Research Report: A Data-Driven Approach To Understanding Your Morality
- Markus Over and Spencer Greenberg
- Oct 15
- 45 min read
Updated: Oct 16
Consider the following hypothetical scenario (warning: some will find it disturbing):
Poe lives alone with their dog for many years, when one day the dog runs onto the street and is lethally hit by a car. Poe, while heartbroken about losing their dog, has heard that dog meat is supposedly delicious, so decides to cook and eat it.
How immoral would you consider Poe's action in this situation? Some people think it's not at all immoral - others think it's moderately so, but not nearly as bad as many other things, whereas still others see it as seriously immoral. In fact, it was one of the most polarizing scenarios in a study we ran. You probably had an intuition right away about how immoral you felt Poe's actions were. But how, exactly, did you make that judgment?
Most of us struggle to explain what principles underlie our moral judgments. Our values and morality are very close to the core of our identities, and yet, we often have surprisingly little ability to explain what our moral compass truly looks like. So, at Clearer Thinking, we set out to make it easier to understand what drives your moral judgments. The result of our efforts can be found in a free tool that we've launched, called Understanding Your Morality, which you can find here:
This tool gives you personalized insights into your own moral judgments and explains how they relate to the moral judgments of others, based on fifteen key moral dimensions:
In the rest of this report, we'll explain our research and summarize the insights from our study of people's moral judgments. Read on, if you want to know:
How we came up with our Clearer Thinking framework of 15 moral dimensions
How our study participants judged the scenario about Poe's dog (and many others)
Why different people so often arrive at very different moral conclusions about the same event
What our findings reveal about the immoral behavior attributed to various celebrities
How moral judgments correlate with various demographic factors
How our 15-dimensional framework differs from Moral Foundations Theory
So, let's get right into it.
Key Takeaways
|
Why Morality is Hard to Understand
One reason it's hard to study people's moral compasses is that people often don't have a detailed understanding of their own moral reasoning. Let's look back at the scenario we mentioned at the beginning of this article, about Poe eating their dog after it was hit by a car. For many people (though not everyone), this scenario evokes a strong immediate reaction of wrongness, or even disgust. Yet, when asked about why they think it's wrong, many people struggle to put it into words.
Even when we can find a good explanation for why we think it is wrong, a crucial fact remains: the strong ethical judgment of the scenario, in many cases, happens so fast that we haven't had time to really think it through. There are lots of possible explanations for this. For example, maybe we're so used to making moral judgments that we've honed our moral pattern recognition to the point where we can make such judgments intuitively (like expert chess players making intuitive judgments about which moves to make, without consciously reasoning about them). Or maybe the reasons we think of, to explain why we judge the situation the way that we do, are simply post-hoc rationalizations – while they may be good reasons for feeling the way we do, they may not actually explain what drives our snap moral judgment. Whatever the truth of the matter is, it's clear that making a moral judgment doesn't require knowing why you made that judgment.
Another example of people struggling to truly understand their moral judgments can be seen when scenarios are carefully designed to seem immoral but avoid being subject to the usual reasons people would find them immoral: such cases can lead to what social psychologist Jonathan Haidt calls moral dumbfounding, where respondents are confident that the action was immoral but can't come up with any reasons for why.
Another tricky aspect of morality is that moral judgment and moral behavior don't always align. When you ask people about their judgments about a hypothetical situation, you may get very different results than when you actually put them in such a situation. For instance, in one of the famous "trolley problem" scenarios, people are asked if they would pull a lever to redirect a trolley to prevent it from killing five people, even though they know that doing so will cause the trolley to kill one other person who was previously safe. While they can answer one way when presented with a hypothetical (most say they would pull the lever), it's hard to know what they'd really do if they saw the trolley barreling forward, had limited time to make their decision, and were filled with the stress and adrenaline of the situation.
Our moral judgments generally seem to be made quickly and automatically, and even though we don't always act in accordance with them, they seem to be one powerful factor that can influence our actions. So we think it's reasonable to study moral judgments in hypothetical situations, as long as we keep in mind that those moral judgments won't always win out over other considerations in the real world.
There have been other attempts to formalize how moral judgments work in humans, the most well-known of which is Jonathan Haidt et al.'s Moral Foundations Theory. We decided to approach things from a different angle. Instead of starting with a theory, we developed a novel statistical method to understand moral judgments, collected data on people's moral judgments through a study, and applied this method to explore the insights the data itself could offer.
The Moral Judgment Study
To run our study and gather the data necessary to predict (and thereby better understand) people's moral judgments, we first came up with a list of 38 moral principles and factors that seemed like they had a shot at covering all the reasons someone might say an action is morally wrong. The goal was to cover as many moral principles and factors as we could find, to cast an extremely wide net. These included things like:
Dishonesty
Illegality
Causing someone physical pain
Disrespecting authority
Next, we created a set of 323 scenarios, each describing a behavior that some people might consider immoral (the scenario from the beginning of this article was one such example). Then, we needed to be able to judge which of the 38 moral principles were relevant to each of the scenarios. For example, here's another scenario:
Poe, a devoutly religious father, tells his eighteen-year-old daughter that she will bring shame onto herself and their entire family if she has sex before marriage and that she must remain abstinent until she has a husband.
Some moral principles that are highly relevant to this scenario might be Christianity, Impurity, or Social Taboos. Principles like Dishonesty or Illegality, however, have little, if any, relation to that scenario.
We decided to use two AIs – GPT-4o and Claude 3.5 – to create a big relevance matching table (using a prompt to evaluate how immoral each scenario is according to each of the 38 moral principles), and had a human decide in all cases of a more-than-slight disagreement between the two AIs (when the two AIs closely agreed, we just used the mean score they provided).
We used a different hand-crafted prompt for each of the 38 principles. For instance, this is the full prompt we used for the Abuse of Power principle:
Consider the following scenario where X takes an action. Evaluate the extent to which X, by taking this action, was abusing a position of power or authority on a scale from 0 (not abusing power) to 10 (completely abusing power). For example, if the scenario was: X, a team leader, assigns tasks to team members based on their strengths and workload capacity. You would respond with an abuse of power rating of 0, as assigning tasks in a fair and reasonable manner does not constitute abuse of power. If the scenario was: X, a supervisor, denies a qualified employee a promotion because of personal dislike rather than merit. You would respond with an abuse of power rating of 3, as using personal biases to make employment decisions may be considered a low to medium level abuse of power. If the scenario was: X, a politician, uses their influence to secure lucrative government contracts for a company owned by their family members. You would respond with an abuse of power rating of 7, as exploiting political position for personal gain constitutes a medium to high level abuse of power. If the scenario was: X, a CEO, embezzles company funds for personal use and threatens employees who attempt to expose the wrongdoing. You would respond with an abuse of power rating of 10, as embezzlement and intimidation of employees represent the highest level of abuse of power. If the scenario was: X poisons the beloved pet of their wealthy neighbor, Y, out of spite for Y's ostentatious behavior. You would respond with a rating of 0 because while this action is extremely immoral and causes significant emotional pain to Y, it does not involve X being in a position of power or authority over Y or Y's pet. Here is the scenario to analyze: <scenario description> Now reply in this format (with nothing else, JUST reply in this format) Explanation of rating: [your explanation goes here] Extent of abuse of power by X: [your 0 to 10 rating goes here] |
We then sent one such prompt for each principle-scenario combination to both GPT-4o and Claude 3.5. We asked the AIs for an explanation of their rating before revealing their rating in order to give them some "room to think", as we conducted this research before reasoning LLMs (which likely wouldn't need this) were available, and hence this approach seemed to yield higher-quality answers. I also allowed us to verify whether the reasoning behind their rating was sound.
