Slashdot Mirror
Do Firefox Users Pay More For Car Loans?
RandyOo writes "Someone wrote in to The Consumerist to report an interesting discovery: while shopping online for a car loan, Capital One offered him different rates, depending on the browser he used! Firefox yielded the highest rate at 3.5%, Opera took second place with 3.1%, Safari was only 2.7%, and finally, Google's Chrome browser afforded him the best rate of all: 2.3%!
A commenter on the article claims to have been previously employed by Capital One, and writes: If you model the risk and revenue of applicants, the type of browser shows up as a significant variable. Browsers do predict an account's performance to some degree, and it will affect the rates you will view. It isn't a marketing test. I was still a bit dubious, but at least one of her previous comments backs up her claims to have worked for a credit card company.
Considering the outcry after it was discovered that Amazon was experimenting with variable pricing a few years back, it seems surprising that consumers would be punished (or rewarded), based solely on the browser they happen to be using at the time!"
Who's saying it is? Correlation is really all insurance rates need to be based on.
which is totally what she said
What the interest rate for IE was?
First of all, if this were an indicator, it would seem logical that Internet Explorer users would trend lower incomes than anyone else. Anyone educated enough to even be using an alternate browser on a PC is probably educated enough to be making more money than your run-of-the-mill user. At least with Safari it kind of makes sense. Anyone using an Apple has enough extra money to waste it on hipster cred.
Of course these opinions are soley those of a Firefox user who likely makes considerably more than the average schlub who's surfing the internet for porn between his shifts at Denney's.
SJW: Someone who has run out of real oppression, and has to fake it.
It doesn't matter in this case; all you need is correlation.
Now I wish I'd set up my car insurance on my own machine rather than while at my mum's house.
which is totally what she said
What rate did they offer for IE6?
Surely that must be good indicator for negligent behaviour?
You base your assumptions that:
1. the browser is telling the truth about the user agent.
2. the browser used it the one the customer uses regularly.
3. your eveluation about "risky" browser is correct.
4. all this makes some kind of sense.
Maybe Computers will never be as intelligent as Humans.
For sure they won't ever become so stupid. [VR-1988]
It's always been an interesting problem with insurance. If they can find some data that reliably correlates to account performance why not use it? Does it matter if it's logical so long as it's true and accurate (note I'm not saying that these particular data are)? Of course if news of certain criterion gets out and causes enough of an outcry to loose them money then they'll stop using it. They are a business after all.
On the other hand the theory behind insurance (not behind insurance companies mind you) is to pool high cost but low probability risk so that everyone chips in a little in the the chance that they might be one of the few who need help. With this principle the idea is to spread out the burden. As companies get more and more accurate predictors of your individual risk and charge you accordingly they start to defeat the entire purpose of insurance. Think of what the end state of a perfectly predictive set of data would be. Everyone pays for exactly what they will cost, in other words it's not insurance at all.
Repeat after me "People who blindly yell 'Correlation is not causation' should be slapped with a trout.'
FTA: "I figured it had just gone up since I received the email. I tried to use their little payment calculator but the flash based widget wouldn't work properly in the Firefox Beta so I loaded up Safari to try and funny enough the rate offered was 2.7%. I checked in Chrome and Opera to see if it was maybe just something wrong with the Firefox beta and Chrome's rate was 2.3% while Opera's was 3.1%."
and "Devin installed fresh versions of the browsers in order to make sure the changes didn't result from different cookie settings. It seems those looking for a Capital One loan should apply through Chrome."
There are no other obvious variables. The only crime you can punish this guy for is not repeating the experiment across other computers. You can try it for yourself to see if it holds true for you as well.
See my journal for slashdot ID's by year. Mine created in 2005. http://slashdot.org/journal/289875/slashdot-ids-by-year
Correlation may not imply causation, but it certainly does waggle its eyebrows meaningfully in causations direction.
Here is a comment from the article. I have not tested this.
bearymore::: ""Wow! I have a Firefox addon which spoofs the user agent. When I go to Capitalone with the default Firefox as the user agent I get 3.50%. When set the addon to tell Capitalone I'm using Internet Explorer, I get 2.70%. When I switch back to Firefox, I get 3.50% again. Keep in mind, I'm using the same browser and simply opening the site in different tabs""
So either deliberate or incredibly bad coding for a financial site.
So what, IE users are not allowed to buy cars?
It figures they seem like a liability, because they are accustomed to their machines crashing
The phrase "correlation is not causation" is strongly correlated with stupidity.
Just turn JavaScript off and you'll get the higher 3.50% every time ...
Seems like something's up with their variables; A different cg variable causes a different rate for the same zip code.
This calls for a Firefox add-on or greasemonkey script that will take such pages, request the pages with different user-agents, and compare the resulting pages for differences. If the only difference is a single numerical value, it should be easy enough to catch. And then print it to the user in a neat table or graph.
I wonder if this correlation has anything to do with AdBlock? Maybe IE users see a lot of ads competing on rate, while Firefox users don't. Or maybe Firefox users pay off their car early and aren't as profitable....
Retaliation for Ad-Block!
http://xkcd.com/552/
Flamebait? Yes.
True? Yes.
See my journal for slashdot ID's by year. Mine created in 2005. http://slashdot.org/journal/289875/slashdot-ids-by-year
http://www.justice.gov/atr/cases/ms_index.htm
Hop to it.
I mean if you use Lynx you must be a bearded freak living in your parents basement.
"Dear literal net, yes I know it will now work with Lynx"
See my blog http://ilovecookes.blogspot.com/ for light hearted technical information.
Installing fresh browser versions doesn't clear your existing cookies...
Insurance is a game of statistics. Yes, some browsers lie in their user agent string. Yes, sometimes people use other people's browsers. But MOST browsers are what they say they are, MOST people use their own machine pretty regularly, and if the insurance company is really doing this (which I take no position on), then you can bet they have statistics to back up their belief that there's a difference in insurance risk, ON AVERAGE, between users of different browsers.
The insurance company couldn't care less if the correlation holds true for every single instance. They know that it doesn't, in fact. But if it holds true often enough, then they can use that data to offer some people a better price on the insurance, because they're statistically less likely to file a claim.
From the text of the email, it seems that the user launched the site once in each browser. The quote was probably changing each time the page loaded (according to some actuarial variable, intentional randomness to see how people reacted to different prices, or a bug in the software making the quote) completely independently of the browser choice.
No kidding!!! What do you say at this point?
"Correlation is not equal to causation" is what an unintelligent person says when they wish to sound intelligent. it's something they once heard they thought was clever, and they think that by aping this simpleminded thought they are adding something to the conversation, when they are just generating useless noise
an intelligent person would actually be looking at the merits or lack thereof of the correlation, and talking about if causation is implied or not
intellectual property law is philosophically incoherent. it is your moral duty to ignore it or sabotage it
Orc mischief to me.
Hope is the currency of fools
Software engineers are prone to lapsing into bad bill paying habits, more out of negligence than need.
Software engineers are more likely to use Firefox than IE.
Don't ask how I know.
they could use that information to find people using apple products. and charge such people 9.5%. as long as the car was titanium white, such people would happily purchase at that rate
intellectual property law is philosophically incoherent. it is your moral duty to ignore it or sabotage it
That is a valid point. However, the OP's "correlation is not causation" is not a valid point. I just get so freaking tired of the Pavlovian response...
And you can run the same experiment on your own PC and determine if the findings are true or not.
See my journal for slashdot ID's by year. Mine created in 2005. http://slashdot.org/journal/289875/slashdot-ids-by-year
Blindly stating 'Correlation is not equal to causation' is not evidence against causation.
I don't understand why anyone still goes to large banks for loans or savings instead of community banks and credit unions. The corporations will nearly always give you a worse deal. And have fun talking to the customer service reps who live half a world away. My checking account is truly free and it earns me interest, like it should. And my credit card has 10% APR. Big banks are parasites. Vote with your wallet, people, it's not that hard!
Technically, i think the W3C standards for HTML 4.0 require a 3.5% interest rate. The other browsers are just providing proprietary rates in a standards breaking battle of oneupmanship. I think HTML 5 attempts to address this issue with the new heavily scriptable InterestCanvas element.
