Sexist Apple Credit Cards

[Sauron]

7 hours ago, SenKa said:

If PCGuy is correct, then it is not broken in a way that disfavors based off sex in such a way that it is designed at such. The sex that has been disfavored have proven themselves to be unfavorable, hypothetically.

From what I gather, the design is flawed in a way that favors this situation. Other credit card providers don't do this, at least not to this extent. Regardless, assuming your credit score should be lower just because you are a woman, even if it were statistically justified, isn't acceptable.

7 hours ago, SenKa said:

Teach an algorithm a certain way it will never know it's wrong, even if it is. Trash in, trash out.

Yeah, so... filter out the trash? That's literally their job as software engineers and data scientists.

5 hours ago, Ehmc130 said:

Wow... People are really running out of things to bitch about. It's completely irreverent what others are getting, just worry about yourself, life's easier that way.  

Yeah, God forbid people care about something as irrelevant as their MOTHERFLIPPIN CREDIT SCORE.

 

Imagine being so unwilling to take women seriously on anything that you argue your credit score doesn't matter.

[LAwLz]

10 hours ago, Sauron said:

On the other hand, just looking at the sources on the wikipedia page for the "pay gap" we find this proper metaanalysis which places the worldwide wage gap at around 30% at the time of publishing with a rate of reduction of about 0.17% per year, which would still place it well above 25% today. Again, the specifics will vary per country and per job/job type - but on average, women earn less than men working the same job.

You're reading that study incorrectly.

That 30% number is the raw gender gap statistic, which is to say, without taking ANY factors into consideration. As the study even says itself, the raw gander gap has been reduced to 30% because the education and training of women in the work force has increased.

 

Quote

Furthermore, our analysis allowed us to investigate the gender wage gap overtime. From the 1960s to the 1990s, raw wage differentials worldwide have fallen substantially from around 65 to only 30%. The bulk of this decline, however,must be attributed to better labor market endowments of females which came about by better education, training, and work attachment.

 

That is to say, the raw data for how much money is being distributed to men vs to women, has decreased from a 65% different to a 30% difference. The reason for this change is because women now has better education and training.

Better education and training = better jobs and salary.

 

What you should be looking at is the "unexplained wage gap", which is another word for "different salary between a man and a woman who are equal".

That number seems to be around 20%, but as the meta analysis also says, they found that a lot of the studies dramatically overestimate what actually belongs in the "unexplained" portion.

Quote

Estimates using panel methods or sample-selection techniques find lower wage gaps; estimates using the Neumark decomposition technique as compared to the Blinder–Oaxaca approach find higher wage gaps. While previous regression results did not show any systematic effects for the unit of wage measure available in the data, the fixed-effects model indicates that the use of non-hourly wages (in general monthly or yearly incomes) results in significantly higher gender wage gaps as would be expected.

A similar situation arises, if experience was not explicitly included in the wage regression: measured unexplained wage gaps are considerably overestimated in such a case.18The panel model also demonstrates more clearly that controlling for occupation decreases the gender wage gap which is no surprise considering that women more often work in low-paid jobs.

That is to say, a lot of the studies (which were used to come to the 20% conclusion) did not actually take years of experience into consideration, nor did they take into consideration what occupation the man vs woman had. Since woman has generally stayed at home more than men to take care of babies/children, they naturally get a lower "years of experience" number, which factors in to their salary. And since women more often work in low-paid jobs, they also earn less because of that.

 

I personally do not believe an inherently sexist based gender gap exits. I firmly believe that a gender gap exists, but that when you take into consideration ALL variables, including things like hours of overtime worked (men work considerably more overtime, and a lot of studies don't take that into consideration) the gap becomes around ~7%, which can easily be explained by things like men being more aggressive and assertive when negotiating pay raises.

Think of it like this, if it was possible to pay women less for the exact same job, wouldn't every single company in the world only try to hire women? I don't think there is a single company in the world that would go "reduce personal costs by 30% with zero effect on productivity and income? Nahh, we don't want that because we hate women".

 

 

Also, the wikipedia page you linked mostly talks about the gender pay gap. Most of those measurements are just "what is the average yearly salary of men in this country, compared to the average salary of women in this country", without taking into consideration even basic things like "what job do they have" or "how much do they work".

When reading statistics about different sexes getting paid differently, it is very important to look into if the study actually compares people with equal jobs and experience. A surprising amount of them do not.

[RejZoR]

6 hours ago, Belgarathian said:

Well yes, but my understanding is that there is data to support that men are higher risk hence higher premiums. 

