those standards and that's where behavior periods come into place and that's what people call irrational behavior it's not irrational from the standpoint that were used to but deviates from that idea of bounded rationality and machines that process information to come up with the optimal utility for themselves so you could be influenced the Attic I could affect tell a player well imagine you took a quiz and because you did much better than everybody else you got the ten dollars but you want to get the $10 if the person accepted offer you might be less willing to offer five dollars because you weren't there or what if somebody you never saw before and you're never gonna see that person again which we call a one-shot game there's a good chance that you might offer lower a MA so you know so depending on the context of the situation or how the money came about it might change your behavior so at that many of you already heard of this that we have two distinctive ways to think system one and system two that's not the only dual processing theory there are many few dual processing theory such as holistic systematic process in the elaboration likelihood model so it's not an uncommon thing in psychology to of process TVs but these theories energy system wanders fast into the efficient habit base they help us to make decisions that are very quick and we don't have to exert a lot of effort making these decisions however if the system is always on and it's you know essentially non conscious you you're not evaluating every step of your decision sometimes the system is not able to help us support that decision or sometimes there is a factor that influences our decisions suggests it's a higher risk decision or is a high involvement decision in that case system to kids in this system is reflective controlled resource consuming good a complex decision tasks it's conscious and it's complex are the students as you can tell this system is very resource consuming so if possible we tend to avoid to use this system which means that the majority of our decisions are influenced or processes in the system one now what the behavioral economics criticism of market research the Christians come from the idea that the vast majority of consumer decisions are based on system one however a large bulk of market research is designed to evoke system two so for example when you ask somebody walk me through your laundry detergent purchase decision because you you were asking question if you come you feel compelled to comply with their request and then you got him oh yeah how do i buy laundry detergent what four foot total store and read the ingredients I see if they're like discounts or some but we know that a majority of those decisions that are low involvement they are not turned away perhaps your previous purchase is the best predictor of your neck Newton force communicator so that's one of the criticism why behavioral economics think that sometimes markets a market research can be misleading alright so what I'm gonna do now I'm gonna show you a few examples of rationality you as I mentioned you guys participate in a survey over fifty eight this concept until yesterday when I process the data so the the survey for the most part for behavioral experiments that I designed based on past research of my own research in order to influence our behavior and show how I but based on my choice architecture I can influence the outcome of your answers so I'm going to show a few I love this effect in querying any adjustment constant contact and free me effects temporal effects and hedonic editing and I'm going to talk about a few theories they help us to predict because behavior phenomena and behavior decision theory is not based on my just observing behavior we you know people that have studied at them we're gonna have a session in behavior phenomena with 30 or 40 minute you know you can get a PhD in this dessert so there is a lot of theories that help us to predict this kind of kind of behaviors we're not going to be talking today about merges because of time constraints but those are the kind of thing that I was given example how governments are trying to influence better behavior or helping people it healthier exercise more or save more energy so let me give you a few examples the idea of anchoring and adjustment is that your judgments might be influenced by numbers that to which you are exposed and also you could think of these are first impression bias you get a first impression about somebody you meet and then it's harder for you to adjust that as you get more information about networks and these are people are unaware of the influence of any occur so as an in I did a series of experiments where I randomly assign you guys to one of two conditions and all the order of the choices have all been randomized so there is nothing here that could be systematically leading to the two turns out except the theories behind such behaviors so for about half of you I say what's the population of the turkey and I ask that to everybody but for half of you I said like is it larger or smaller than 300,000 and for the other half I asked is it larger or smaller than 300 million people okay and then asked what so estimate of turkeys population so I presented with an anchor that was either low or high and you were asked to estimate the population so I was kind of results here the people that were in the low anchor they estimated that the average population of Turkey was 41 million people and the participants in the high anchor condition they estimate that an average is 85 million people the actual populations that made it around 82 million people so but see they and this is statistically significant or with me even with a very small sample I'm trying to show that this was not driven because of like one of a few of lives so I have be able to consistently influence your