Afterwards, we asked each of our 359 study participants to judge a different subset of around 130 of these scenarios, saying how immoral they thought the actions described were, on a scale from 0.0 (Not at all immoral) to 5.0 (Extremely immoral).
We examined each participant individually and fitted a personalized linear regression model to forecast that person's moral judgments. For each participant, their personalized model produced 38 coefficients - one per moral principle.
Notably, we added one constraint to the linear regression: we disallowed negative coefficients. The reason is that while it's sensible for someone to not care about a moral principle, we don't believe that real people invert moral principles, finding that violating that principle makes something more moral, the way a supervillain from a cartoon might (e.g., "the more suffering this causes, the more moral I think this action is!")
In practice, this means that we ran with an iterative least squares approach, where we trained the model for a given user, then dropped all the principles (for that user) that had negative or 0 coefficients, and trained the model again with only the remaining (positively weighted) principles. We repeated this until only positive coefficients remained.
Finally, having that trained personalized linear regression model for each study participant allowed us to do two things:
We could predict our participants' moral judgments of any scenario. For instance, if our AI ratings indicated that a scenario was high in emotional pain, and the participant's emotional pain coefficient was also high, then that would increase the score our model predicted the participant would give the scenario. We excluded a subset of their ratings from the training set so that we could use them to test the predictive capabilities of our personalized regression models.
This also allowed us to reevaluate our 38 moral principles and reduce them to only the most informative ones, which ultimately led us to our Clearer Thinking framework, with its 15 dimensions.
Identifying the 15 Moral Dimensions
As described above, we started our study with 38 moral principles. After we gathered all the data from our study participants, we reduced that set and ultimately identified a subset of 15 highly predictive moral dimensions.
Coming up with such models always entails many trade-offs, a major one being accuracy versus simplicity. Using all 38 principles would have been slightly more accurate overall, in the sense of making slightly better predictions. But it would also make the framework more complex and prone to overfitting. Reducing the principles to the 15 most predictive and easily interpretable dimensions seemed like a sweet spot, where predictive accuracy was minimally affected, and the model was still simple enough to allow us to derive meaningful conclusions for individual users.
We used three criteria to eliminate principles:
Vagueness. If a principle was vague, it was unclear how to interpret it or what it meant to score highly on it. Such principles were removed. One principle we dropped for this reason is the "Golden Rule" (treating others as one would wish to be treated). It may be a good heuristic to follow, but it's so broad and generic (since different people will disagree about how they wish to be treated) that it doesn't really help differentiate a person's moral intuitions from those of others.
Predictive accuracy. If including a principle caused little to no change in the predictive accuracy of the models compared to not including it, the principle was removed. This was, e.g., the case for the Kantian rule of treating another person as an end in themselves rather than a means to an end, which didn't improve the overall predictive accuracy of our models at all.
Prevalence. If the principle rarely had a coefficient above zero (i.e., it seemed to predict the moral judgments of very few people), it was removed. A principle that we dropped for this reason was Physical Pain: only a very small ratio of users benefited from this principle being part of their predictive model (above and beyond the other principles that were already included).
For the final selection of dimensions we obtained this way, we then evaluated the predictive performance across all our study participants. We computed a median R² of 0.47 (equivalent to a correlation of 0.68) and a mean R² of 0.38 (equivalent to a correlation of 0.60), indicating that the models were quite predictive and captured much of the variation in people's judgments.
These are the 15 dimensions we ended up selecting in our final set based on these criteria:
Abuse of Power: Being concerned about the misuse of authority, evaluating actions based on power exploitation.
Authority: Valuing respect for legitimate authority, prioritizing adherence to established roles and responsibilities.
Christianity: Valuing alignment with modern Christian teachings, evaluating actions against contemporary religious standards.
Dishonesty: Valuing truthfulness in all interactions, judging actions by their honesty and transparency.
Emotional Pain: You value protecting others from emotional suffering, evaluating actions based on the mental distress they cause.
Harm to Vulnerable: Caring about protecting vulnerable people, judging actions by their impact on those least able to defend themselves.
Impurity: Judging actions based on whether they are impure or degrading to the person committing them.
Inequality: Valuing equal treatment, evaluating actions based on the absence of discrimination or exploitation.
Loyalty: Valuing loyalty to people and groups, judging actions by how they honor or betray commitments.
Prejudice: Being sensitive to discrimination, evaluating actions based on bias against groups.
Property Rights: Valuing the respect of ownership rights, evaluating actions based on violations of what belongs to others.
Social Contract: Valuing societal cooperation, judging actions by commonly agreed-upon rational rules.
Social Taboos: Prioritizing adherence to social norms, evaluating actions based on their violation of deeply-held cultural or moral boundaries.
Unfairness: Values fairness in all situations, assessing whether actions are just and impartial.
Utilitarianism: Weighing overall outcomes, assessing whether actions create more total harm than benefit.
Note that the only religious principle used was Christianity, because our research was in the US, where Christianity is by far the most popular religion (as of 2023, Gallup polling indicates that 68% of US adults identify as Christian, 22% as not religious, and the next highest religion was 2%). We were not confident we could get large enough samples of other religions to draw meaningful conclusions compared to other demographics. Additionally, Christianity bundles together a number of moral ideas for those raised in it that then become correlated (e.g., humbleness, the ten commandments, forgiveness, piety, and so on). In other cultures, this would have to be changed to include other religions, of course.
Our New Tool: Understanding Your Morality
Running this study didn't just allow us to learn about our study participants. It was also the basis for a new, free, interactive tool that can help you better understand your own moral intuitions and learn how your morality compares to that of our wider audience.
At first, the tool follows a very similar process to our original study: you'll rate a set of scenarios (at least 45) by how moral or immoral you think they are. Based on this data, our tool then uses our linear regression approach to find the coefficients that best predict your moral intuitions. This way, you'll learn what your moral judgments say about how you weigh the 15 moral dimensions.
The tool will also show you how "compressible" your moral judgments are, within this model. That just means it will tell you how well your moral judgments can be captured by our 15-dimensional model. If the model predicts your answers accurately, it suggests that our 15-dimensional framework can tell you a lot about how you make moral judgments. If the predictions are less accurate, it may mean your moral judgments follow different principles or that they are less consistent than our model would need them to be to make accurate predictions.