 

The data surrounding the investigation seems to be that only gender is significantly different, particularly when it comes to credit scores, income, etc as they were all join applications with shared accounts. 

So, because some other men are risky drivers, how does that make me the same? Based on gender? That's pretty sexist I'd say regardless of what "data" says... Premiums should be calculated per individual. Everyone starts at the same level and if you had several crashes in which it was your fault, the premium should increase FOR YOU. If I'm just being male or not being at fault for crashes or not having them in the first place, why should I pay higher premium? This is actually a thing and despite people "talking" about it, nothing is really getting done about it coz fuck men. As always.

 

But one woman had her credit limit lower than her husband and whole frigging world is up in Defcon 5 rage mode even though it's most likely actually calculated from income, expenses and current employment stats, like every bank that I know does. Which, unless she has absolutely perfectly identical employment, income and current/past credits to her husband, it won't be the same. But no one gives a shit about any of it because woman is being discriminated and everyone should drop whatever they are doing right now and talk about THIS "issue"!!!!!

[WereCatf]

11 minutes ago, RejZoR said:

But one woman had her credit limit lower than her husband and whole frigging world is up in Defcon 5 rage mode even though it's most likely actually calculated from income, expenses and current employment stats, like every bank that I know does. Which, unless she has absolutely perfectly identical employment, income and current/past credits to her husband, it won't be the same. But no one gives a shit about any of it because woman is being discriminated and everyone should drop whatever they are doing right now and talk about THIS "issue"!!!!!

Someone really has some fricking fragile manhood, if he has to start raging like that when any article about women being treated wrong comes out.

[Kisai]

55 minutes ago, LAwLz said:

 

That is to say, a lot of the studies (which were used to come to the 20% conclusion) did not actually take years of experience into consideration, nor did they take into consideration what occupation the man vs woman had. Since woman has generally stayed at home more than men to take care of babies/children, they naturally get a lower "years of experience" number, which factors in to their salary. And since women more often work in low-paid jobs, they also earn less because of that.

 

Basically AI just learns the feedback loop and leans into it. Just like how Marketing learned that pink aisle/blue aisle marketing made more money by reinforcing gender-bias, because moms were more likely to buy more for their kids if they knew which one it was for. It wasn't that they were making more sales, but that they were affecting the buying habits. So a lot of toys that were more gender neutral never wound up in either aisle (really LEGO is about as gender neutral as you can get) but instead got separate boy and girl versions of basically the same thing that only differentiated in color.

 

What is really desired is for algorithms and AI to actually use different algorithms that are blind to each others nature and select scoring and sorting algorithms based on different things.

 

For example. If a job requires responsibility, you would want someone who is responsible right? Well statistically women are more responsible than men. However if your algorithm is simply looking for the cheapest pushover of a person to fill a disposable job, that might also be a woman. 

 

[justpoet]

1 hour ago, Sauron said:

that you argue your credit score doesn't matter.

Credit scores don't matter unless you're dumb enough to go into debt.

[Trik'Stari]

Unpopular fact: There are differences between men and women.

[rcmaehl]

Machine learning and algorithms reflect the biases they were created in. If you train them on biased data, even if statistically accurate, you will get a biased system.

[Founders]

5 hours ago, Sauron said:

Yeah, God forbid people care about something as irrelevant as their MOTHERFLIPPIN CREDIT SCORE.

 

Imagine being so unwilling to take women seriously on anything that you argue your credit score doesn't matter.

You’ve mistaken... I’m not arguing that your credit score isn’t important, it is. What is irrelevant is worrying what other people’s credit scores are. Your second point in no way reflects my views, you’re jumping to conclusion and making assumptions.

[The1Dickens]

It would be interesting to see, in the homes where the wife has the lower credit limit, who files as head of household on their taxes. Also, in the case of home ownership, who is on the title first, and how they are vested.

 

As for the US credit reporting, of the 3 bureaus, TransUnion, Equifax, and Experian, they all gauge your credit-worthiness (credit score) differently, there is no requirement to report to all 3 of them, let alone a single one. And each entity that extends a line of credit may use the score of one bureau over another, even if one is higher than the chosen one.

 

Also, little known fact, you get a free copy of your credit report once per year from each of the bureaus.

[Taf the Ghost]

6 hours ago, RejZoR said:

So, because some other men are risky drivers, how does that make me the same? Based on gender? That's pretty sexist I'd say regardless of what "data" says... Premiums should be calculated per individual. Everyone starts at the same level and if you had several crashes in which it was your fault, the premium should increase FOR YOU. If I'm just being male or not being at fault for crashes or not having them in the first place, why should I pay higher premium? This is actually a thing and despite people "talking" about it, nothing is really getting done about it coz fuck men. As always.