estimation by just present with a low or high angle I also did that by asking a question about Mahatma Gandhi so I asked if Mahatma Gandhi died before or after the age of either 140 years old or nine years old both are very extreme estimates I find it hard to believe that anybody believed them Mahatma Gandhi died one is 140 140 or when he was nine years old and then I asked you to estimate the what age madman that he was when he died well he was 78 when he was assassinated so the people in the low anchor when I asked if they were before or after 9 years old they estimate that his age was sixty eight point three and the people in the high anchor they were the estimate was at 82 again a significant difference in vs me I was asking exactly the same question to everybody but I offered an Amber that influence in poo the judgments in that direction that was you know either lower or a higher so also why don't you show you an example in a marketing context so I said imagine went to the grocery store and you saw a sign stating the phone so for half a pill I presented this information by snicker bars for your freezer I didn't provide any anger for the other half I influenced 10 I try to influence your behavior by saying light the sign said by 18 snicker bars for your fries so as yes something like a Snickers bar snicker bars how many punches so the results show that my store signage worked because those in the no ankle condition on average bought two point four bars but those on the high anger condition when I said buy 18 bars on average they bought 5.5 four bars so it's not like just you know in judgment that might be relevant like guessing the age of again D but something with a potential implication for marketing behavior I can try to do something more extreme adjust after I cast half of USA think about a number one and ten and write his number in the box below and the other half I asked to think about a number between nine thousand nine hundred ninety one ten thousand and write it down and then I asked how many African countries are member of the United Nations so on the lower anchor condition the estimate was 18.8 the actual number is 54 and in the high anchor condition the estimate was twenty four point four so even something that was related to the question right the previous Emperor they were kind of related to the question was able to influence behavior so another example I give you this example of our software that's being developed by University and I wanted to estimate won't be a fair price for a small business very mean to half of the participants in the lower anchor condition I offer this price yet well I want to actually know that the personal version cost just much and the family version cost as much but in the high anchor condition I added a high-end core Enterprise version that cost 299 m so when I got your estimate of the expected price so the people in the lower anchor conditions thought on average the price for a small business should be $75 and the people in the high anchor condition that had the enterprise information they estimated the price being 124 dollars so by providing the anchor we have this innate need to compare things to something and that's what often drives these types of types of results all right so implications for marketing you can create anchors you know they're really big behavior if you have a markdown price you can increase the per customer limit you know say Oh only only 20 units which if you wanna sell more units it's not limited because nobody would buy that number of units for that product but it could potentially influence the amount that they buy and a product that has multiple use a high high end for example implications for consumer research and insights the magnitude of values may influence response anchor effect the answers to questions may be anchored to priorities and values kind of a cockade over in fact Disney happen regardless of the meaningfulness of the anchor just like I asked you to write down a large or a small number all right let me show you some other examples of such what we'll call the irrational behavior about the fashion personality so I'm going to give you an example of what's called contextual decoys imagine a situation where you are deciding between two two wines one that has a rating of points and cost to $12 69 cents and one day is rated at 73 points and is six dollars and sixty nine cents here's what the rating scales mean based on my some research that I did online so let's see it under this situation 70 percent of the people exposed to this context here then choose the 87 points are at 12 69 well now imagine that your context change and I add another bottle that is 67 points which is lower than this one and also is more expensive this is considered decoy because this alternative is completely dominated and should not influence the decision between these two bottles here because it's low in quality and is higher in price than this option but what often happens that shifts the perception of this alternative view and then I'm not saying it's gonna increase by 20% but I'm trying to give you a sense of the direction of the change so the choice proportion for these bought of wine here mining tree increased as a function of this completely irrelevant option in the set so that's why it would be if this happen you'd be considered irrational behavior well what could also happen is that I would have an option that's on the higher end but this option is still below the level quality for the relative to this option here and it's more expensive so it's again a completely dominated option and should not influence your choice however what we see often times is their preference for desorption here increases as a function of this decoy what's interesting and also is that this effect may happen even if the options are not available imagine you went to Amazon's