Lastly, in addition to giving you these insights based on our new Clearer Thinking framework, we will provide you with personalized insights about Moral Foundation Theory – a different framework popular in academia that aims to capture the essence of people's morality by reducing them to a different (and even smaller) set of moral principles.
We hope that the information you get from using the tool will provide a basis not only for gaining new insights about yourself and your moral intuitions but also for reflecting on whether you endorse those principles upon reflection. After all, our moral intuitions are not set in stone; we may be able to deliberately shape and adjust them over time through regular reflection.
Try the free tool here:
What We Found
Nobody Relies on All 15 Dimensions
One might imagine that most people value all possible moral dimensions to different degrees. Most of us surely have some preference for fairness, honesty, preventing harm, and so on. Right? Well, not quite. Here are the ratios of how many users ended up with a coefficient greater than 0 for the 15 moral dimensions:
Even the most widespread dimension, Dishonesty, didn't have a positive coefficient for everyone: 14% of users had a coefficient of 0! And the other dimensions had even higher zero rates.
Does that mean people really don't value these dimensions at all? Well, not necessarily - the dimensions are not perfectly distinct from each other. So, you may care a bit about some dimension, but only to a degree that is sufficiently explained by other, related dimensions. For this reason, even a person who cares somewhat about Dishonesty may occasionally end up with a coefficient of 0. Still, the much lower ratios of many of the other dimensions indicate that, for some people, dimensions such as Social Taboos or Utilitarianism just don't contribute meaningfully to their moral compass.
Taking a different angle on the data, we can also ask what ratio of study participants had any given number of positive coefficients (out of 15, as there were 15 dimensions). For instance, a hypothetical user whose moral judgments were perfectly predicted by a single dimension - say, Prejudice - would end up with only one positive coefficient, as all others would end up being 0, since the corresponding other dimensions would add no additional predictive power for that person. People's moral intuitions tend to be more complex, though, so in practice, we see that most participants ended up with some medium-sized subset of the 15 dimensions in their personalized regression having positive coefficients:
Here, we see the number of dimensions on the x-axis, and the bar above each one indicates the percentage of study participants who had positive coefficients for that many dimensions out of 15. This chart tells us that:
For more than half of the people, 8 or fewer dimensions were sufficient for their personal linear regression model, meaning that the remaining dimensions added no further predictive power.
Around 90% of participants required no more than 10 dimensions.
Not a single study participant had more than 13 dimensions present in their regression model.
So, clearly, there is some variance in how many of our 15 dimensions are required to predict the moral judgments of different people. And yet it appears that not a single study participant in our study relied on all 15 dimensions strongly enough that they all ended up being predictive of their judgments.
Disagreements and Extremes
Earlier, we referenced the scenario around Poe's dog. In the study, it was phrased as follows:
Poe's dog was killed by a car in front of their house. Poe had heard that dog meat is delicious, so Poe cut up the dog's body, cooked it, and ate it for dinner.
How would you rate the immorality of this action, on a scale from 0.0 to 5.0?
In our study, this is one of the scenarios where we can see disagreement between study participants most clearly. Let's have a look at the distribution of ratings that people gave this scenario:
One interesting observation about it is that the two most common ratings for this scenario were 5.0 and 0.0 - the two opposing extremes. More than a quarter of respondents (26%) chose the highest immorality rating possible, while 8% chose the lowest possible rating, meaning they found this scenario to be not immoral at all.
Looking at the 15 moral dimensions, two that seem particularly relevant to this scenario are Impurity and Social Taboos. Would you expect these to be in any way predictive of what judgments people made for this scenario?
It turns out that these dimensions do explain the disagreement that we've observed above quite well: if we look only at users who have a positive coefficient for Impurity and Social Taboos, then we get an even higher ratio of 5.0 ratings (45%), and the ratings from the lower half of the spectrum are greatly reduced. But if we look only at users whose Impurity and Social Taboos coefficients are both 0 (meaning neither dimension is at all predictive for these users), we end up with a rather small sample size of only 15 study participants (as out of the 118 participants who rated this scenario, only 15 had both Impurity and Social Taboos coefficients of 0) - but yet, half of the original 0.0 ratings remain, while the 5.0 ratings have almost entirely disappeared:
This lends further empirical support for the moral dimensions being meaningful predictors, while also showing how strongly different groups of people can disagree in their moral judgment of such scenarios.
So, this single case already demonstrates several interesting insights that, as we'll see, show up in many of our scenarios:
Disagreement: There is enormous disagreement on many of the scenarios, with some people finding them extremely immoral while others tend to find them not the slightest bit concerning.
Extremes: Many users tend to choose the extreme values, 0.0 and 5.0, rather than the more nuanced values in between. In the above scenario, 35% of users chose the extremes. Across all scenarios, 26% of all judgments taken together were either 0.0 (8%) or 5.0 (18%).
Non-standard Distributions: It's quite common in studies where a large number of participants generates a distribution of numbers that you end up with something reminiscent of a (possibly skewed) normal distribution (also known as a bell curve because it's shaped like a bell) - in fact, the Prevalence of Positive Coefficient Counts chart we saw earlier is one example of this. As we see in our histograms for scenario judgments, though, many scenarios end up with a distribution of judgments that is very far from a bell curve (although there are a few exceptions). This suggests that it may be a mistake to assume human morality has a single, consensus answer that individuals will then deviate from in either direction. Instead, there's a plurality of very different views that can lead to entirely different conclusions on any given scenario.
Let's look at another scenario:
Poe borrows money from their friend Avery and, prior to getting the loan, claims they will pay Avery back within 30 days, even though Poe knows that realistically, they will only be able to pay the money back within 90 days, not 30.
Again, consider forming your own judgment about this scenario first, ranging from 0.0 (Poe's actions are entirely acceptable) to 5.0 (Poe's actions are maximally immoral).
Here's how immoral study participants thought it was:
This is one of the scenarios where people are less strongly divided and extreme judgments are less common: here, only 8% of participants selected the most extreme values (10 out of 130 chose 5.0, and only a single person chose 0.0). However, it raises another interesting consideration: given that the second-lowest value selected here was 1.2, it's plausible that the single 0.0 rating may have occurred by accident. There are several reasons why some numbers obtained through such a survey may not accurately reflect the participants' true beliefs:
People may sometimes just misclick without realizing
They may misinterpret the direction of the scale
Sometimes, they may just misread the scenario
Some rare participants may even take such surveys in bad faith and don't genuinely consider their answers at all
We aim to screen out any participants acting in bad faith when we can detect them, but it's not always possible. And the number of accidentally incorrect numbers is naturally hard to know. In fact, 77% of our scenarios had judgments that simultaneously included both extremes (0.0 and 5.0), and while for many of them this probably does reflect the true judgment of participants, in some cases, like the one above, there may be other causes (although we should be careful when making such assumptions – even this single outlier may be indeed the result of a person's honest reflection of the scenario).
Lastly, let's look at one more example:
Poe and Avery are platonic friends, but secretly, Poe often masturbates while imagining engaging in sexual acts with Avery.
