This graphic shows average debt categorized by email provider. Average age for Gmail users is 33 and 47 for Comcast. Yeah, that sounds about right.
[via We Love Datavis]
What your email says about your finances | Ingenial
Looks like I’d better switch from gmail to comcast in the hope of reducing my student debt and increasing my credit rating.
My goodness…what utter crap!
You obviously saw straight through this as I did; noting the average ages of each… you mention gmail and comcast, but I’m confident you would keep working straight down with Yahoo coming in around 13.
ah… embarrassed – just noticed average age below the first graph. please strike prior comment
Very interesting, but a little iffy graphically. For the first graph, the Yahoo bar is a little more than half the size of the Comcast bar, even though the actual difference in credit scores is not nearly that. And the Percent of On-Time Payments graph is more or less meaningless.
yeah, i don’t know what’s going on there. i have a feeling the bar length originally included the shaded regions where the ages are. i find it’s best to take in all of these sorts of graphics witch a big bag of salt.
I would like to point out that for the “percent on-time payment” statistic all data points have three significant digits (e.g. comcast 98.7%) but msn only has two: 98%. Do you not have good enough stats on msn or do you actually mean 98.0% ?
Re-iterating the point made by @jonemo, I’m not sure I even see a sample size anywhere here. In the first section, this is particularly blatant.
The differences are made to seem really large although the middle three providers are only a few points from the mean. None of them are a full 30 points from the mean and the fact about $28b doesn’t quite put credit scores into context.
This is clearly not intended to be a rigorous analysis and I think the graph is asthetically pleasing. I just wish they gave some sense of what was significant from a statistical and practical standpoint.
Following up on my previous comment: Since the graphic is not by “you”, there is no point directing the question at “you”. I should have said: “…do _they_ not have good enough data or…”
A couple things I want to mention.
First, most people have more than one e-mail address. When I sign up for stuff online, I don’t give out my primary e-mail address (Gmail), but instead use one I’ve dedicated to online sign-ups and junk (Hotmail).
Second, if the average age of Gmail users is lower, it doesn’t surprise me at all that they have less total debt and less CC debt. This is because younger people are probably less likely to own a home (and thus have a mortgage), and have had less time to build up lasting debt over the years.
Note that the date is (1) self-reported, and (2) based on people registered with Credit Karma, so who knows how representative that is of the population at large?
That said, I just turned 53 and use Gmail as my primary email. so thanks for making me feel 20 years younger!
I agree with some of the other commentators. this is a nice viz of the average credit scores of Credit Karma users by email provider. You can’t draw any conclusions about the relationship between email provider and credit IN GENERAL based on this data.
Thanks for posting our infographic. Spirited discussion. A few answers to your question: this is not self report data, it is factual data from the user’s credit report. The sample size was well over 400K. And most importantly, we never suggested any causality. As a matter of fact. The original post has the following disclaimer:
“Our analysis is for informational purposes. The data shows correlations for a number of reasons and is based on averages. As anyone who has taken a statistics class knows, causality and correlation are very different.”
We just thought it was a fun set of averages.
Thanks for making this clear. I think your method reduces many of the potential biases mentioned above. Yet, you still don’t explain why you drop one significant figure on MSN. Dropping the tailing 0 reduces the accuracy from the interval [97.95 98.04] to [97.5 98.4].
A 400k sample size sets off both the “With a Big Enough Sample the Null Hypothesis Can Always be Rejected” and the 1936 Literary Digest Memorial “Sample Size Doesn’t Compensate for Sample Bias” alarm bells. How are your customers different from your competitors or people who don’t use a credit monitoring service?
It was 98.0%. The designer wanted to make it “pretty” as much as it was “interesting”.
I’ll be sure to use this next time I teach about spurious relationships. But yeah, it is kinda pretty.
Can’t reply to a second level comment so I’ll do it here. You are absolutely correct in that we have an inherent bias in that we only have data for people who monitor their credit through Credit Karma.
There are clearly spurious relationships in the data. For example, the higher student loan debt with Gmail maybe associated with the lower average age of those user. We tend to pay off our student loans by the time we are 50.
Since you all seem to like data so much, I would love to hear some recommendations on our next infographic. We have over 1.2MM credit reports with over 200 data points in each.
Soniq.org Links » What your email says about your finances | FlowingData
I’m wondering why those with ‘no mortgage debt’ are not counted in the data. I can possibly see not counting renters, but if you have paid off your mortgage one would think that would be a good financial sign.
They are counted in the data. They simply are not counted in the average mortgage debt numbers.
Thank you for pointing that out!
Recently I started using GMAIL more than Comcast… I guess I’m getting younger and getting more debt now!
What Your Domain Says About You « The 401st Blow :: Thoughts On Media
These differences are all statistically insignificant, aren’t they? In other words, aren’t the differences just flukes?
With the sample sizes used, I suspect a t-test will show the differences are statistically significant probably at the 99% CI. But as others have pointed out, these difference are due to the spurious relationship of the data.
Conclusions are very tough to draw from this data. Can you present this report broken down by age group instead (i.e. for 25 – 34 yos: 55% use Y!, 29% use Gmail, etc)?
and then also include the breakdown of all the other data included above ;)
Have you realized the first bar chart has not 0 baseline? For me it’s a major failure…
What your email says about your finances « Collections History
Become a member.
Learn to visualize your data.
From beginner to advanced.
What you get
We don’t all start our work days at the same time, despite what morning rush hour might have you think.
The data goes back to 1960 and up to the most current estimates for 2009. Each line represents a country.
There are rules—usually for specific chart types meant to be read in a specific way—that you shouldn’t break. When they are, everyone loses. This is that small handful.
Moving on from the most trendy names in US history, let’s look at the most unisex ones. Some names have …