 

But one woman had her credit limit lower than her husband and whole frigging world is up in Defcon 5 rage mode even though it's most likely actually calculated from income, expenses and current employment stats, like every bank that I know does. Which, unless she has absolutely perfectly identical employment, income and current/past credits to her husband, it won't be the same. But no one gives a shit about any of it because woman is being discriminated and everyone should drop whatever they are doing right now and talk about THIS "issue"!!!!!

One could argue that this really was just a vehicle for the combination of humble bragging & "flashing cash" while also being able to "play the victim". This is like prime "Wives of ultra-rich Men complaining about life" on Twitter.

 

I'm also still at a loss for why these very rich people are even applying for Apple Credit Cards. Are they that hard up for Status Symbols that their Apple Account (not the devices, the account) is the Status Symbol?

[PCGuy_5960]

15 hours ago, Sauron said:

It's broken in a way that disfavors based on sex, so by definition it's sexist. Sexism doesn't need to be the consequence of direct hate to be harmful. Obviously it's not sentient and it doesn't have a personal vendetta against women, nobody is saying that.

I mean an Algorithm is dumb, it doesn't know what a woman or a man is. Plus, it would be naive to assume that such an algorithm does't take into account more than just gender (could also take into account age, spending habits, how much money the woman makes relative to her husband, etc). What I mean is it's probably not discriminating against women, but it is discriminating against groups that it thinks are less profitable for the company.

 

14 hours ago, SenKa said:

If PCGuy is correct, then it is not broken in a way that disfavors based off sex in such a way that it is designed at such. The sex that has been disfavored have proven themselves to be unfavorable, hypothetically.

Yeah pretty much.

 

Obviously, I have no idea how the algorithm works and this is all just speculation on my part. I'm just pointing out that the algorithm is probably a lot more complicated than people think.

7 hours ago, Sauron said:

From what I gather, the design is flawed in a way that favors this situation. Other credit card providers don't do this, at least not to this extent. Regardless, assuming your credit score should be lower just because you are a woman, even if it were statistically justified, isn't acceptable.

There is also a chance that other algorithms are biased in such a way that they seem more unbiased to humans.

[Drak3]

5 hours ago, Trik'Stari said:

Unpopular fact: There are differences between men and women.

Yeah, but feelings are more important than facts in 2019.

 

 

 

 

 

 

 

/s

[Sauron]

6 hours ago, LAwLz said:

That 30% number is the raw gender gap statistic, which is to say, without taking ANY factors into consideration. As the study even says itself, the raw gander gap has been reduced to 30% because the education and training of women in the work force has increased.

Fair, but the low reduction rate statistic is specifically about discriminatory factors. The conclusion clearly states that while adjustments lead to a lower wage gap estimate it never even suggests that there is no such gap. I'd also argue that there are quite a few societal factors that cause that raw wage gap beyond individual women's choices and while that's beside the point in terms of wage gap, it's definitely part of the point when talking about credit scores.

6 hours ago, LAwLz said:

What you should be looking at is the "unexplained wage gap", which is another word for "different salary between a man and a woman who are equal".

Reading further into the wikipedia article, particularly the per country sections, the adjusted gap seems to be around 10% in western countries such as Germany, lower if you account for motherhood but I would argue that's not a very good reason to pay people less. Regardless, it's present. There's also a study commissioned by the US Joint Economic Committee which shows with very little doubt that at least some discrimination is afoot there and that, regardless, there is a consistent unexplained gap across the board.

6 hours ago, LAwLz said:

I personally do not believe an inherently sexist based gender gap exits.

Well I'm sorry, the science disagrees. We can argue all day about how large the discriminatory gap is but its existence is pretty much undeniable.

6 hours ago, LAwLz said:

the gap becomes around ~7%, which can easily be explained by things like men being more aggressive and assertive when negotiating pay raises.

That's just a guess and it definitely doesn't explain women getting more jobs when the applicant's sex isn't specified. I would also argue that men have more opportunities to negotiate said raises due to social norms and expectations. What you're highlighting is society wide sexism, which doesn't really help your point. 7% less money for no good reason isn't really something to brag about.

6 hours ago, LAwLz said:

Also, the wikipedia page you linked mostly talks about the gender pay gap. Most of those measurements are just "what is the average yearly salary of men in this country, compared to the average salary of women in this country", without taking into consideration even basic things like "what job do they have" or "how much do they work".