website you saw not wine but any other product you have fortunes that are meaningful in your options that are not meaningful even if Amazon said sold out not available it's likely to influence your perceptions of of this project so the question is why this happened so we are making a decision on a to attribute claim here okay so the first dimension it is price desirability so the cheaper bottom has a price that's more desirable and perceived quality the more expensive borrow has a less desirable price but has a higher desirability in terms of perceived quality so what happened when I introduced user to get desorption is less desirable in terms of price that's why I choose this to the left of this bottom in this lowering quality so that's why it's below the target bar well what happened is that the perception of just a bar of change because we think you put people you know things in scales so we judge things relatively and then what happen is that the perception of that bottle there are 73 points and 6.60 69 cents increases in terms of prices our ability perhaps quality and then now that option is not as far off in terms of perceived quality and is more desirable relative to this boring that's why choice proportions tend to change towards this opportunity the same is happening on the more expensive for the more expensive bottle where I introduce a decoy a completely dominated choice in terms of prices our ability in quality but that makes me a rearranged perceptually I put this product on my scale and then they change the perception this wine perhaps becomes more procedural higher quality and also more desirable in terms of price and that's why the choice proportion for these options for this option here tends to change so a famous story you might have heard that the economists used to have desorption where you could buy economies concepts curve subscription for $59 print and web subscription for $125 in a print subscription $425 this option is completely irrelevant is completely dominated by the other options so but what happened when I did I really did an experiment showing exactly this kind of same at the end when you have this decoy here the choice proportion for this option here the clinton web is 84% and for the economy subscription only is 60% however if you remove that decoy you can see a flip in the preferences now more people want to buy the economy own subscription and fewer people want to buy the premium of subscription which front revenue standpoint is not ideal for the economy which means that that decoy it is there and it's helping the economies so why is it irrational because that option is completely dominated and but is it helps you to make a decision by creating more standards of comparison alright I did something similar to you guys so I offer you a credit card that was either 1% or 2% cash back in cost 20 or $40 energy ideally I would have the chance to calibrate so we started 50/50 so and then in one condition I told people there there was a third option there was again a decoy because was point eight cash which is larimar version but it was more expensive and in another condition I said that there was a another credit card there was one point eight percent which is lower than the two percent and also it was more expensive than the forty dollars that you pay for the 2% on the 1.8% and $50 there was no there was no change in your presence so ideally I would have to calibrate this in order to be able to define what are the levels that I'm going to be able to make you shift your preferences but for the second decoy the cheaper decoy where is 0.8 percent at $30 there was some change and so about 40% of the people shift their preference from the 1% increase the desirability of the option there was one percent $20 so just using that data and using the knowledge of perceptual of decoys I was able to influence alright so another thing that I did I said I told you like well we got this wine from Brazil and they chose Brazil not only because I'm originally from Brazil but also because this is not known as a producer of wine so that will kind of take away your range of price distribution and then I said he's and then I asked you like okay so you got the god of wine you went to a store and you saw a section of Brazilian wines and in their section the price is over 0 - or the fall so in one condition water called ask you most prices were below your wine which was $15.99 in the right skew condition most prices were above the price of your one but what I was asking you to tell me is how we perceive the price the quality and the generosity of your friend for giving you a bottle that cost $15.99 so everybody was answering the question about $15.99 so so this data is not based on a big fan of decoys because they are not talking about the dominated alternatives here based on our duties range frequency theory and some of my research publish at you know consumer research so what so here's the prices then I ask you to judge the price the expensive needs the quality and the generosity of your friend so on the left you can dition where most of the prices were below the price of your target bottle you were much more likely to think that the wine was expensive then in the right SKU condition when most of the prices were high again why is there in rationality because $15.99 is $15.99 no matter what when I asked about Polly same thing in this condition here you are like that you think the wine was higher in quality and here you also you are thinking that your friend was more generous so range frequency theory has this name is because their range of the extremes of the distribution in the density of the distributions can affect the perception of prices you can see how you could you know if you're more profitable for you to sell this product here you would increase the density or extend the range of the options here