This is how our study participants judged it:
This is one of the scenarios that study participants, on aggregate, found comparatively harmless. Still, 5% of respondents assigned the highest possible immorality rating, and even though 15% found it completely fine, 37.5% assigned a judgment from the upper half of the scale. So, even though this is far from the most controversial scenario we've had, the disagreement amongst the US population that took part in our study was still very large.
All these examples help illustrate that moral judgments are indeed highly heterogeneous, with a non-negligible percentage of people simultaneously judging the same scenario as either maximally harmless or maximally immoral. These findings give some credence to the model of people subscribing to different moral principles: if some people weigh, say, purity very highly, while others don't think it has any moral relevance, this would explain why, for certain scenarios, some people think of them as highly immoral, while others find them perfectly acceptable.
We Can Create Standardized Scales About the Average Perceived Immorality of Actions
While study participants disagreed about the relative importance of the different dimensions, the survey results allowed us to find representative examples (for our US audience) for average ratings from 0 to 5. We also surveyed participants on how to best match suitable descriptive terms to these six different levels of immorality, which informed our descriptors in the table below. For example, zooming in on Dishonesty, we can now point to concrete example scenarios that received the corresponding average ratings in our study, which allows different people to calibrate on what these numbers mean exactly.
We created similar standardized scales for the average perceived immorality of actions in the domains of Prejudice and Sexuality (the latter not being one of our 15 dimensions, but we still got useful results for representative scenarios within that area). Below is a table with one representative example per rating for the three domains. You can also find the full tables with more details in the appendix.
Rating | Descriptor | Example Scenario (Dishonesty) | Example Scenario (Sexuality) | Example Scenario (Prejudice) |
0 | Fine | X, a grandparent, tells Y, their grandchild, that they have the most beautiful smile they've ever seen, even though Y has a missing tooth. | X and Y are romantic partners, and they engage in consensual sexual activity. | X, a manager, in an attempt to avoid subconscious bias from impacting their hiring, has all applicant names and photos removed from resumes before evaluating those resumes. |
1 | Unsure, Lean Fine | X tells their child, Y, that their pet ran away even though the pet actually died because X thinks that it will be easier for Y to handle emotionally. | X flirts with their co-worker, Y, who is in a position of equal power to them, but after Y responds with body language that indicates a lack of interest, X permanently stops flirting with Y and, from then on, acts just like a co-worker. | X, a supervisor, provides flexible work arrangements to Y, an employee from a historically marginalized age group, to accommodate different life stages. |
2 | Unsure, Lean Unethical | X, a student, tells another student, Y, that they couldn't find the notes for the lecture, even though X does have the notes, to encourage Y to take their own notes. | X and Y are close platonic friends, and X gives Y a long and deep hug to say hello, but it is a longer and deeper hug than Y desires, and the hug makes Y feel uncomfortable. | X, a manager, assumes that Y, an employee of a religion that is uncommon in that area, celebrates the most common holiday in that area. |
3 | Unethical | X, a gamer, cheats during a casual online game and denies it when confronted by Y, another player. | X makes a highly sexual joke in front of a person they manage at work, Y, though the joke is not about Y. | X, a customer service representative, unknowingly provides inferior service to Y, a customer, because of Y's gender, age, or race. |
4 | Very Unethical | X takes credit for one of their work colleague's ideas during a business meeting to their boss, pretending it was their own original concept even though they know their colleague came up with it. | X and Y are having consensual sexual activity, and X asks Y if they can do a different sexual activity that they had never engaged in before. Y says no, but X does it anyway. | X, a police officer, conducts biased profiling, targeting Y, an individual, because of their gender, age, or race. |
5 | Extremely Unethical | X engages in a one-year-long extramarital affair and repeatedly lies to their spouse about their whereabouts and activities to conceal their infidelity. | X slips a drug into the drink of their new acquaintance, Y. And then once Y is so intoxicated as to no longer be capable of consenting, X engages in sexual activity with Y. | X calls the police and falsely accuses their neighbor Y of a crime because X hates all people who look like Y. |
One advantage of such standardized scales is that they enable people to communicate how unethical some person's behavior was, without having to mention any details about their actions. Suppose you want to warn others about a person, but you don't want to reveal what exactly happened, e.g., in order to protect the victim, or because you have been sworn to secrecy. In that case, you could point to such a scale and mention that their behavior would rank, say, somewhere around 4.
Standardized scales can also be used to quantify the perceived severity of actions in a more objective way, relative to a given population (in our case, our study participants from the US). Consider, for instance, cases where celebrities are implicated in scandals involving sexual behavior.
A Scale of Public Opinion About Celebrity Immorality
Public judgments are often colored by people's feelings about the individual, which makes evaluation tricky. To address this, we anonymized reported actions that several well-known figures were alleged to have engaged in and asked participants to rate the immorality of those actions using our standardized Sexuality Immorality Scale - under the assumption that the allegations were entirely true.
In one example, accusations of Kevin Spacey sexually assaulting someone in a bar while intoxicated were rated (after anonymization), on average, as 4.2 / 5 on the scale - likely a much more objective measure than you might get when asking people about this case directly, because their views on Kevin Spacey himself might influence their opinion.
Below is an overview that includes some other high-profile cases and how a US audience rated their alleged actions after anonymization:
Celebrity Immorality Scale
This chart shows the average immorality rating scores given by Americans on a 0 (not at all unethical) to 5 (extremely unethical) scale when participants were provided with neutral descriptions of real accusations made against each celebrity. Study participants were not told who the accused was and were asked to assume for the sake of their rating that the accusation was entirely accurate. So these ratings only reflect the perceived immorality of the behavior described in each accusation if the accusation is completely factually accurate. We make no claim about the factual accuracy of the accusations.