There are plenty of studies that correct for these factors, I mentioned some of them above but there are more if you'd take a deeper look.

 

Though again, regardless of what job they have such a wide overall gap is indicative of some serious societal imbalances.

5 hours ago, Trik'Stari said:

Unpopular fact: There are differences between men and women.

More like irrelevant fact. There are differences between you and me despite the fact that we're both men, that doesn't mean one of us should earn more or less for doing the same job or be automatically given a lower credit score.

2 minutes ago, PCGuy_5960 said:

There is also a chance that other algorithms are biased in such a way that they seem more unbiased to humans.

...yeah, that's kind of the point. All algorithms are biased in that they apply different weights to different factors. Bias itself isn't a problem. Harmful bias, on the other hand, is. The algorithm optimizes for whatever goal you give it while staying within the boundaries you set - even if it were objectively more profitable to eat toddlers that wouldn't make it an acceptable recommendation and whoever designed the system would have to disallow this scenario (and see a therapist).

7 minutes ago, PCGuy_5960 said:

Plus, it would be naive to assume that such an algorithm does't take into account more than just gender (could also take into account age, spending habits, how much money the woman makes relative to her husband, etc).

This doesn't explain the difference between similarly aged partners upon getting the card for the first time. Also again, other banks don't do this, or at least not as blatantly.

9 minutes ago, PCGuy_5960 said:

What I mean is it's probably not discriminating against women, but it is discriminating against groups that it thinks are less profitable for the company.

I understand what you're saying, that doesn't make the situation better.

[spartaman64]

7 hours ago, LAwLz said:

Remember when Amazon's machine learning recruitment was called "sexist" for recommending hiring men over women? As it turned out, in Amazon's case the machine wasn't even being fed the gender of the applicant. It looked at a ton of different variables and it just so happened that far more men than women were suitable for the job, in a completely unbiased way.

But we humans then looked at the result, and since we humans LOVE to find patterns, we applied our own bias of "women are just as capable as men and should therefore be recommended 50% of the time!" and then concluded that it must be sexist.

 

Do we even know what data was used to determine the credit score for the Apple credit card? It can't be sexist if it doesn't take gender in as a factor. And if it does then it probably looks at statistics to draw conclusion. Like someone said earlier, nobody has a problem with men having to pay more for car insurance because "statistically they are higher risks". If that's okay, why shouldn't it be okay to give women a lower credit limit if they statistically and in general doesn't have the same spending power/credit score as men?

except they share a bank account so they have the exact same spending power and she has a better credit score. if its like 5% difference you can chalk it up to variations and stuff but 10 times more credit limit? and 20 times more credit limit? like wtf is up with that. i dont think they intended to make the algorithm sexist but its sexist nonetheless. i dont think loaning money to women has 20 times more risk than loaning money to men as i emphasis again they have the same spending power so the only factor i can think of to lend more or less money is risk. Lets put this into perspective if the wife has 50,000 credit limit the husband would have 500,000 credit limit or in woz's case 1,000,000

 

[PCGuy_5960]

13 minutes ago, Sauron said:

...yeah, that's kind of the point. All algorithms are biased in that they apply different weights to different factors. Bias itself isn't a problem. Harmful bias, on the other hand, is. The algorithm optimizes for whatever goal you give it while staying within the boundaries you set - even if it were objectively more profitable to eat toddlers that wouldn't make it an acceptable recommendation and whoever designed the system would have to disallow this scenario (and see a therapist).

I didn't say that it was OK. I agree that they should fix the algorithm so that the difference in credit limit is minimized. I just disagree that this is rooted in sexism, it's just an algorithm doing what it's programmed to do, maximize profits and minimize losses.

22 minutes ago, Sauron said:

This doesn't explain the difference between similarly aged partners upon getting the card for the first time. Also again, other banks don't do this, or at least not as blatantly.

Different income could explain it though. If let's say the man made 5x more than the woman the algorithm may assume that the woman doesn't really need the same credit limit. 

[Guest]

12 minutes ago, PCGuy_5960 said:

I didn't say that it was OK. I agree that they should fix the algorithm so that the difference in credit limit is minimized. I just disagree that this is rooted in sexism, it's just an algorithm doing what it's programmed to do, maximize profits and minimize losses.

Different income could explain it though. If let's say the man made 5x more than the woman the algorithm may assume that the woman doesn't really need the same credit limit. 

In the case of many reported both parties have joint bank accounts and credit cards. Income is shown between them. As far as I am aware, income is only revealed to the credit reference agencies via bank accounts and not via tax returns. As such in a joint bank account situation both parties should be treated equally. Obviously something is awry, and I am sure it will be rectified. These things need to come out into the open so they can be investigated and fixed. 