because you know sorry I will do the opposite I would increase the density density here because $15.99 would be perceived as less expensive and I would sell more of that product so another thing another example is going to give in terms of rationality or you recognize the power of free I asked have a view you know if you saw lean truffles being so for first hands in her case for myself and other half if they're sold by for 3.3 cents and free from economic standpoint I'm lowering both price by once and so we shouldn't change the you know proportions of preference so what I saw is that when they are being sold for four cents in one cent eighty-one percent of participants chose the VIN travel however when they offer a three cent and free that the proper proportion of people they are now chose lint decreased by twenty two twenty two percent so the so you know the takeaway here like if you can offer something for free there to have like a big bump on your especially if you're offering a bundle another concept I want to talk to you about is this idea how our preferences across time consistent well they do have a hyperbolic discount is are like wow we're we're very patients in the short term but they're a lot more patient in the long term so well usually happen I didn't do this experiment but if I had done would be like would you try to have a hundred dollars right now or $100 $110 tomorrow $110 tomorrow because of this hyperbolic discount doesn't sound as much more however when I offer like would you rather have $100 100 days from today in a hundred ten 101 days from today usually people will pick the because here the discounting is slower and you don't see that you are losing too much by by waiting again classical economics assume you have well-established preferences I just showed a bunch of examples how I can influence your preferences based on the way I manipulate the context or the way I manipulate time perception I'm going to keep temperatures even in interest of time so in terms of implications for marking sales is dominant options decoys with features similar but in fear to the most profitable offering skew the distribution to influence an option to make you more attractive and whenever possible blend of products in a way that one of the products is for free in terms of application for consumer research and insights context may strongly influence judgment the skewness of distribution of X perception about stimuli and various stimuli dominance of alternatives in a survey such as decoy effect perception of value in choice alright just going to be at the last block of the presentation I also did a bunch of small experiments here where I for example I ask you this question for so you must be one of Julian's trusted choice a you have a 50% chance of winning $1,000 in a 50 chance of winning $0 choice B you have a hundred percent chance of winning $450 well which option has a greatest expected value if you believe in classical economics option we should be the one we choose because option Wei has an expected value of $500 which is a 50% chance of option B is has an expected value of $450 let's see how you behave not surprisingly based on what I know in terms of behavior theories the vast majority of you were workout called bomb bounded irrational or bounded rational or irrational and chose the risk-free option even though that option has a lower expected I say like okay let me flip the scenario here I say now instead of winning you have a 50% chance of losing a thousand in the 50% chance of losing zero dollars and charge do you have a hundred percent chance of losing four hundred and fifty dollars so remember right here you chose the risk-free option now in terms of expected value option B is better okay because your expected value for that is only losing 450 while in the honey you're expected balance losing $500 what happened here is that we saw a flip in the result right now people are willing to take a greater risk just to have a chance of not losing mine so what happened that were I just show like when we're making choice under risk we are risk averse in the domain of gangs and we're risk seeking in the domain of losses so here we were risk averse we went with the risk free option but here we were risk seeking just because we have such a loss aversion that were really to take a bigger loss the risking having to lose a thousand dollars just so I have a chance to lose nothing so that's another question I asked you about your trip and then you went to the hotel to a hotel when you decide to order food and the you and your friend there was two pizza places in that town they both had pizza a large pizza for 20 dollars in that 12 wings for $20 and but they had different promotions the first one was offering life by the large pizza for 20 bucks which is the food price and get the wings for 10 bucks which is discount the second picture here was Tommy get a pizza for 10 bucks discounted if you buy the wings for 20 dollars now my way what way you look at this it is you're paying $30 either way so for a classical rationale standpoint you should be indifferent I should get a 50/50 percent here distribution this scenarios based on my research on price discount in framing banana population so what happened is that you guys behaved as I predicted you prefer desorption here even though you're paying $30 either way I'm going to show you why you prefer that option the reason why you prefer the option is based on what called value function from prospectivity so just theory predicts that the our value function on the domain of gains is concave in this less is cheaper than the vastest so the pleasure that you get from a $10 game see this magnitude here this magnitude here the dash line is smaller than the pain of losing $10 so let's imagine scenario C the pizza was more or less at a reference price like zero there's no gain or