Maybe it is just the tinfoil hat person inside of me, but I wonder if there's more to it than just risk assessment. Consider the fact that each browser maker has a vested interest in preferential treatment by websites. ATM, the most hotly contested treatment has to do with how the website looks in a particular browser(ever wonder why Facebook runs faster in IE?). But as we move towards more standards compliant browsers, that area will be slowing diminishing(especially if IE6 goes the way of the Dodo bird). So what's next? What's going to drive users to use one browser over another, besides being pre-installed on particular OSs & devices? What about having special deals for those that use their browser? This becomes especially true if the browser maker also can control how the user finds a vendor's product like Google & M$ can. All it would take from Google is to put a vendor 1st in a search for a car loan and for the bank to check which browser for someone to get a better rate just for using Chrome.
While it may be true that users of different browsers tend to have varying levels of credit worthiness that tracks to what browser they use, this creditworthiness will be more clearly (and accurately) reflected by other criteria. As an example, while people with a particualar credit score range may be more likely to select a particular browser, it seems improbable to me that people who select a particular browser are more or less likely to default on debt than people with the same credit score who select a different browser.
The truth is that all men having power ought to be mistrusted. James Madison
Sure, but in this case it makes perfect sense. Obviously anyone using Firefox is a filthy open source hippy who demands everything for free. If you tried to loan money to a Firefox user he'd probably just spend it all on weed and then claim your demands to pay back the money violate his privacy in some way.
Opera users aren't good people to lend to either, since none of them have any income. They spend all of their time scanning the Internet for stories about browsers (any browser) so they can jump in and extol the virtues of Opera. This leaves no time to hold down a real job.
Safari users are good people to lend to because, since they're using an Apple product, you already know they're accustomed to paying huge premiums for slightly shinier versions of various consumer goods. All you have to do is send the bill in an elegantly designed box and they'll pay it without question every month.
Chrome users are the best because their close relationship with Google shows they've already given up on the whole concept of privacy and will gladly supply you with any information you ask for. Plus, all you have to do is tell them you're not evil and not only will they allow you to do whatever you want no matter how evil it is, they'll actually defend your actions to others!
Clearly browser choice is and should be a significant factor in the lending business.
Now, at least we know there is room for negotiation on the interest rates.
Don't pay more if you don't have to.
He who knows best knows how little he knows. - Thomas Jefferson
It is sometimes. For example, they should be charging the Safari users more, as those Mac fanboys will pay anything if it's marketed right.
"He who can destroy a thing, controls a thing." --Paul Atreides, Dune
Dammit and I used FF to apply for a car loan. Turned down, naturally, because I leaned the other way.
Used to work there. Capital One has done this for many, many years. They used to send snail mailers with different rates to random households in the same neighborhood. The CEO is one of the early proponents of experimental marketing.
One commenter says:
"it sounds like an A/B test. when you arrive at the site, you're assigned a random rate. they do this because they want to be able to see how customers respond to price increases, decreases, special offers, webpage designs, etc.
after your first arrival, every following time you reload or return to the site, you'll see the same rate, since you'll keep the persistent cookie. if you delete your cookies and see the same results every time, there could still be other means for them to be sure it's you, so there's a chance they're using those tricks (flash cookies, for instance). reinstalling the browser may not even fix this.
of course it's possible they're distinguishing by user agent, but i think it's very unlikely."
Sounds plausible.
I can tell you that this was the case for me on Zillow, when it comes to calculating the monthly payment. I and my brother thought it was quite strange but assumed it must be some coding bug. IE gave lower numbers than firefox or chrome. I don't remember what the numbers were.
The entire insurance industry is based on the principle of retaliation against customers who cost the company money and rewards for those who dutifully pay more than they claim. Any statistical basis they can use to figure out which customer is which ahead of time... they'll use. The only ones they (usually) won't use are those prohibited by law; instead they look for some other factor that correlates strongly, and use that instead. So if they aren't allowed to use race, they'll use neighborhood... if they aren't allowed to use neighborhood, they'll look for something else. Maybe ISP, or IP address block, or OS, or.... browser. They don't care, as long as there's a good correlation. It's not that they hate Firefox users, any more than they hate people of a particular race or ethnic group or neighborhood or religion or credit score or driving record... they just don't care. Like any "good capitalist" they want to maximize profit.
http://alternatives.rzero.com/
And if the information returned Internet Explorer, could they offer a lower rate, believing that the car being purchased will be in the shop more often?
"Correlation is so highly correlated to causation that we're forced to conclude that correlations cause causation."
Still cracks me up. Probably caused by^W^W^WHighly correlated with reading hundreds of threads like this on Slashdot...
I suffer from attention surplus disorder.
Only when used in idiotic contexts like the first post.
In this case, they are looking for patterns, and the actual cause is irrelevant, if they notice patterns that indicate higher risk, regardless of the root cause, they follow them.
Now, there are correlations (violent video game players have a higher incidence of violent activity) that, by themselves, do not imply causation (further tests are needed for causation). When someone tries to use these as an implication of causation, then the phrase "correlation does not imply causation" is quite intelligent.
And, I just implicitly defended Capital One. I feel morally dirty now. Thanks a lot.
Self proclaimed typo king, and inventor of the bear destroying coffee table (patent not pending).
I just spoke to my wife, who spent 10 years working for Capital One at their corporate headquarters in Richmond (*). When I mentioned the gist of the story to her, her response was "I believe it", which actually floored me as I am always telling her about various conspiracy stories I see online. She immediately offered up that they use various bits of financial information to determine what % interest rate you would be charged and suggested that the Safari users would be seen as having more money (as Mac owners) and hence a lower risk.
(*) As an aside, if you haven't been to the Capital One campus in Richmond then you are missing out. The selection of restaurants is amazing. They have outdoor basketball and volleyball courts. And they even have a frickin' treehouse where you can go and sit outside with your laptop and do your work via Wi-Fi. The downside is that corporate doesn't instill loyalty as pretty well everyone in Richmond either knows someone who was downsized from Capital One, or is downsized themselves.
The only better work conditions nearby would be CarMax, which for their staff they don't offer fixed length vacations. You get to choose how long you take off each year - they assume that if you are professional enough to do your job that you are professional enough to know when your job is done!
Or it could be that firefox users tend to work in the tech field, a field with notoriously short and unreliable job durations. It could be that firefox users are rated lower due to more frequent (if short) bouts of unemployment.
"I used IE6, and I didn't get a rate. In fact, i don't think the page exists at all. This may be fake"
What rates did lynx/elinks garner? What about IE 6? Does faking a useragent do enough?
After reading about this, I am completely {outraged}{amused}{indifferent}{turned on} by this practice.
Javascript error. Aborting script.
Any "Weakest Link" jokes you'd like to make, as long as you're hitting these phenomena at the height of their popularity?
Yeah, I'd feel dirty if I defended Capital One, too.
They're local here and known by many people as Crapital One for their firing sprees and tendency to make employees disappear. They're also the fastest bank in the nation to sue their own customers. (I sell data to bankruptcy lawyers who keep up with this kind of thing.) They're the last bank I'd go to for a credit card or a loan anyway.
This is indeed true. I've experienced it this past September. However, I think it is not just based on browser, but also based on geographic location. I checked their site from my home PC with Firefox (east coast state), and to be a bit safer, I VPNed in to my office with my work laptop, still using Firefox, to actually fill out the application. To my surprise, the rate was higher (the exit node to the Internet is a north, central US state for the corporation). I called my wife in to the room to show here two computers, both on Capital One's auto loan home page, showing her the two different "as low as" rates on the page. I'm glad I didn't go with them for the auto load - seedy bastards.
I recently noticed the same type of happening when I was shopping for a Verisign SSL cert (clients *sighs* don't get me started on why). When I use Safari or Firefox, the price for a 1-year "Secure Site" SSL cert (with site seal) is $499 - however, when I switch to Chrome or IE8, I get a price of $399. I only had to buy the one, and so wrote it off as a fluke - but I just re-verified that this is still happening for me (tested on both Mac and Windows), given the news on this article.
This smacks me as being seriously wrong - now I have to test all browsers when buying something online, to be sure I'm getting the right price? And no, I'm not going to change my default browser habits, just to get lower pricing...
Worrying works!! 99% of all the stuff I worry about never happens
So the question becomes whether this is also true for other products and services. Mortgages? Insurance? Physical products? Ancillary question: do these companies not realize the potential PR damage stuff like this does?
The crimes of eBay are a disgrace to it's pig latin heritage!
The phrase "correlation is not causation" is strongly correlated with stupidity.
Yes, but correlation is not ... oh, wait ...
Quidnam Latine loqui modo coepi?