Extremely Unethical | ||
Harvey Weinstein average rating: 4.9 | Accusation:HW is male in his 60s and a renowned film producer, and EC, a female in her 30s who is a model and actress. During a film festival event, HW unexpectedly arrived at EC's hotel room. There, HW coerced EC into performing oral sex and subsequently raped her. | Legal result:Convicted of rape in two cases, one of which was thrown out due to the trial judge "unfairly allowing testimony" against Weinstein based on allegations that weren’t part of the case. |
Bill Cosby average rating: 4.8 | Accusation:BC, a male in his 60s, renowned as a comedian, who administered unknown pills to AC, a female in her 30s, a national team basketball player. These pills rendered her semiconscious and immobile. Subsequently, BC touched AC's breasts and crotch and placed AC's hands on his penis. Upon regaining consciousness several hours later, AC discovered her clothing scattered around the room. | Legal result:Cosby was convicted but the conviction was overruled because "an agreement with a previous prosecutor, Bruce Castor, prevented Cosby from being charged in the case." However, in 2023 there were 9 women who filed sexual assault allegations against Cosby. Cosby also lost a civil suit alleging sexual assault. |
Vin Diesel average rating: 4.6 | Accusation:VD, a renowned male actor in his 50s, who entered the hotel suite of his female assistant, AJ, in her 30s against her wishes. VD proceeded to grope AJ's breasts and kiss her chest after taking her to the bed. He then coerced AJ into touching his private parts and masturbated in front of her. AJ was subsequently terminated from her position hours later. | His attorney's statement:"[he] denies generally and specifically, each and every allegation" |
Very Unethical | ||
Louis CK average rating: 4.2 | Accusation:LC, a male comedian in his 30s, who invited JW and DM, both female comedians in their 30s, to his hotel room during a comedy festival. Upon their arrival, LC asked if he could expose his penis. Initially perceived as a joke, JW and DM laughed it off. However, LC proceeded to undress completely, exposing himself and masturbating. When JW and DM attempted to leave, LC blocked the door with his body until he finished masturbating, then allowed them to exit the room. | His response: "These stories are true. At the time, I said to myself that what I did was okay because I never showed a woman my dick without asking first, which is also true. But what I learned later in life, too late, is that when you have power over another person, asking them to look at your dick isn’t a question. It’s a predicament for them." |
Nigel Lythgoe average rating: 4.2 | Accusation:NL, a renowned TV show producer in his 70s, engaged in inappropriate behavior with PA, a famous singer in her 60s, with whom he had no prior romantic involvement. In the first incident, NL groped PA's breasts and genitals and attempted to kiss her in an elevator, but PA left when the doors opened. The second incident took place during a dinner at NL's house, where he tried to kiss PA, who pushed him away and promptly departed. | His attorney's statement:"Lythgoe did not harass, bully, or sexually abuse PA," her claims are "pure fiction" and the "worst form of character assassination." |
Kevin Spacey average rating: 4.2 | Accusation:KS, a renowned movie star in his 40s, and TM, a film director in his 30s, were at a bar together. While TM was ordering a drink, KS approached him, forcefully putting his arm around TM and grabbing TM's crotch. TM removed KS's hand and went to the bathroom. KS, seemingly intoxicated, followed TM, making inappropriate remarks. TM pushed KS away and left the bar. | Response from a representative:"Kevin Spacey is taking the time necessary to seek evaluation and treatment." |
Bill O'Reilly average rating: 4.1 | Accusation:BR, a male TV show host in his 50s, holds a position of authority over AM, a female TV producer in her 30s. During dinner, BR made inappropriate remarks about vibrators and suggested he needed a younger lover, while touching himself. He then made a suggestive comment when AM left the table. In subsequent instances, BR initiated calls with sexual content, including masturbating and discussing sexual items. He also made unwelcome advances, including proposing a threesome with AM and her friend. Despite AM's objections and reminders of their professional relationship, BR persisted in making lewd remarks and pressured AM into engaging in sexual conversations and activities. | Reporting on his response:He "emphatically denied the claims", insisting that "his wealth and fame make him a target for such accusations". And he "denounced the press that has reported on the allegations -- and the ensuing settlements -- as corrupt and desperate to take him down" (source: CNN) |
Jonah Hill average rating: 4.0 | Accusation:JH is a male actor in his 20s, and he met AN, a female 16-year-old actress, during a house party. When they were drinking, JH made a joke about AN's legal drinking age, "Can you even drink this?" At one point, AN wanted a cigarette, and JH said that he had one in his car outside. JH and AN went outside to grab the cigarette. JH grabbed the cigarette from his car but didn't give it to AN. When they both walked back to the house, AN asked for the cigarette. JH said nothing and slammed AN to the door and stuck his tongue forcefully into AN's mouth. AN pushed JH off and ran inside the house. | His attorney's statement:"[AN's] story is a complete fabrication." |
A.Schwarzenegger average rating: 4.0 | Accusation:AS, a famous male movie star in his 30s, met Y during a work meeting. Y is a female movie studio secretary in her 30s. After the meeting, AS slipped his hand under Y's skirt and grabbed her buttocks, and said, "You have a very nice ass; I'd love to work you out." | His response to general accusations about his behavior:"My reaction in the beginning, I was kind of defensive...Today, I can look at it and kind of say, it doesn't really matter what time it is. If it's the Muscle Beach days or 40 years ago, or today, that this was wrong...It was bullshit...Forget all the excuses, it was wrong." |
Morgan Freeman average rating: 3.7 | Accusation:MF, a famous male actor in his 70s, worked together with Y, a female production assistant. During the work, MF comments about Y's figure and clothing; MF also rests his hand on her lower back and rubs her lower back. In one incident, MF kept trying to lift Y's skirt and asking if Y was wearing underwear. | His response:"Anyone who knows me or has worked with me knows I am not someone who would intentionally offend or knowingly make anyone feel uneasy. I apologize to anyone who felt uncomfortable or disrespected — that was never my intent...I did not create unsafe work environments. I did not assault women. I did not offer employment or advancement in exchange for sex. Any suggestion that I did so is completely false." |
Aziz Ansari average rating: 3.6 | Accusation:During a date night, AA, a male actor and comedian in his 30s, engaged in sexual advances with G, a female photographer in her 20s. AA kissed G, touched her breast, undressed both himself and G, and attempted to initiate sexual intercourse multiple times despite G's expressed discomfort. G resisted and expressed her reluctance, but AA persisted, pressuring her to engage in various sexual acts, including oral sex. Despite G's objections, AA continued to push for sexual activity, leading G to feel pressured and uncomfortable. Eventually, G left AA's apartment feeling upset and tearful. | His response:"There's times I felt scared. There's times I felt humiliated. There's times I felt embarrassed...And ultimately, I just felt terrible. That this person felt this way...[while it was] completely consensual" "[I] took her words to heart and responded privately after taking the time to process what she had said." |
Study participants saw only the accusation (with names being replaced with "X" or "Y") and were asked to rate the described actions' immorality. Study participants did not know that these actions referred to concrete cases that (allegedly) happened, nor who these actions were about.
It is, of course, not possible to be certain that all of the accusations here are true - while some have resulted in a legal conviction, or were admitted to by the perpetrator, others have been denied or partially denied by the accused. We make no claim about the factual accuracy of the accusations. The ratings apply to the behavior of these celebrities only insofar as the accusations are accurate.
How Our 15 Dimension Framework Compares to Moral Foundations Theory
Moral Foundations Theory (MFT) is another framework that aims to explain people's moral intuitions based on a set of core foundations. The framework, spearheaded by Jonathan Haidt and others, originally proposed five foundations: Care, Fairness, Loyalty, Authority, and Sanctity. Later, Liberty was added as a sixth principle, and Fairness was eventually split into Equality and Proportionality to better differentiate between "equal outcomes" and "outcomes based on merit", which can be viewed as two distinct forms of fairness. Thus, the original five foundations had grown to seven. With Honor and Ownership, two other principles have been proposed as contenders for being raised to "foundationhood", but as MFT is a dynamic framework, the debate around this is ongoing.
In our tool and research, we didn't only work with the Clearer Thinking 15 Dimension framework but also with MFT. This way, users of our new tool get insights into MFT as well, plus we are able to compare the two frameworks and even test some of the empirical claims that were made by developers of MFT. For all this work, we relied on the following seven MFT foundations: Care, Equality, Loyalty, Authority, Ownership, Sanctity, and Liberty. The reason we didn't include Proportionality and Honor was that our 323 moral scenarios were not well-suited to distinguish them sufficiently.