no loss now in the wings I me the wings expensive so they would feel like a loss then so if I if I give the discount to the pizza what happened is that the perceived gain will be this much but if I give the discount to the wings the perceived avoided loss would be they're much larger and that's why people tend to prefer this option here you feel like you're you know it's less of a of a loss reference points the reference points are relative you know if I offer you if two companies offer you a contract that paid you sixty thousand dollars in the first year 70 in eighteen the second and the third or 80 thousand the first 70 in the second and 60 in the third most people will choose this increasing trend because you adapt to a reference point in 70 it's a game relative to the reference point in eighties again relative to set however the optimal decision here would be to choose offer B because of time value of money you are getting more money upfront but it's very hard for people to take that option and this also is idea of gains and losses in framing happens in the marketplace see this example here where this offer here is frame as 101 percent fat and this option here is frame as 99% fat free because fat heaven has acquired a negative value you know throughout the years this option is probably gonna be seen as more negative than this option here even though they are exactly the same same product let me just show one more example of the data I don't wanna go too much over my my time I asked you also that mr. M is the B bought a lottery ticket mister a won $100 the miss mr. B won $50 when mr. would a left the store he realized had partner illegal zone and received a $50 parking ticket as who is happier at the end of the day remember the bad McGinnis a states of wealth it's what matters they have the exactly state of wealth they should be equally happier but you guys expect mr. B the person who didn't lose anything would be happy and that's exactly because of what we're talking about that the pain of losing $50 is much larger than the value that you have of retaining $50 and that's why he hurts you let me see just kind of behaviors I can so he don't everything is this idea that we edit things in our minds to make us happier so for example mr. a instead of saying I won $100 and then I lost 50 a better way say like I won $50 today you make you feel better so think they help us so for example segregate multiple games so if you I bet he'll be happier if you have choose lottery tickets or $50 each then one ticket of $100 so segregate things I gave multiple losses that helped us to cope better with the pain or enjoy more against so as to a question I want to see hedonic editing from your data so I thought there was this part of whining but for 25 bucks and now it cost $75 and I study how would you tell yourself you're drinking when you drink that bottle what you know how do you edit this behavior in order to make ourselves feel better so the optimal rational answer would be $75 which is the price to replace that problem 19 out of 58 chose the option which is about 1/3 so but there's people like I it's 0 we spayed their account is closed so if you say I profit 50 bucks some people thought like it's 25 bucks plus interest and somebody like is 25 $25 so different people use different strategies to justify why they were drinking $75 so let me just go to my adviser can open for a question so just like any other tools that you have in your tool set no use behavior economics where necessary or understand how can influence your results okay or neuro marketing or whatever - but with that goal to solve a problem a research problem that we we flew insane this sheet so I use this analogy do not do the wall you know so you can use your shiny new power to don't create a research problem so you can use your neuroimaging or don't create a research problem so you can use your new behavior so some additional reading so these are very popular books and here's my contact information like to now open to questions thank you very much Marcus it was really interesting so we have three questions thus far first one this was back in reference to when you were talking about anchoring low versus high anchoring and he has this is Craig these are averages but what was the median or did you control for outliers so what I did was a quick clean so I took like a the two extreme bands okay the lowest and so for example somebody there estimated that turkey population was two billion people and somebody says that's 80 people right so I removed from this kind of data the two extremes however if I had outliers that would increase my standard error if a by increasing my standard error would make it harder for me to find statistically significant differences right because the outliers would in inflate less than air so the fact that I found is that this very significant difference in the world way below point zero five you're on point point zero one month so they mean that even if there were you know outliers they were not enough to to show me that there were true difference between those two groups mm-hm next question is from Praveen hi professor this is interesting thank you wonder why 1.8 percent of 50 dollars did not work as an effective decoy I think this was in reference to the banking yeah the cash back so I think because so there was already in a [Music] preference for higher rewards over price so I think I would have to calibrate more where line I think if I kept at 1.8 percent when increase the price to $90 okay that might have influenced I think that was just just not different perceptual difference between those two options that influenced their zone so because also you have to think welcomes of a percentage in price it's a concave curve right so the difference between 1 & 2 is usually perceived to be greater than the difference between 2 & 3 so a