"Correlation is not equal to causation" is what an unintelligent person says when they wish to sound intelligent. it's something they once heard they thought was clever, and they think that by aping this simpleminded thought they are adding something to the conversation, when they are just generating useless noise
an intelligent person would actually be looking at the merits or lack thereof of the correlation, and talking about if causation is implied or not
Perhaps, but even considering true that most people who use that phrase are unintelligent wishing to sound intelligent, It doesn't mean that all people using that phrase are unintelligent and using the phrase out of context. To support what I'm saying: "correlation is not causation"
True, but where there is correlation you have to look for causation. Occam's razor -- Can you come up with a better explanation as to why one would get different rates using different browsers? If not, then you have to suspect causation. If so, then you have to find another cause.
Free Martian Whores!
Tests with car loans:
IE 8: 3.10%
FF 3: 2.30%
Chrome: 3.50%
Firefox user agent switcher, it reports to web sites that it is IE 6-8, Googlebot 2.1, MSNbot 1.1, Yahoo Slurp and Iphone 3.0, out of the box; and you can add any user agent you want too, even make up your own. I'll have to play with it more than I have, a lot of sites let Googlebots through their paywalls, but not user browsers.
Apocalypse Cancelled, Sorry, No Ticket Refunds
Is what rates do you actually get? Just because they print a rate on their site doesn't mean you may not get a higher or lower rate when you actually get approved for the product, after they've looked at things that actually matter.
I have trouble believing a random poster that claims to have worked there that this actually influences things. Really? Then where is the actuary data on this, and why isn't it something asked on home loans? Anything that has a significant predictive value int terms of loan performance, banks want to know. Given that home loans are some of the most major loans, they check everything they legally can. If browsers really were such a major predictor, they'd ask on your home loan app.
To me this looks like an error. I cannot believe that the browser a person uses is a good enough performance predictor that they'd offer a 1.2% difference in interest rate (which in the context of these loans is a 35% difference in interest). That is like 100 FICO score kind of interest difference, or more. No way the browser someone uses accurately predicts loan performance to the point of being worth it.
Just remember that the rates on the site are advertising rates. Who know show the hell they decide what it should show or why JS is involved, but that it hows it has nothing to do with the rate you'd actually get when you apply. Not only is the rate they show always the "Very best credit," rate but it often has other provisos. For example when I was looking at a home equity line of credit, the bank had a rate listed on their site that, when you dug in to it, only applied if you got a $250,000 or larger line of credit AND took out $100,000 or more AND kept it out for several months. Ok well that isn't unrealistic (sometimes you finance home construction with such a loan) but clearly most peopel aren't getting that rate.
This is just the advertised rate, which is always based on parlor tricks and weird math. Your actual rate is calculated after your application and credit check are reviewed, and neither of those things care what browser you use.
This makes sense if you think about it. The bank wants to advertise a rate that is appropriate to you, but it doesn't know much about you until you apply. So it has to guess based on the very limited information it does have. Otherwise -- and this still happens all the time -- you see an ad that says 'Low low 2% APR!' and then you apply and find out that you qualify for 6% APR and get pissed off. If looking at your USER_AGENT reduces the spread between what they promise and what they get, it makes sense for them to do it. But don't let that get in the way of your populist rage.
Not anymore, now that this story is circulating. Maybe they'll be able to detect Firefox users who have spoofed their useragent string, and give them even higher rates. :)
So far I tested this in Safari and IE7. Make sure capitalone.com is not open and clear your entire browsing history, cache, everything. In Safari, you can just hit "Reset Safari". In IE, after clearing all that, close the browser, reopen, and visit capitalone.com. Seems to pick randomly 2.3, 2.7, 3.1, or 3.5. You may get the same number again, just try it a few times. Can anyone else replicate this?
Google provides the data they use to base their pricing on, and Chrome gets you the best deal?
Interesting.
Insurance companies don't care about causation. Correlation is all that really matters to them. If people with names beginning with 'A' have more accidents, insurance companies are going to give them higher premiums. The causation doesn't matter.
And not actual rate differences. Remember that nothing they list there is anything but marketing. Your rate isn't your rate until you actually apply. A number on a website is just promotion/information. It isn't a binding offer of any kind. They'd need a credit check at the very least before they'd be willing to make an actual offer.
Now, as you said, the one and only concern with loans is established risk. What factors (that they are legally allowed to consider) increase or decrease risk? All they are interested in is coming up with a probability of default, and then deciding if they want to make the loan and at what rate based on that.
Well browsers just don't seem to play a role. I've heard of no studies, no evidence, that choice in browser indicates anything with regards to handling money. Without such research, it would be a meaningless metric to use.
Also notice that the rate difference is LARGE. Something like that means a significant risk difference. Even if browsers factored in, you really think it'd be that much? You think they'd offer over a percent just based on the browser?
Not hardly. Unless someone can provide some actuary data showing that this has been looked at scientifically, I do not believe it is actually a deciding factor, just some Javascript being squirrely.
LOL. When they started showing the huns/vandals/whatever in their commercials, I knew which company those thugs *really* worked for.
I saved a lot of money on my car loan by switching to Internet Explorer.
Reinstalling your browser doesn't clear your cookies; *especially* your flash cookies, which are troublesome to clear anyway. The only fair test that I would really believe would be changing your user agent string and hammering the reload button. I may try this myself later. Other's anecdotes are less convincing than my anecdotes (to me anyway). I need to see it; or I need to see data from a controlled experiment; I need to see one of these companies admit it; or have the FTC explain that they've done it. You know.... something credible.
Interesting notion though.
Imagine if you weren't allowed to use roads because a bus company complained about your driving 3 times. --skunkpussy
Isn't this whole argument rather moot? I mean, OK, the guy goes to websites for various loan providers, and depending upon his browser, he claims that he is shown different interest rates. Big deal. It's not as if the actual loan is going to be based on the interest rate that was quoted in the little "payment calculator" tool, since any lending institution is going to have to do a credit check on anyone before they loan him $30,000.
Looks like this website's idea has been taken a little further than people expected it would
There's still causation there. If you looked at rates and noticed that through random chance interest rates were higher for people whose names started with "A" you'd have a point. In your example there is causation. The raising of rates was a conscious effort. The fact that it was based on poor methodology is beside the point.
The original article wasn't arguing whether there was correlation between browser use and trustworthiness to repay a loan, it was arguing that there was a correlation between rates and browser use. Whether the loan companies premise was valid is a different issue.
only an idiot finds it necessary to make note of the obvious. and the idea that "correlation is not causation" is only a remarkable revelation to an idiot
if you are in a roomful of moderately intelligent adults talking about the weather, no one needs to say "condensation causes clouds". if someone were to say this, they would immediately be labeling themselves as a fourth grader
so i suppose saying "correlation is not causation" is of value if the average slashdot reader is a 12 year old. but otherwise, making note of correlation not being causation in a conversation such as this thread is of no value whatsoever, other than conveniently labelling yourself for everyone else's sake as the idiot in the room
intellectual property law is philosophically incoherent. it is your moral duty to ignore it or sabotage it
Anyone dumb enough to accept any financial services from them deserves whatever they get coming.
Typically the value you get when you multiply the amount of money spent on relentless TV marketing with the amount of money spent on nonstop direct mail campaigns is inversely proportional to how trustworthy their financial products are.
In that case (as in the one in the article) names starting with A would actually be causation for the increase. What people should be saying is correlation doesn't equal good cause for causation. If I hate people whose names start with a certain letter, and I punch you because your name starts with that letter, the cause for me punching you was because your name starts with that letter. It isn't a good reason for me to do it, but that phrase (correlation != causation) was never intended to imply that.
Was the comment about correlation and causation not aimed at the way loan companies work in general rather than the way this guy experimented with different browsers?
Or in the case of fields like marketing and insurance, causation is irrelevant if the correlation is strong enough.
If teenagers have a 1% chance of getting into an accident in their first year being insured with the company, but ones with red cars have a 3% rate, you don't have to prove any causal relationship between car color and ability to drive, you just play the numbers and raise rates accordingly. Business decisions don't have to be rational, they just need to be right 51% of the time to stay in business.
I've seen this with Netflix before. I don't know if it still does this, but you get offered a month trial when using IE and only a two week trial when using Firefox.
IMHO, privacy concerns like this (and panopticlick) are a primary reason I prefer my browser to not send the USER_AGENT header. It's optional according to the HTTP spec, and most sites work fine without it. For those that don't, my useragent is a phrase, such as "I am not a script" for Wikipedia. Very occasionally I have to set a real one, so I just use WebPositive's.