One big difference between these two frameworks is how they were created: while our framework is entirely empirical (and based on US data only), the MFT framework is grounded in theory (and is designed to be applied internationally). Also, we're pursuing different goals: we strive for the accuracy of predictions of people's judgments in the US, whereas MFT aims to provide a theoretical underpinning to morality across cultures. So, it was not our intention to come up with a "better" framework than MFT. Instead, we wanted to cover somewhat different use cases, such as prediction and empirical verification, and hence went with our distinct approach.
So, what are the empirical claims around MFT that we tested?
One of the most widespread claims about MFT is that the political orientation of people in the US predicts which moral foundations they care about: The theory is that people from the political left care much more about Care and Equality than about the other foundations, while conservatives, on the other hand, are said to value the remaining foundations more strongly, on average. And indeed, our data, based on our methodology, shows this to be pretty accurate. To test this, we used our system for predicting each participant's judgments, using the MFT foundations rather than our 15 moral dimensions. Then we looked at the correlation between these foundation coefficients and political progressivism:
In our study, we found that Care, Equality, and Liberty were quite strong predictors of political progressivism, with the other foundations being weakly correlated with and predictive of political conservatism. Liberty/Oppression is a bit of a surprise here, but previous MFT research helps make sense of this data point: they found that both progressives and conservatives value liberty but emphasize different aspects: conservatives focus on freedom from government interference, while progressives emphasize protecting vulnerable groups from domination. Our correlation findings might reflect that our liberty-related scenarios were weighted more toward progressive interpretations of liberty rather than conservative ones, suggesting an opportunity for future research to capture both aspects more fully.
Demographic Findings
In addition to checking how the political spectrum relates to the dimensions of MFT, we also gathered all kinds of demographic and psychological data and checked how these relate to the different dimensions in both frameworks, MFT and our own.
For this, we checked the general correlations between certain demographic and psychological markers and the moral dimensions and, in some cases, trained additional linear regression models to predict people's traits based on the moral dimensions. These linear regression models then gave us coefficients, which, similar to the table shared above, allow us to see how a given moral dimension affects the trait while statistically holding all other dimensions constant.
Political Ideology and Clearer Thinking's 15 Dimensions
Our Clearer Thinking framework revealed several strong correlations with political conservatism:
Christianity showed the strongest positive correlation with conservatism (r = 0.39), indicating that the scenarios our model associated with Christian values are strongly linked to conservative identity
The Prejudice dimension showed a strong negative correlation with conservatism (r = -0.31), suggesting that progressives were more likely to consider discrimination and bias when making moral judgments
Emotional Pain and Utilitarianism both showed notable negative correlations with conservatism (r = -0.28), indicating conservatives were less likely to base moral judgments on emotional suffering or on calculating the greatest happiness for the greatest number
Social Contract and Unfairness were less emphasized by conservatives (r = -0.24 and -0.20, respectively)
The table below shows the correlations between political progressivism and the 15 Dimensions of Moral Judgment by Clearer Thinking:
Feminist Identity and Moral Dimensions
The following sections explore correlations between moral dimensions and other demographic factors (e.g., feminist identity, religious affiliation, urban/rural living, personality, etc.) These are exploratory findings, so please note that they have a higher risk of being false positives than some of our other results, especially since many hypotheses were tested for this section.
Self-identified feminists (n = 96) showed:
Moderately strong positive correlations with CT's Emotional Pain (r = 0.23) and CT's Harm to Vulnerable (r = 0.22), highlighting an emphasis on empathy for negative emotions and protecting those who are less able to protect themselves
Positive correlations with CT's Utilitarianism (r = 0.18) and CT's Unfairness (r = 0.19), indicating a tendency towards considering the extent to which actions cause suffering/reduce happiness and equity in resource distribution
Negative correlation with CT's Impurity (r = -0.20)
Negative correlations with CT's Loyalty (r = -0.17) and CT's Christianity (r = -0.15), suggesting a potentially critical stance on religious/patriarchal structures and rejecting group conformity pressures
Religious Identity and Moral Dimensions
Self-identified Christians (n = 167; Note: 'Christians' here refers to a demographic group and is different from CT's Christianity moral dimension) showed negative correlations with CT's Utilitarianism (-0.17) and CT's Emotional Pain (r = -0.16), and a positive correlation with CT's Christianity (r = 0.28) and CT's Impurity (r = 0.26)
Self-identified atheists (n = 53) showed a positive correlation with MFT's Equality (r = 0.27)
People identifying as spiritual but not religious (n = 49) showed a positive correlation with CT's Abuse of Power (r = 0.19)
Big Five Personality Traits
Openness to experience correlated negatively with MFT's Sanctity/Degradation (r = -0.19), meaning people who are higher in Openness were less likely to make moral judgments based on concepts of purity or disgust
Extraversion correlated positively with CT's Authority (r = 0.15)
Emotional stability (i.e., low neuroticism) correlated negatively with CT's Emotional Pain (r = -0.17), possibly because emotionally stable people experience less emotional distress and may be less attuned to it in others
Agreeableness correlated negatively with MFT's Authority/Subversion (r = -0.18) and CT's Abuse of Power (r = -0.15)
Economic Trust Games
Giving behavior in the Trust Game correlated positively with CT's Inequality (r = 0.16) and CT's Christianity (r = 0.15), and negatively with CT's Loyalty (r = -0.16). In this hypothetical game, participants proposed splitting a theoretical $100 with an anonymous partner. The partner could accept or reject the offer (if rejected, neither received money). Then, the amount participants offered was doubled by researchers, and the partner then decided how much, if any, to return. Our 'Giving behavior' metric refers to the share of the initial $100 that participants allocated to their partner. Note that no real money was actually exchanged.
Other Findings
Self-identified environmentalists (n = 80) showed positive correlations with MFT's Equality (r = 0.18), MFT's Care/Harm (r = 0.15)
Living in more urban areas showed positive correlations with MFT's Care/Harm (r = 0.17), CT's Emotional Pain (r = 0.19), and CT's Utilitarianism (r = 0.15) — suggesting city residents may place a higher priority on empathy and considering others' suffering compared to rural residents
Higher donations to charity in the past 12 months showed a positive correlation with CT's Impurity (r = 0.19)
A higher number of visits to churches, temples, or other places of worship in the past 30 days (excluding weddings and funerals) showed a positive correlation with MFT's Sanctity/Degradation (r = 0.18) and CT's Impurity (r = 0.20)
Socioeconomic status correlated negatively with CT's Prejudice (r = -0.17)
Education also correlated negatively with CT's Prejudice (r = -0.17)
Income and gender showed relatively weak correlations (r < 0.15) with all moral dimensions, suggesting that these factors may play a smaller role in shaping moral judgments
Note: These findings are based on a sample size of n=359. We focused on correlations of r ≥ 0.15, a threshold that ensures a p-value of less than 0.01 – indicating that, for each result considered individually, a result at least this extreme would be obtained less than 1% of the time if there actually was no correlation. Given the multiple comparisons being made (and the lack of other adjustments to account for this), this more conservative approach increases the robustness of our conclusions – even though, as mentioned, this section is more exploratory and may contain false positives.