little bit of calibration and knowing better the audience would probably have helped and I would be able to move in both directions yeah more clarity there this question is from Brian he asked have you seen the dunning-kruger effect with any of these decision-making methodologies ie are some more effective when people who think they're more effective at decision making but aren't I haven't seen him but I also not magic expertise right so so because I'm randomly assigned people to conditions there is no reject to to believe that there was more expert in one condition than it is possible that the effect is going on but they give you the effect is not systematic in one condition versus the other so it's you know there is no reason to believe that would be however so in that extender in that sense that can only be seen as noise in the data which would make it harder to get the result however if I had measured expertise and had enough sample to test that I could test whether there you think mm-hmm but next question from Kevin yes have you for perhaps how do you foresee some of the aspects of loss aversion or conceptualization of free changing in postcode nineteen world also said very informative presentation great use of data collection thank you thank you I think right now we are so I think one thing that that the Kobe 19 has done is that has I would say we are probably using more system mu then system one because there's potential risk outcomes for our behavior think I'm think of myself I never thought about like wiping grocery stores or now putting the wiping grocery store bags are not putting grocery store bags on the counter and put on the floor so all this the fear in the potential risk of the outcome being negative yes heighten our perception which means were more likely using system to I don't know for for how long or you know it is just like a a temporally fact that's gonna happen for a short period of time but because of the potential risk and negative outcome that is likely making us I don't know morning have you seen yourself doing thinking about things you didn't used to think before oh yeah sure yeah I mean the grocery store is like a you're going into battle yeah so that means that we're more likely using more system to which means is more resource depleting which means your product more tired at the end of the day and my health issues right because you're constantly processing negative information when you usually would not do that yeah that's true next question is from Molly what is your take on the ethics of manipulating consumer behavior based on a rational decision-making so I think I mean marketers will try to do that to their benefit and and I think like educating consumer that SSI go to the at is that it's it's an important issue but I also also think in terms of policymaking because a lot of these things that do use their national levels to try to expensive like you know by brightening the lights around produce makes people consumer bridges you know by showing your behavior in terms of electricity consumption relatives your neighbors improve the consumption of energy so well like anything there's like you know a negative sign how we can treat consumers to behave in a way they will favor me but we have to not at the best interest in terms of well-being by the consumers but we can also influence consumers to behave in a way that improves their welfare mm-hm okay we may have time for one more question that was brilliant Marcus thank you can you speak more I'm more so on how often companies are using the metrics available in the market to shape their attainment of KPIs rather than the necessary tools to measure if they are attaining their KPIs ie crafting the message to meet the measurements that are more commonly available rather than actually measuring against their objectives is this an issue where have you seen opportunities for a pivot for major companies I think that's always an issue that's how incentives are aligned right so imagine I can nowadays brand managers are highly reward for short term results in terms of sales for what they do they're going to be like a lot more in sales promotion they're not ver tight because the effect of advertising is long-term I might not be here two years from now as a brand man however you know if every quarter I need my orden meet or exceed my sales goal I will be promoted or I'll get a new the new job right so what happened like well if their company does not align incentives so that people so people who behave in a way that will accomplish that goal they will be you know short side they're behaving you know the best way to provide the best result in the short term and that also speaks to the issue about I mean I'm sure all of you are in the insight you miss your mark in this er you pro have seen clients they have a a priori favored outcome right so they're they want to test two advertising campaigns but they already think campaign a is bad and then you're gonna collect the data hopefully supporting them if the data does not support that there's a glitch and they say no we need more research so that's calling in psychos that is called confirmatory hypothesis biased because you have a a priori hypothesis like just like the kpi's I will bias Li look for information that will support my hypothesis rather than thank you so much market so super interesting and we do have other questions that we don't have time for but is it true that you'll be around for some time yeah after the session okay perfect so everybody feel free to reach out to Marcus and get some more great information about his research because it was really great and I think all of us would be able to use this and think about bias going forward when we're crafting our research questions and thinking about our approaches so once again thank you so much for your time the Mayan have a good day bye