In any case, I see no benefit to me, as a user, for a site to know my browser and operating system. Perhaps not knowing will cause them to write their webpages to standards rather than to specific browsers.
I wonder if they also use any kind of geolocation / IP Lookup to change these rates as well... It seems to me that targeting localities would make more sense than assumptions based on user-agents
Has anyone tried clearing cookies and filling out the form again in the same browser? Or filling the form in the same browser but at a different IP? Or they could have a load balancer and the result varies according to which server you get.
I have never tried to get a car loan online, but my experience with mortgages is that the advertised rate they give you online is completely pulled out of their ass, and has no resemblance to the final rate they actually give you. Of course by the time they tell you the actual rate you don't have time to get a different loan and meet the purchase timeline you have agreed to with the homeowner.
As a general principle, no, you don't. As a rule of thumb, you are probably safer assuming that a correlation between two variables is the result of a common cause. E.g., in humans, height over 6'5 is strongly correlated with usage of urinals. At no point should you bother to rule out causation (in either direction) in this case.
A more practical example is the media's constant repetition that healthier people have more sex. Of course, they jump to the sexiest but dumbest possible conclusion, that the sex causes the health. But we don't even have to assume that the health causes the sex; in fact we can intuit that health must be only partially the cause, since many healthy behaviors will increase attractiveness completely aside from their health benefits. (Especially diet and its effect on body shape.)
The general principle is that any one cause will have myriad effects, all of which will be correlated with one another. By default, assume a common cause.
Unless this was observed by only changing the user agent string, I would speculate this is related to having different browsing history for each browser. If Chrome is the best deal, was this their preferred browser that they happen to use more often?
I tried this in Firefox. I got random rates from 2.3% to 3.5% by clearing cookies and reloading the page. Same behavior in IE/Chrome and Safari. The rate is randomly generated and tied to a tracking cookie. Reloads without clearing cookies leaves the rate intact. As someone already mentioned, browser re-installs aren't going to clear your cookies.
is there any evidence that this is supported by google data whoring? is there statistical and fiduciary grounding for this, and how is this legal? this is clearly a form of discrimination. and I note the article states that googlers received the premium rates ... why would this be, unless google had some input at the planning or implementation stages? they could easily skew the results and I believe they have already previously done so, when tying their products together, or referring each of their products together ...
Come on, give the average slashdotter some credit! Do you really think slashdot readers are goign to RTFA or even test it themselves? >;)
Step 1 - clear all cookies from capitalone.com
step 2 - load auto rate calculator - 36-mo rates are 3.50, 5.09, 4.84
step 3 - clear cookies from capitalone.com
step 4 - load auto rate calculator - 36-mo = 2.70, 4.09, 3.94
clear again, load again gives 3.10, 4.49, 4.34
Firefox 3.6.12, OS X 10.6.4. Note that I browsed away and then entered the URL on the navigation bar - if I just tried a reload it didn't seem to work. As someone mentioned, this is no different than sending different offers to people in junk mail, to see which ones they will respond to - it's just much more visible (and annoying) to the potential customer.
Correlation will make for defensible decisions. Ones where you can look to your peers and say something like, "Look, they were browsing with Chrome. How could I know they would default their loan?" And your loan granting colleagues will all nod their heads and sympathize. Correlation may even help one to make decisions with more predictable outcomes, even when the causative factor is unknown. While trying to anticipate the future correlates with well with being human, past performance does not guarantee future results. Either way, I'm switching to Chrome until the data indicates otherwise.
Wait. Stop scrolling for a sec. O.K. Thanks. - P
I actually like my Capital One MC.
It's one of the few credit cards that doesn't charge a fee for international transactions.
I save 2-3% over using my Amex whenever I travel abroad.
Jesus used to be my co-pilot, but we crashed in the mountains and I had to eat him.
It actually sounds quite likely to me, and makes sense a bit. If you go by the stereotypes you've got:
Safari: Mac users, i.e. not a huge amount of brains but lots of cash, easily duped, faithful.
MSIE: Windows users, i.e. stupid, plus they're using the browser that came with their machine and is known for poor security, so not tech-savvy at all, Joe Public make be lower income.
Firefox: Linux users, i.e. more intelligent, possibly not as much disposable income. If its Windows FF then more brains than MSIE users.
Chrome: see Firefox, plus could be smartphone users so tech-savvy possibly business people and higher income.
Opera: well who uses Opera these days, very niche, so possibly used to paying more for things. Could be smartphone users so tech-savvy possibly business people and higher income.
These are pretty stereotypical (and I expect I sound like an arsehole) but I expect that's what the ratings are based on, not scientific research.
I agree its very naughty to bias, but I wouldn't put it past the credit industry.
#include <sig.h>
You're arguing a different scenario, which would be the headline "high accident rates correlated with high insurance rates". In that case, yes, having lots of accidents causes insurance companies to charge more.
The Quirkz Handbook of Self-Improvement for People Who Are Already Pretty Okay
Repeat after me "People who blindly yell 'Correlation is not causation' should be slapped with a trout.'
There is a correlation between the number of people who drive expensive cars and the number of people who's kids go to University. Therefore the government could increase the number of University graduates by having a scheme to buy poorer people expensive cars.
I worked briefly in car insurance, and one of the many questions asked by one of the top insurer was how long the client usually kept his car. (Note that this was an information asked from brokers about their customers, not direct b2c)
Turns out people who renew their car often take better care of them so as to maximise resale value, and consequently produce less claims. It wasn't a big difference but it was apparently statistically significant.
I'd like to know if there are also different rates depending on which OS you're running. Or for that matter, what about a mobile device (iPhone, iPad, Android) versus a regular computer.
Netscape users are not good because they are likely so old, that they are libel to die soon anyway causing you to pay out...
Gopher users are not good because they are dead, and the dead don't pay premiums.
Mosaic users...
Disable javascript and cookies, wipe cookies and then go to the site and reload and reload and reload...
You'll get different rates with the same User-Agent.
I mean, there would be no reason to stop at the browser used? The mode of internet access---heck, even IP address---could be factored in as well.
To-do List: Receive telemarketing call during a tornado warning. Check.
People who blindly yell 'Correlation is not causation' should be slapped with a trout.'
Yeah, but where would we get a trout? For me, fishing is NOT correlated with obtaining a trout, it's correlated with getting angry or drunk.
A detestable practice when taken to the extreme. Especially when in electronic world where consumer's data is increasingly used to exploit the consumers themselves, rather than help businesses provide better services.
IANAL, It looks like price discrimination is only illegal if it hampers competition, but not if it only unfairly treats a certain segment of consumers.
Thanks for explaining what "correlation is not causation" means. I never knew that before... ... you might want to read the word "blindly" in my first sentence; it sets the tone for the rest of the argument.
See my journal for slashdot ID's by year. Mine created in 2005. http://slashdot.org/journal/289875/slashdot-ids-by-year
The general principle is that any one cause will have myriad effects, all of which will be correlated with one another. By default, assume a common cause.
Lead by example. What's your common cause as to the assumption in this situation?
Sounds like a good opportunity for a firefox plugin.
Got Code?
Why should the interest rate depend on a credit check..? If I was advertised a certain rate on the website and then they turned around and said "actually, the real interest rate is 5%", I'd tell them where to go.
which is totally what she said
Who Cares about Capital One? thier fees are outrageous and support sucks... Screw Capital One
It's Capital One. They are probably just fucking with people.
Why, without your clothes, you're naked, Miss Dudley!
People who blindly yell 'Correlation is not causation' should be slapped with a trout.
No, no, no. While it *is* true that people who have been slapped with a trout have, at an earlier point in time, blindly yelled 'Correlation is not causation,' from this one cannot immediately conclude that the yelling directly resulted in them being slapped. Further study, possibly a randomized double-blind trial, would be necessary to conclude that the two events are linked.
they could use that information to find people using apple products. and charge such people 9.5%. as long as the car was titanium white, such people would happily purchase at that rate
You're being funny, but Mac users do have statistically higher incomes, as you'd expect. They'd probably get a lower rate from these clowns, since higher income people bring less risk to the table.
Ok the real reason that FF users are being charged more, is because it is open source. You can take it, and modify it anyway you want and do not have to pay. ALSO, FF users tend to use more open-source software whenever they can. That is a HUGE SAVINGS overall, and insurance companies want the money that you are saving from not spending it on software.
Just because it works, Doesn't make it right. - JTM
Variable pricing based on browser is not unusual. Vista Print always give me different prices on their promotions based on which browser I use.