Why This Might Matter (for You and Society)
Morality lies at the heart of how we judge others and how we act ourselves. Yet, it's difficult to measure how we make those judgments or to articulate what underlies our judgments. Our new model of 15 moral dimensions helps bring structure to what would otherwise be a messy process. It offers a concrete, data-driven framework that transforms moral judgment from a matter of intuition into one of reflection and understanding.
But why should this matter?
On a personal level, understanding your own moral compass can lead to surprising insights. The 15 moral dimensions identified in our framework offer a mirror to reflect on the moral principles you do and don't endorse. Maybe you care more about loyalty than you thought. Or perhaps your moral reactions are strongly driven by concerns about inequality or emotional pain. By making these patterns explicit, the Understanding Your Morality tool helps you become more self-aware. Not just about what you judge, but what may drive those judgments. This kind of reflection can lead to more consistent, intentional moral decisions in everyday life.
We also hope that research like this can help bridge divides. One of the biggest sources of conflict – especially across political, cultural, or religious lines – is that people often talk past each other when it comes to morality. We assume others are simply wrong or misguided when, in fact, they may be prioritizing entirely different moral dimensions. For example, someone who focuses heavily on authority and fairness may come to very different conclusions than someone who emphasizes liberty and harm to the vulnerable. Neither is necessarily "less moral" in an objective sense; they're just drawing on different internal frameworks. By quantifying and comparing these moral profiles, we can better understand where others are coming from and possibly find more empathy and common ground in the process.
On a societal level, this work aims to provide a new approach to studying morality: one that is empirical, flexible, and highly personalized. Unlike traditional frameworks that start from theory and generalize outward, our approach began with actual human judgments and built the framework from the ground up, based on empirical data. This shift makes it possible to analyze how morality varies across different populations, such as political groups, demographics, religious affiliations, or personality types. It also lays the groundwork for more rigorous cross-cultural research in moral psychology by applying these ideas in different countries: instead of assuming a one-size-fits-all model of ethics, we can explore which dimensions hold up across cultures and which are more culturally specific. Our research so far has been focused on a US audience, but we hope that the methodology will be used in other cultures and contexts as well. If you'd like to use this approach in your own research, just reach out – we're happy to provide you with code to apply the method.
What Should We Take Away From All This?
At Clearer Thinking, helping people make sense of their own judgments and behavior has long been one of our key interests. We've created a number of tools aimed at this – from Understanding Your Philosophical Beliefs, to The Intrinsic Values Test, to Uncover Your Guiding Principles. Understanding Your Morality is our latest step in this direction, and we hope it helps people see their own ethical intuitions in a new light.
We'd love to hear your thoughts – on the tool, on the research, or on the ideas and questions it raises. And if you're a researcher yourself, we hope you'll consider building on our work, such as by applying it to new cultures, new questions, or even new kinds of minds.
Appendix: Full Standardized Scales
Here's the full standardized scale for dishonesty:
Dishonesty Immorality Scale | ||
X, a deceitful online scammer, fakes their identity when speaking to an elderly person, Y, in order to steal all of Y's life savings. | 5 Extremely Unethical | X engages in a one-year-long extramarital affair and repeatedly lies to their spouse about their whereabouts and activities to conceal their infidelity. |
X, a contractor, accepts payment from Y for contract work that Y would like done, even though X does not plan on providing the services, and X stops responding to Y as soon as X receives the payment. | X, a social scientist, completely falsifies research data in order to publish a paper in a top academic journal. | |
X, a salesperson, sells a faulty product to Y, claiming it's the latest and most reliable version, even though X knows the product won't function properly. | 4 Very Unethical | X takes credit for one of their work colleague's ideas during a business meeting to their boss, pretending it was their own original concept even though they know their colleague came up with it. |
X, a politician, makes promises to voters in a public speech, even though X knows they won't possibly be able to fulfill those promises if they are elected. | X, a student, secretly copies Y's math homework and hands in the work as though X had completed it on their own. | |
X, a gamer, cheats during a casual online game and denies it when confronted by Y, another player. | 3 Unethical | X borrows money from their friend Y and, prior to getting the loan, claims they will pay Y back within 30 days, even though X knows that realistically, they will only be able to pay the money back within 90 days, not 30. |
X claims they haven't seen their sibling Y's missing item when asked, even though they have seen it because X is annoyed at Y for something that happened earlier in the day. | X, a party host, is later confronted by a friend, Y, about why Y wasn't invited. X pretends to have forgotten to invite Y when, in reality, X purposely didn't invite Y. | |
X, a colleague, pretends to be busy when their friend Y needs help because X isn't in the mood to help. | 2 Unsure, Lean Unethical | X tells their friend Y that they're in a taxi on their way to meet Y, even though X hasn't actually left their house yet, because X is embarrassed about how late they are running to meet Y. |
X tells their child, Y, that the store ran out of their favorite snack, even though X actually just forgot to buy it. | X, a student, tells another student, Y, that they couldn't find the notes for the lecture, even though X does have the notes, to encourage Y to take their own notes. | |
X, a student, claims to Y, a classmate, that they studied all night for the exam, but in reality, they barely reviewed the material. | 1 Unsure, Lean Fine | X claims to their friend Y that X can bench press 200 pounds when, in reality, X can only bench press 150 pounds. |
X agrees with their co-worker Y's suggestion during a meeting (in an attempt to maintain harmony and avoid conflict) even though X has some doubts about the suggestions. | X tells their child, Y, that their pet ran away even though the pet actually died because X thinks that it will be easier for Y to handle emotionally. | |
X tells their partner Y that they look great in an outfit, even though X has reservations about the outfit, to avoid hurting Y's feelings before an important event. | 0 Fine | X tells Y (a guest in X's home) that they make the best cookies they've ever tasted, even though X finds them a bit too sweet, to show appreciation for Y's effort. |
X, a neighbor, compliments Y's garden, even though it's not well-maintained, just to be friendly to them. | X, a grandparent, tells Y, their grandchild, that they have the most beautiful smile they've ever seen, even though Y has a missing tooth. |
The full standardized scale for prejudice:
Prejudice Immorality Scale | ||
X physically attacks Y on the street because X hates all people of Y's race.* | 5 Extremely Unethical | X calls the police and falsely accuses their neighbor Y of a crime because X hates all people who look like Y.* |
X engages in relentless mocking and belittling of Y because of negative stereotypes of Y's gender, age, or race. | X, a judge, displays clear bias in legal proceedings, delivering harsher judgments against Y solely because of Y's gender, age, or race. | |