 

I also find it quite funny that the very rich also have credit cards. One would think they do not require them and instead would just use debit accounts.

 

What was more alarming in the article was the other case where there appears to be discrimination in other companies based on race. That will be interesting to follow in the courts.

[DrMacintosh]

12 hours ago, mr moose said:

When a couple uses a single account it is impossible to know who is more responsible with money just from the numbers.

 

That’s why a credit history exists. Your credit score is tied to your social security number in the states. Your potentially bad history with going into debt isn’t wiped out because you opened an new bank account with your more financially responsible SO. Yes, even if you work the same job and the same hours. That would be ridiculous. 

[spartaman64]

2 minutes ago, DrMacintosh said:

That’s why a credit history exists. Your credit score is tied to your social security number in the states. Your potentially bad history with going into debt isn’t wiped out because you opened an new bank account with your more financially responsible SO. Yes, even if you work the same job and the same hours. That would be ridiculous. 

And thats why they specified that the wife has a better credit score

[DrMacintosh]

4 minutes ago, spartaman64 said:

And thats why they specified that the wife has a better credit score

Unless the Banks algorithm is released and there is a specific male or female flag/variable/etc. which directly tanks a women’s credit score, it’s going to be way easier for the Bank to prove it’s not sexism than the accusing party is going to be able to prove that it is sexism. 

[spartaman64]

11 minutes ago, DrMacintosh said:

Unless the Banks algorithm is released and there is a specific male or female flag/variable/etc. which directly tanks a women’s credit score, it’s going to be way easier for the Bank to prove it’s not sexism than the accusing party is going to be able to prove that it is sexism. 

I dont think theres an if female then lower credit limit in the algorithm code but that's not required for it to be sexist. Just like how after the civil war southern states didn't make laws that say black people can't vote they made laws like if you are not a 2nd generation landowner you can't vote or if you can't pay the voting tax then you can't vote. Those are still racist even if they didn't mention a race. If the bank has something like if bought make up then lower credit limit that's still sexist imo. And like I said before I dont see what factor could make lending money to females 20 times more risky than to males

[SpaceGhostC2C]

I love to see people discussing "the algorithm". "The algorithm isn't sexist", "the algorithm inherits the biases"....

"The algorithm" is a tool. Despite its name, "AI" is just a tool. Saying the algorithm did this or did that is like saying "I didn't put that nail there, the hammer did!".

 

Essentially, how a company and the people working for it made decisions A,B,C doesn't matter. Ultimately, those people, and the company by extension, made the decision. The tools they used are their problem.

 

And when it comes to discrimination, it doesn't matter if women earn less or more or whatever, or even whether credit scores are different ex-post for women, what matters is if being a woman (or a male for that matter) in itself was used as a decision variable. You can have discrimination against women and higher credit limits for women. Take a simple linear rule: 

 

CreditLimit = b₁ Wage + b₂ Assets +b₃ CreditHistory + b₄Sex

 

As long as b₄ != 0, you have discrimination. Women may have lower scores because of lower wage, assets, history and discrimination. Or they could have higher limits due to higher wage, assets, or credit history, but not "high enough" due to discrimination. Women could benefit from discrimination if b₄>0, but still is discrimination.

The real question is whether sex, or a proxy that ultimately represents sex after controlling for all other variables, is a variable in making this tool. Because, if it is...

[Brooksie359]

19 hours ago, Curufinwe_wins said:

Look. This isn't a "men have higher driving rates, and crash rates so they pay 25% more car insurance".

 

This is "Man and woman share all accounts and funding sources, woman has higher credit score than man... woman has 1/10th the credit availability on this particular card."

 

That is far far far beyond different spending habits and income and straight into a blatantly sexist algorithm.

 

Which is not uncommon (in fact, it's extremely common) in ML based algorithms. Turns out machines are extremely good at finding, replicating, and amplifying our own systemic biases based on training sets.

I mean both practices are sexist.....

[DrMacintosh]

3 minutes ago, SpaceGhostC2C said:

CreditLimit = b₁ Wage + b₂ Assets +b₃ CreditHistory + b₄Sex

 

As long as b₄ != 0, you have discrimination.

That’s really what it comes down to and a lot of people don’t seem to understand that. 

[spartaman64]

2 minutes ago, Brooksie359 said:

I mean both practices are sexist.....

Yep they should target the factors instead like they already take into account how much you drive. But at the same time I don't think men have 10x higher insurance premiums