Why should the interest rate depend on a credit check..?
Are you being serious? Have you never secured credit before in your life? ANY lending institution is going to require a credit check prior to lending you any money. You'll notice those advertisements usually have fine print somewhere which indicate "rates as low as" regarding the rate(s) quoted. Your credit score always comes into play, whether or not you realize it. If it didn't, banks would have to offer the same rate to a guy who defaults on all his loans that they offer to a guy who's paid everything on time for years.
Does this mean that Chrome users are CHEAP?
Do I smell lawsuit or what?
Sorry, You'll Have to load the image. You're welcome to mirror it though.
http://www.glowfoto.com/static_image/04-083228L/3809/jpg/11/2010/img5/glowfoto
Why should the interest rate depend on a credit check..? If I was advertised a certain rate on the website and then they turned around and said "actually, the real interest rate is 5%", I'd tell them where to go.
And they'd tell you where to go.
And you'd have to walk, because you didn't get the car loan.
Repeat after me
after me
after me
after me
after me
after me
after me
after me
after me
Let q be a radix > 1. I am in ur base-q, killing 10 d00ds.
An HTTP connection not only provides the user agent (browser) employed but also the operating system. I would wonder if there are any correlations between Linux users and increased or decreased risk. The choice of an OS would seem far more behaviorally significant than the choice of a browser.
How is it that you know so very much about what a prototypical unintelligent person thinks? Do you claim to be a mind reader? I suspect there is a simpler conclusion to be drawn.
An intelligent person would also capitalize and punctuate.
Telling people that correlation does not equate to causation is just a reminder to others that the fact one things occurs after another doesn't automatically mean there's a connection. For you to call such people "unintelligent" is bizarre, especially in light of how many articles Slashdot posts that do imply that two events are connected when they aren't necessarily (and often skewed toward the agenda of the community here; e.g., there have been articles claiming that piracy boosted sales without skepticism, while articles that claimed piracy hurt sales were mocked, often by posters stating that--you guessed it--correlation does not equal causation).
I don't know why you have a chip on your shoulder against users of the phrase or why you consider it useless noise. When you say that an intelligent person would look at the merits of the implied causation, that's what they're doing when they tell you that just because something happened around the time of something else, it doesn't mean there's a connection. In other words, there is no obvious causation at play.
I'd further suggest that people who accuse others of being unintelligent and trying to sound clever are in fact trying to sound clever themselves.
Pretty sure you just get a random rate so they can test out conversions. I don't believe this has anything to do with browser type. And you probably keep getting the same rate on the same browser based on cookies.
Yeah, I'd feel dirty if I defended Capital One, too.
They're local here and known by many people as Crapital One for their firing sprees and tendency to make employees disappear. They're also the fastest bank in the nation to sue their own customers. (I sell data to bankruptcy lawyers who keep up with this kind of thing.) They're the last bank I'd go to for a credit card or a loan anyway.
Yeah, but you have to balance that against the fact that they have funny vikings in their ads. That counts for a lot when you're choosing a credit card, you know.
A good backend would have retained the user's IP address and then gone up to the "uses multiple browsers" high-risk rate.
You can't use race in US finances but you can use proxies. Race is a very important predictor however.
Browser choice is going to distinguish folks by race. --Measure it.
"Correlation is not equal to causation ... always." However, in business, there are lots of things that seems to correlate and are used to make decisions all the time. I've heard "we don't know why this works, but it does" in meetings all the time, so we use it.
I seriously doubt the browser used has much to do with the actual rate provided to a person once their personal data is known, it is just for a teaser ad.
OTOH, Capital One has never struck me as a 100% ethical bank or lender based on their passed advertising campaigns. They seem to target "income challenged" people. OTOH, I doubt any business is 100% ethical no matter how hard everyone tries.
I wonder what the reasoning was that lead to Safari users getting the lowest teaser ad rate? Apple users seem to be prepared to pay much more than others AND pay a premium for excellent customer service. They also probably have higher income levels than most, since most people will not spend $1500 on a $600 PC or $2500 on a $1100 laptop like what Apple sells. There's nothing wrong with that, and I do it for non-computer things all the time.
I tried viewing the Capital One auto loans page with a program on one of my web sites which fetches and displays pages, refusing all cookies and removing all Flash, JavaScript, etc. The browser string sent is "SiteTruth.com site rating system"; it's not pretending to be a browser. This, of course, is a diagnostic service we run to see pages as our web crawler sees them. The only state information the site receives is the IP address, which is not changing. On successive tries, I received:
So the rate returned is randomized.
I would question the legality of advertising random numbers as interest rates.
Similar case when I signed up for Netflix. In Firefox I was offered a 2-week free trial, while IE was 1 month.
LRN 2 SWM
If this were about car insurance ads instead of car loan ads, there'd be a great joke about linking what browser they use to the likelihood of crashes... ...but as it is, I've got nothing. :(
Friend: "The NIC is misconfigured..." Me: "No prob, I'll just telnet in and fix it." *Silence*
Now, there are correlations (violent video game players have a higher incidence of violent activity)n (further tests are needed for causation).
Apparently they've already established the causation link:
http://onlinelibrary.wiley.com/doi/10.1111/1467-9280.00366/abstract
On Slashdot, it's stupid and irrelevant more often than not. People just post it because they have read it before and seen it get +5, insightful. You see it posted every single time there's a story about some indicator of something or other. So the correlation is certainly there.
I think there might be a weak causal relationship as well, since so many slashbots seem to just break down and stop thinking when they come across indicators and other correlations, believing the right answer is to mindlessly harp on about the same old meme, and thus spread it -- in the same mindless form. Most people intuitively understand the difference between correlation and causation, but when they are infected with this nasty little meme, they seem to lose the ability to understand what a correlation is.
so i suppose saying "correlation is not causation" is of value if the average slashdot reader is a 12 year old. but otherwise,
It is when you continue try to make a relation between who uses that phrase and the dumbness of who uses that phrase out of context. Since you are continually making a "correlation with causation", I'm pointing the irony of your comments. See that I'm not saying you are ironic, I'm pointing the irony of your comments.
To reach the value of an argument you can't attack the people who uses that argument as a way of reaching the real value of the argument. That's not being honest. You have to discuss the argument itself.
other than conveniently labelling yourself for everyone else's sake as the idiot in the room
Based on your last phrase, perhaps it's too much to ask for you to discuss arguments rather than attacking people who use them.
Where this really sucks is credit reporting.
Credit reporters have lenders as clients and have a financial incentive to accept whatever information lenders report, regardless of accuracy, and to make removal of inaccurate negative information as difficult as possible.
Lenders have an incentive to have as much negative information as possible on clients as it allows them to charge higher interest rates.
CapitalOne is the bank that's been advertising for new business for years with TV ads featuring a gang of pillaging barbarians demanding to know "What's in your wallet?" while wrecking everything they touch. CapitalOne was of course central to destroying the global economy, like every other large US bank.
Why wouldn't CapitalOne do stupid, selfserving banking practices, in secret? They've only made $BILLIONS by doing that for years, and nobody's stopping them. And with the new Republican House of Representatives, there's only going to be lots more barbarism in banking.
--
make install -not war
What's the argument FOR causation in this case?
That firefox sneaks a look at your account number and steals the money so it can go pay for porn while you're at work making it harder to pay your loan?
Eschew Obfuscation
When someone tries to use these as an implication of causation, then the phrase "correlation does not imply causation" is quite intelligent.
Nah; it requires no more intelligence than a 1-line perl script to generate that reply. Actually, it's doable with a 1-line sed script, which makes it even less a sign of intelligence.
Now if it required something like a 2-line Prolog program, I might be willing to take it to imply some intelligence. But doesn't take even that much processing power to match for a list of thesaurus entries associated with "correlate" and generate a "correlation does not imply causation" reply.
(I sorta like the reply I once say, to the effect that correlation may not mean causation, but it is often the universe's way of saying "Hey, look over here; there's something interesting going that you might want to study." ;-)
Those who do study history are doomed to stand helplessly by while everyone else repeats it.
but wasted far more time
- I'd prefer not to.
So how on earth is someone that has all these browsers on their computer any way a good person to lend to?
Besides, isn't Capital One's usual measure of creditworthiness related to the number of intact kneecaps you have, and how much you value them?