X, a police officer, conducts biased profiling, targeting Y, an individual, because of their gender, age, or race. | 4 Very Unethical | X intentionally excludes their colleague Y from collaborative projects because of Y's gender, age, or race. |
X, a landlord, unconsciously charges higher rent than normal to Y, a tenant, because of Y's gender, age, or race. | X, a financial advisor, decides not to provide loans to Y because of their gender, age, or race, influenced by statistical data suggesting higher default rates for people of Y's gender, age, or race. | |
X, a sports coach, unconsciously provides less encouragement and recognition to Y, an athlete, because of Y's gender, age, or race. | 3 Unethical | X, a customer service representative, unknowingly provides inferior service to Y, a customer, because of Y's gender, age, or race. |
X, a hiring manager, considers Y's application more favorably, based on data they had read about individuals from Y's gender, age, or race who are statistically more likely to stay longer in the job. | X, a friend, comments on Y, an indigenous person, about Y's exotic appearance. | |
X, a manager, assumes that Y, an employee of a religion that is uncommon in that area, celebrates the most common holiday in that area. | 2 Unsure, Lean Unethical | X, a person on a dating app, is less likely to send a message to Y due to subconscious biases towards Y's gender, age, or race. |
X, a supervisor, praises Y, an Asian colleague, for being good at math (a stereotype sometimes held about Asian people). | X, a teacher, offers additional resources and attention to Y, a student from a historically oppressed ethnic group, to foster inclusivity. | |
X, a supervisor, provides flexible work arrangements to Y, an employee from a historically marginalized age group, to accommodate different life stages. | 1 Unsure, Lean Fine | X, an event organizer, invites Y, a speaker from a historically underrepresented gender, to ensure a diverse range of perspectives and experiences. |
X, a teacher, incorporates diverse cultural celebrations into the classroom, fostering an appreciation for different traditions among students like Y without bias related to gender, age, or race. | X, a friend, organizes gatherings that cater to various preferences, creating an inclusive space where Y feels comfortable, irrespective of gender, age, or race. | |
X, a supervisor, ensures equal access to training opportunities for employees, including Y, irrespective of gender, age, or race. | 0 Fine | X, a manager, in an attempt to avoid subconscious bias from impacting their hiring, has all applicant names and photos removed from resumes before evaluating those resumes. |
X, a supervisor, is fair in their dealings with all the people they manage.* | X, a teacher, is kind to their students regardless of the student's backgrounds* |
Scenarios annotated with * were not part of the original immorality study. We evaluated these in a separate study in cases where, for our three standardized scales (sexuality, dishonesty, and prejudice), our existing scenarios didn't sufficiently cover the whole 0-5 range. For instance, for sexuality-related scenarios, only two scenarios fell into the 0 category, so we came up with a few particularly harmless scenarios and asked people to rate them, so that we could fill our standardized scale with sufficiently many examples.
And finally, the full standardized scale table with examples for sexuality:
Sexual Immorality Scale | ||
X breaks into the home of Y and then physically forces Y to have sex despite Y's cries for help. | 5 Extremely Unethical | X slips a drug into the drink of their new acquaintance, Y. And then once Y is so intoxicated as to no longer be capable of consenting, X engages in sexual activity with Y. |
X is a platonic friend of Y. One day, Y gets drunk and passes out, and despite never having discussed this before, X pulls down Y's pants and fondles Y's genitalia while Y is unconscious. | X, who is an adult, befriends a 16-year-old, Y, who lives in their neighborhood. And X tells Y that if Y really cared about X, they would have sex, and Y agrees to have sex, which they then do. | |
X and Y are having consensual sexual activity, and X asks Y if they can do a different sexual activity that they had never engaged in before. Y says no, but X does it anyway. | 4 Very Unethical | X is standing on a very crowded subway train where everyone is packed together, and X purposely touches their crotch against the body of a stranger, Y, because X finds it sexually exciting to do so. |
X and Y are platonic friends, and while X is visiting the home of Y, and when Y is doing something else, X secretly sneaks into the bedroom to sift through Y's underwear because X finds it sexually exciting. | X is a professional masseuse, and while massaging their customer Y at the massage parlor where they work, X starts massaging Y's genitals, hoping that it will lead to a much larger tip from Y, but Y tells X to stop, so X stops and resumes a normal massage. | |
X and Y are in a long-term romantic relationship, and X spends a long time trying to convince Y to engage in a sexual act that they don't want to engage in. And Y finally consents despite still not wanting to do the act. | 3 Unethical | X shows sexually explicit photographs of a stranger that they found on the internet to their friend Y without first asking person Y if they are okay with seeing the photographs. |
X makes a highly sexual joke in front of a person they manage at work, Y, though the joke is not about Y. | X flirts with their co-worker, Y, who is in a position of equal power to them, and after Y responds with body language that indicates a lack of sexual interest, X continues to flirt with Y in future interactions, though X does not escalate beyond flirting. | |
X and Y are close platonic friends, and X gives Y a long and deep hug to say hello, but it is a longer and deeper hug than Y desires, and the hug makes Y feel uncomfortable. | 2 Unsure, Lean Unethical | X rests their hand on the thigh of a close platonic friend, Y, during a conversation, even though Y has never demonstrated any sexual interest in X before when their prior friendship has not involved any physical contact, but X makes no further advances on Y. |
X puts their hand for a few seconds on the lower back of a new acquaintance, Y, during a conversation. | X and Y are friends, and X incorrectly perceives Y as being sexually interested in X. And while hanging out, X suddenly kisses Y on the lips, which surprises and shocks Y, but X immediately realizes it wasn't wanted and apologizes to Y and doesn't try it again. | |
X flirts with their co-worker, Y, who is in a position of equal power to them, but after Y responds with body language that indicates a lack of interest, X permanently stops flirting with Y and, from then on, acts just like a co-worker. | 1 Unsure, Lean Fine | X and Y are new acquaintances, and they consensually start to kiss. Then, after a few minutes, while still kissing, X reaches their hand into Y's clothes and touches Y's genitalia. And Y enjoys this, so let it continue. |
X puts their hand for a few seconds on the upper arm of a new acquaintance, Y, during a conversation. | X is walking on the sidewalk, and as a stranger, Y, walks past, X says to Y, "I find you very attractive - would you like to go get coffee with me?". | |
X flirts with their friend, Y, but after Y responds with body language that indicates a lack of interest, X permanently stops flirting with Y and, from then on, acts just like a friend. | 0 Fine | X and Y are romantic partners, and they engage in consensual sexual activity. |
X and Y are on the date; X asks Y, "Can I kiss you?"; Y says, "No"; so X doesn't do it.* | X and Y have been dating for a long time, and when they get home, X kisses Y as a greeting.* |
Scenarios annotated with * were not part of the original immorality study. We evaluated these in a separate study in cases where, for our three standardized scales (sexuality, dishonesty, and prejudice), our existing scenarios didn't sufficiently cover the whole 0-5 range. For instance, for sexuality-related scenarios, only two scenarios fell into the 0 category, so we came up with a few particularly harmless scenarios and asked people to rate them, so that we could fill our standardized scale with sufficiently many examples.