I don't have any version of IE, but do have a few browsers installed on Ubuntu 10.04 amd64. FWIW, my IP resolves to somewhere in Finland. Here are the rates offered:
Opera 10.63 = 3.50% new cars, 5.09% used
Konqueror 4.4.2 = 3.50% new cars, 5.09% used
Firefox 3.6.12 = 3.10% new cars, 4.49% used
Chromium 9.0.568 = 2.70% new cars, 4.09% used
Epiphany browser 2.30.2 = 2.70% new cars, 4.09% used
Hitting refresh did not change the rates offered, even if all cookies were deleted.
Those who can make you believe absurdities can make you commit atrocities. - Voltaire
When you see an advertised interest rate, this rate will be footnoted with the acronym "O.A.C.". Perhaps you should look up what that means...
Xenon, where's my money? -Borno
I cleared all cookies from Chrome, Opera, and FF, and went to http://www.capitalone.com/autoloans/
Chrome and Opera both gave me rates of 3.5%.
Firefox gave me 2.7%.
I proxy my FF sessions through a ssh tunnel to a personal Linux server I have, which is located in Michigan. So to see if that had an effect on it, I cleared cookies again and then disabled the proxy settings. Restarted FF and went to the URL again. This time I got a rate of 2.3%.
I went back to Opera and Chrome, deleted cookies and restarted again. Now Opera was 2.3%, and Chrome was 3.1%. Did it again, and Opera was still at 2.3%, and Chrome had dropped to 2.7%. It's as if this site is just coming up with totally random numbers.
I called the Federal 'regulators' responsible for forbidding such shenanigans, the phone would just ring and ring and ring.
Well there is always trouser trout.
There really doesn't need to be an argument for causation. Common causation resulting in correlation is perfectly adequate for insurance usage, firefox becomes a proxy for the root cause, which is probably management skills (e.g., you have poor management skills, so you can't keep yourself from getting viruses with IE, and wind up having to move to firefox).
"Who is the Journal of Quantum Physics going to believe?" --Stephen Hawking
That's entirely true. By selling their expensive, government provided cars, they would be significantly more likely to be able to afford college.
"Who is the Journal of Quantum Physics going to believe?" --Stephen Hawking
When you see an advertised interest rate, this rate will be footnoted with the acronym "O.A.C.". Perhaps you should look up what that means...
Ohio Athletic Conference?
...they just need to be right 51% of the time to stay in business.
I agree with your general point, however since this is a thread about correlation/causation, I'm going to point out that this statement is not really accurate. The profit of a company's good business decisions must exceed the losses from their bad decisions...how often those decisions must be correct depends on the nature of business. Consider an investor as a simple illustration: if one invests equally in 10 blue chip stocks that on average yield 10%, one bad decision (say one stock was WaMu, considered safe until it went belly up) can wipe out all of the profit from the other 9; on the other hand, one could invest equally in 100 different options or futures, and have 99 lose their entire value, but have the remaining one return so wildly that it makes up for all of the other losses. In other words, depending on the risk/reward balance of the business decisions in question, you can be wrong far less than half the time and fail, or you can be right only occasionally and still profit.
If you prefer, think of Microsoft...only a handful of their products/ventures make up the vast bulk of their profits, the majority only break even or fail miserably.
Momentarily, the need for the construction of new light will no longer exist.
only an idiot finds it necessary to make note of the obvious. and the idea that "correlation is not causation" is only a remarkable revelation to an idiot
Right, but the problem is that we have slashdot editors posting stories, and most of the top level responses are to those editors. So it might legitimately be a remarkable revelation.
"Who is the Journal of Quantum Physics going to believe?" --Stephen Hawking
I wonder if this has to do with how long people have used various browsers. People are probably less likely to default at the beginning of a loan than they are several years in, and all of the browsers listed except IE and FireFox have had somewhat recent large increases in use, thus increasing the percentage of Chrome and Safari users, in particular, that are in the first 12 months of their loans -- and thus less likely to have already defaulted.
Or it could be something else entirely, but that seems plausible to me.
A more practical example is the media's constant repetition that healthier people have more sex. Of course, they jump to the sexiest but dumbest possible conclusion, that the sex causes the health
Indeed, that makes less sense than "I really feel like shit, can we do it tomorrow?" How much sex is a cancer patient getting?
However, this is a correlation that may or may not have causation; if exercise is good for your health, than sex is too, because sex is excellent exercise, almost as good a swimming (very low impact aerobics).
So there could be a two way causation; being sick leads people to having less sex, and more sex makes one healthier. Unless you catch the clap or something...
Free Martian Whores!
Give the guy a break. He's not guilty that you're all bitter and smart. Peace
Put simply, correlation is a necessary, but not sufficient, condition for causation.
The Freelance Wizard
People who blindly yell 'Correlation is not causation' should be slapped with a trout.
What do I get?
In any case, I don't see why browser version is any less valid a pattern to use for insurance rate policies than age.
The data says younger drivers get into more accidents. They are therefore higher risk drivers, and thus by default get a higher insurance rate.
The data says Firefox users get into more accidents. They are therefore higher risk drivers, and thus by default get a higher insurance rate.
Now, it could probably be because Firefox users tend to be younger, and younger drivers tend to get in more accidents, and a rough guess for the cause of that correlation would be lack of experience, but that would require in-depth study and is far beyond what is needed to accurately set insurance rates.
So the cause is lack of experience, but the correlation is to age and browser usage. It could well be something completely different for FF, since I would think Chrome would correlate to young drivers also. The point is, knowing the root cause doesn't really matter. All you need to know is that they correlate in order to set up risk tables.
Security is mostly a superstition... Avoiding danger is no safer in the long run than outright exposure. - Helen Keller
You shouldn't shop for insurance online anyway. You need to talk to agents and shop around in person / over the phone to get any good estimate of how much you'll owe. Looking online is a good start for ballpark figures, but that's about it.
It would be an absolutely moronic input to use for your insurance rates. If firefox users tend to be younger, then just use the person's freaking age.
But this isn't even about insurance rates. It is about loan rates. So your point is even further off base.
You get slapped with a manatee.
See my journal for slashdot ID's by year. Mine created in 2005. http://slashdot.org/journal/289875/slashdot-ids-by-year
I don't know why you have a chip on your shoulder against users of the phrase or why you consider it useless noise.
I do.
It's because people use it without having any idea what it actually means.
Case in point, the first post on this thread.
The post is technically correct: Correlation does not imply causation. However, it completely ignores the fact that insurance companies could care less about causation. They aren't out to prove that using FireFox makes people get in more accidents. That is obviously not the cause of the correlation. All they care about is the fact that people who use FireFox get in more accidents than people who don't. That is absolutely true (supposedly), and it is all that is necessary to set up a risk table, upon which insurance rates are based.
Insurance companies also take your driving record into account. You may meet all the risk factors available (drive a red sports car, under 25, not married, yadda yadda), but if you've never had so much as a traffic ticket since you got your learner's permit when you were 14 you are going to have a pretty low rate. You may hit all the risk factors, but you've shown that you obviously aren't a risk.
Security is mostly a superstition... Avoiding danger is no safer in the long run than outright exposure. - Helen Keller
I think belthize rode the short bus to school.
I kid! I kid!
Security is mostly a superstition... Avoiding danger is no safer in the long run than outright exposure. - Helen Keller
Ahh Netscape.
Such a great browser... as long as the only thing you had to compare it to was Internet Explorer.
Security is mostly a superstition... Avoiding danger is no safer in the long run than outright exposure. - Helen Keller
I prefer banks paying me interest to the other way around.
Yeah, but where would we get a trout?
Where every good fisherman gets them, the store...
Learn to love Alaska
Stewie... is that you? Man that sounds just like Stewie.
Reminds me of that time that Peter ____________.
soylentnews.org Go there to enjoy the people!
Actually it very likely affects the risk tables.
A risk table is basically an educated guess as to what a particular customer will end up costing the company.
Pretty much any easily measured correlation is perfectly valid for a risk table. And the fact is, they work. It doesn't matter one bit why young drivers get in more accidents, the fact is young drivers get in more accidents. No need to find the cause, just charge young drivers more.
Same with browsers. It doesn't matter why FireFox users get in more accidents (it almost certainly has absolutely nothing to do with FireFox). The fact is, they get in more accidents, and are therefore a higher risk. It's a marker, and apparently a statistically significant one.
Security is mostly a superstition... Avoiding danger is no safer in the long run than outright exposure. - Helen Keller
It doesn't mean that all people using that phrase are unintelligent and using the phrase out of context. To support what I'm saying: "correlation is not causation"
I've not heard anyone on the Internet ever use it intelligently. You could say "whenever I drop a ball, it falls to the earth" and some moron will point out that correlation doesn't equal causation. Technically correct, but useless for all applications I've ever seen. Do you know how the scientific method works? I'll give you a little spoiler. Correlation == causation. If you can show a correlated link, explain that link, challenge everyone on the planet to prove one case where that link is wrong and no one can and no one can conceive of any possible situation where that correlation wouldn't hold true, then it is essentially accepted as fact. So there are indeed times where correlation does equal causation. So someone posting something known to be false to look smarter makes them look dumber. The only time it would be the slightest bit relevant is when they go on to list confounds or otherwise address the initial issue. But there is a strong negative correlation between those who use the phrase in question and those who offer any arguments in support of the phrase in question's application to the point at hand.
Learn to love Alaska
And your sarcasm is really not clever at all. I could go on and be as sarcastic as you to try to prove I'm clever, but I won't.
The point is they've earned the name by many local ex-employees. I didn't go into details, but did you ever see the site crapitalone.com while it was still up? It, along with the number of IT guys I knew that had worked there made the point very well about what kind of culture ran throughout the company.
But if you insist on just dwelling on what you think I thought I made up to be funny so you can prove you're oh, so clever, please, feel free to continue.
And of course Lynx users are all living in a cabin in the woods, so they don't need a car loan in the first place!
I've abandoned my search for truth; now I'm just looking for some useful delusions.
Actually, Apple used to charge $100 dollars more for the black MacBook that was in every other way identical to the white one, and even Dell charges more for colored notebooks. Obviously pricing is based on whatever the customer is willing to pay, not on any rational calculation of value.
I've abandoned my search for truth; now I'm just looking for some useful delusions.
I agree with the gist of what you're saying. However, they don't need outside, independent research for their purposes. They've given out loans before, and many people pay their loans online. It's quite easy for them to use their own internal data to see that people who use browser x to pay their loans default at y rate.
I doubt there has been much in the way of independent research for car accident statistics for my zip code, but insurance companies use zip codes for determining rates all the time. I'm sure Capital One's actuaries are up to the same kind of stuff with regards to loans. As to the actual interest rate, while the rate shown on a website isn't at all binding, there are still valid reasons for changing based on the browser. For example, let's say that their own data has shown that people who pay using Firefox are most likely to default, they can show a higher rate hoping that you'll go somewhere else instead of with them, so they'll be less likely to get a new defaulting loanee.
Stop Global Warming!
Just say no to irreversible processes!
The correlation of high rates to the letter "A" in that example was caused by the belief that it had some effect. There was a correlation and it had a cause. The fact that the cause was ludicrous is beside the point.
An example of correlation not implying causation would be if somebody sampled all rates and discovered that the letter "B" had the highest rate due to nothing more than random chance.
I understand the point you and the parent are making but in this context it's invalid. There is a correlation and a cause ... it appears higher rates do in fact vary by browser type and that's caused by the decisions of loan makers not random chance. Whether those decisions are in turn based on valid correlation is a different argument.
Can they do that, I mean i know they physically can, but legally? Is there not some sort of discrimination going on here that could result in a class action lawsuit?
There is a correlation and a cause ... it appears higher rates do in fact vary by browser type and that's caused by the decisions of loan makers not random chance.
Wait, I think this is the problem. Everyone else is arguing whether or not "your riskiness as a borrower" is either correlated to or caused by "the browser you use". You seem to be arguing that the variation in quoted rates is caused by companies choosing different rates. This may be why you think everyone else is crazy, and we're saying you're not making sense.
You are correct that the rates are changing because companies chose to change the rates, and not because they're just random. However, that's not the correlation/causation anyone else is trying to discuss.
I do also want to stand firm on the idea that a statistical correlation specifically means the items really are linked, and it's not just chance. The definition of "correlated" requires that the link between two items cannot be chance. There has to be a real link.
The Quirkz Handbook of Self-Improvement for People Who Are Already Pretty Okay
Compare the the exchange rate they charge to convert foreign currency to USD with a card that would charge the conversion fee...that needs to be considered in the cost as well.
//m
"Hello ladies.....take a good look at your man. Now back at me. Back to your man. Back at me. Unfortunately, he is not me. But with Google Chrome browser, he can get bank loans as if he were me. I'm in a bank, with loans for that thing you want. Now they're diamonds. Anything is possible when you use Google Chrome. I'm in a new car."
Who's saying it is? Correlation is really all insurance rates need to be based on.
Well, like any other price insurance rates are based on elasticity (PED). Risk just sets the break-even point; clearly if you sell at break even you're not making a dime, and insurance isn't really a true commodity where price is everything but has a fair degree of elasticity. Browser may be a risk correlation, but it may also be an elasticity correlation just as well... (In other words, Firefox users may be willing to pay more or comparison shop less than Chrome users, or have fewer options for whatever reason.) Look up "price elasticity of demand" - for some reason Chrome won't let me paste a link on Slashdot.
Technically this is about loan rates not insurance rates, but we could take it as self insurance by the lender against the risk of default.
To your point...
Correlation *in the future* is what insurance rates need to be based on. Correlation in the past may or may not imply this.
Causation would tend to imply correlation in the future, as would causation of both elements by a common cause that's unlikely to change quickly (say, a person's level of responsibility). Statistical happenstance does not imply future correlation - just because there was some correlation between people who flipped heads at the beginning of last year and default rates doesn't mean there will be next year. Causation by factors likely to change also doesn't imply things about the future, and can drown out other relationships.
The question is which of these are represented in the relationship between browser choice and defaults. Whatever was the case in the past, given how easy browser choice is to change (particularly for visiting one particular site) I expect in the future the primary cause of choice of browser will be knowledge of its influence on rates.
> I've not heard anyone on the Internet ever use it intelligently.
You haven't read enough slashdot threads yet, then. We get fairly regularly bombed by weak studies that make outrageous claims of "x and y are correlated, therefore x causes y". The typical (and CORRECT) response to most of these is indeed the quote you hate, followed by "how do you know y doesn't cause x? because you sure as hell didn't test for it" and "how do you know z doesn't cause both x and y? because you sure as hell didn't test for it".
Primary recurring example: studies of [anything] and violence. Because clearly Doom made the Columbine shooters into monsters, they couldn't possibly have already liked violence and that led them to Doom, right?
This is not to say that no one ever uses the phrase wrong. It's just to point out that a different set of people is continually failing at statistics and needs the phrase beaten into them by an ASCII 2x4.
> "whenever I drop a ball, it falls to the earth"
That preserves the time data, and thus has a vastly stronger argument for which is cause and which is effect. Way better than a poll of "have you ever done x? have you ever done y? Oh my god, one of these must be causing the other, even though we didn't ask which order you did them in and threw out all the results of people who only did one!" Or this actual topic's issue, where one guy did the test on one computer.
There was a pattern in the United States that home prices never dropped simultaneously across the entire country; not in 80 years. So if you packaged bonds together with a mix of mortgages from across the country, your risk was expected to be very, very low. You could even package sub-prime mortgages this way, safe in the knowledge that your income stream from one part of the country would always cover the income stream from another.
Of course, this particular pattern stopped being a pattern around 2007. Oops.
Can you imagine what the interest rate for Safari users is?
'The tyrant will always find pretext for his tyranny.' - Aesop's Fables
It isn't quite that straightforward.
Suppose that A correlates strongly with B and B correlates weakly with C. This would cause A to correlate weakly with C. If you're interested in A, you know B, and the correlation between A and C is essentially what would be expected given the A-B and B-C correlations, you would probably ignore the A-C correlation, as it won't tell you anything about A that you can't already deduce more accurately from B.
More precisely, if A correlates with B more strongly than with C, you would test the A-C correlation for specific values of B (disaggregation), and may well find that there is actually no A-C correlation when disaggregating for B. If you measure the A-C correlation without considering B, B would be a "confounding factor". The art is in identifying the most important correlations and disaggregating for them, so that you don't overestimate the effect of weaker correlations.
Obviously pricing is based on whatever the customer is willing to pay, not on any rational calculation of value
"What the customer is willing to pay for it" is the only rational calculation of value. Everything else is subjective.
Socialism: a lie told by totalitarians and believed by fools.
$ for i in {1..2000}; do curl https://www.capitalone.com/ 2>/dev/null | grep "as low as" -m 1 | cut -d ">" -f 3 | cut -d "%" -f 1; done | sort | uniq -c
420 2.30
499 2.70
428 3.10
653 3.50
So for 2000 samples (with the default curl user-agent) we get
2.3% (~21%)
2.7% (~25%)
3.1% (~21%)
3.5% (~33%)
Now let's try it again with a firefox for windows user agent.
$ for i in {1..2000}; do curl https://www.capitalone.com/ --user-agent 'Mozilla/5.0 (Windows; U; Windows NT 6.1; en-US; rv:1.9.2.11) Gecko/20101012 Firefox/3.6.11 Mozilla/5.0 (Windows; U; Windows NT 6.1; en-US; rv:1.9.2.11) Gecko/20101012 Firefox/3.6.11' 2>/dev/null | grep "as low as" -m 1 | cut -d ">" -f 3 | cut -d "%" -f 1; done | sort | uniq -c
395 2.30
536 2.70
450 3.10
619 3.50
So for 2000 samples (with the firefox user agent) we get
2.3% (~20%)
2.7% (~27%)
3.1% (~23%)
3.5% (~31%)
This deviation does not seem statistically meaningful to me, I would conclude that user agent is not relevant, at least not between firefox and curl.
"The crows seemed to be calling his name, thought Caw."
The typical (and CORRECT) response to most of these is indeed the quote you hate, followed by "how do you know y doesn't cause x? because you sure as hell didn't test for it" and "how do you know z doesn't cause both x and y? because you sure as hell didn't test for it".
You haven't read enough slashdot threads to see that such comments are vastly inferior in numbers than the unsubstantiated assertions of "correlation doesn't equal causation." Even worse, those comments you mention show up every time, often listing confounds that are explicitly stated in the article as having been identified and corrected for. Or, the article was a 5th party summary that didn't include anything useful and there are some pretty obvious confounds that everyone asserts proves them wrong because some inaccurate summary doesn't state that they were corrected for.
"Correlation does not equal causation" can be left off for all intelligent responses pointing out such effects. People who use it are invariably using it to look smarter, not to share some insight no one has heard of. And it's wrong more than right, so anyone who uses it has already dealt a blow to their own credibility.
That preserves the time data, and thus has a vastly stronger argument for which is cause and which is effect.
Still a correlation. Still a "law" of physics. And anyone that says "correlation does not equal causation" after that is a complete idiot that is, in fact, wrong. Since "correlation does not equal causation" can be proven wrong easily, then it is a nonsensical statement added with no value.
Learn to love Alaska
"There are no other obvious variables."
Yes, actually there are. From my testing (in a different post) I showed that you get different rates with each new load as long as you aren't sending cookies. Whatever he was doing to test cookie-wise was wrong (which should have been obvious considering he was re-installing browsers to clear cookies, wtf?).
So yes, this was absolutely a case of correlation != causation.
"The crows seemed to be calling his name, thought Caw."
You can be right 1% of the time if your margin is high enough.
I don't think there is necessarily any correlation between the number of years someone's been capable of taking out loans, and the browser they use.. though someone who uses a niche browser is probably someone that spends more time researching their options in everything they do.
which is totally what she said
So far I've just avoided getting any loans besides my student loan, which is a fixed thing in Scotland, the rates are the same for everyone.. so I really wouldn't know.
which is totally what she said
Richard Fairbank, is that you?
There's a SCRIPT tag which has been commented-out that uses "offermatica/mbox.js", which is the old name for an Omniture product facilitating Multivariate Testing (MVT) and A/B testing. Although this script has been commented-out, it shows that Capital One is familiar with the technique and likely experimenting on their traffic through other means (e.g., server-side rather than JavaScript). Note that Omniture has been acquired by Adobe, so expect to see this type of experimentation happening more frequently. Presumably, such functionality will be "seamlessly" integrated into their content creation frameworks. See whichMVT.com for a list of vendors.
Ah my karma for a mod point.
Are you being serious? Have you never secured credit before in your life?
Yes I am, and no I have not.
ANY lending institution is going to require a credit check prior to lending you any money
Sure, they can refuse a loan, but how is changing the interest rate going to help make things better, unless they actually lower it for people who usually struggle to make payments?
which is totally what she said
Actually, I've either bought my cars outright (used, of course), or I can use company cars if I wish :p
which is totally what she said
Well, now that I know you are being serious, I can hardly give you too much of a hard time since you haven't been through it yet. This is a very basic explanation of how it works.
In a nutshell, you're charged a higher interest rate if your credit score is lower (and thus, you're perceived to be more of a risk....that is...you might not pay for your stuff). So, they charge more to recoup the money they've loaned to you faster (in case you default) and so they can get more profit by taking the "risk" on you. It's sort of like investing in the stock market, but in reverse. The lender is investing in you as their stock, and the interest is their dividend.
Although it f*cked the US the people who did it still made out pretty well, so I would say it was a really good strategy for them even if it stopped being a pattern in 2007
As long as capital 1 stops this practice when it is no longer true they are being smart (asses, but smart)
www.RacquetUp.org - Helping Detroit Youth
And someone who uses safari probably payed a ton for a Mac (I am not judging whether it is worth it or not, whether a better product or not, so please dont start that discussion) and probably has more money to throw around, so less likely to need a loan, and thus would only do so for a low rate
www.RacquetUp.org - Helping Detroit Youth
US should delete amazon from the internet
XD
Not really.
Now if they could make funnier Vikings than Monty Python, THAT would count for something!
You lost me on that one
Safari users get a cheaper rate because thay are prepared to pay more????
"chrispederick.com,
The personal web site of Chris Pederick and home of the Web Developer browser extension"
Kind of like the "large Hardon Collider"
Apocalypse Cancelled, Sorry, No Ticket Refunds
Perhaps the sequence of requests coming from the author's IP address had an effect, not the user-agent. In other words, every time he applied from the same IP, they fiddled with the rates. Or perhaps they based the rate on how quickly he filled in certain fields, e.g. those who dilly-dally on the "annual income" field might be exaggerating, and will thus get a higher rate--or maybe a lower one, to entice them to get in over their heads! So the user-agent is simply one of many variables.
It was a Family Guy reference.
"And they'd tell you where to go." To Crapital One via Comcrap, no doubt.
Mod Me Up. You'll make a grown man cry.
Correlation is all that really matters to them.
Unless the correlation is with race, religion, national origin, or other "protected" differences.
Actually I'm sure the actuaries wouldn't have any problem using all the data available but he lawyers, public relations people and government regulators won't allow it.
Would you believe that some people think that even things with proven causation (e.g. pre-existing conditions) should be protected from pricing differentials.
I guess the car analogy in this case would be demanding to be given the same low rate on a car loan after telling the bank that the reason you're buying a new car is that your previous one has just been repossessed.
That group of bovine standing over there appears quite portentous. That's right it's an ominous cow herd.
If you get a loan from Capital One you're a moron anyways.
I'm pretty sure this violates discrimitation laws. Not all browsers have the same accessibility features (chrome has practically none while firefox has many). So since people with movement or vision impariments have to use the browser that gets the higher rate they are in fact giving them a higher rate based on their dissability. I would assume that they will now be sued by the ACLU for violating laws all over the country.
You don't even need causation in the mix for people to delude themselves. Correlation can be mishandled all by itself.
IIRC, the correlation between who wins the Super Bowl (AFC or NFC) and how the economy does next year is very very high. Would you make bets on the economy based on who wins the Super Bowl?
If you look hard enough, you'll find apparent correlations that were a short-term fluke. With enough variables, there *have* to be some coincidences.
Sometimes past results predict future results ... and sometimes not.
Of course, the insurance and auto companies are going to analyze whether or not a correlation is statistically significant -- chi-square tests etc.
About CarMax not having fixed vacation - I think it's super-clever from their side. I know that I constantly accumulate more vacation time than I take and I have hit the company cap sometimes and took time off just not to lose it - and if I leave they have to pay me all the vacation time I have accumulated. However if they don't have a fixed vacation - nothing of it applies, they can just rely on tech people being workaholics (which happens quite often) and not take too much vacation (if one does, you can always fire him...) - but they don't pay any additional money for unused vacation time! This is pure genius.
-- Si hoc legere scis nimium eruditionis habes.
Thanks for remembering. I wrote an article for the Fall 2000 issue of 2600 Magazine explaining how to change payment amount between the shopping basket and the checkout of a wine website. A friend claimed that he got a delivery of vodka from said website after dividing the total by 10. After 10 years, that claim remains unverified. However, the wine website, belonging to a national retailer, was taken offline shortly after the article was published. The article was also plagiarized extensively.
The article demonstrated that third party credit integration was placing too much trust on client software.