Fun with Graphs!

Earlier today, juliewolf wrote a scathing diary criticizing a dishonest graph posted on MyDD's front page.  In the spirit of juliewolf's brilliant excoriation of the dishonesty that has become so typical around these parts, I'd like to have some fun with graphs!

Here are some graphs I came up with.  All of the data is legit.  I will use them to distort the truth and mislead my readers, while maintaining that I am being honest because the data itself doesn't lie.

(graphs below the fold...)

-----

"Did you see how big Obama's delegate lead is?  It's huge!"

-----

"Hey -- it looks to me like the votes-per-delegate counts are basically equal."

-----

"Wow, even though there's no such thing as the popular vote in the Democratic primary process, it looks like Obama is way ahead of Hillary!"

"Oh, I forgot -- Since the 'popular vote' is not a real number, I gave Obama an extra 400,000 votes because Edwards endorsed him."

-----

"Look at how many contests Barack Obama has won.  It's clear that he's absolutely destroying her!"

-----

"Considering how few remaining undeclared superdelegates there are, it looks impossible for Hillary to regain her lead among the supers."

Update [2008-5-26 0:9:18 by chinapaulo]:

Holy smokes -- top of the reclist and a veiled dig from Jerome on his last post! That seriously makes me so happy. Fifteen minutes, indeed. Thanks a ton!

If I can be at all objective, I'd like to say that this is one of the most enjoyable and least vitriol-filled posts I've ever read here on MyDD. Though I wish more Clinton supporters would come along for a laugh. I will give double phantom mojo (still can't rec or rate) to any reasonably deceptive graph that uses actual statistics to show that Obama is losing somehow. Perhaps a little laughter is the start to the kind of unity that's been talked about here for a while?

Tags: Barack Obama, clinton, Graphs, Hillary Clinton, obama, primary campaign, statistics (all tags)

Comments

61 Comments

Re: Fun with Graphs!

Come on and join in!  Post your own graphs and show us how to mislead with statistics!

Consider this an equal-opportunity diary where you can use whatever misleading image you want to support a half-truth or outright falsehood about your candidate or 'the other guy'.  I want to see some creativity here.

by chinapaulo 2008-05-25 01:48PM | 0 recs
rec this up and tip jar the poster folks

by kindthoughts 2008-05-25 02:41PM | 0 recs
Re: Fun with Graphs!
Oh man that gave me such a good laugh! :p I'm so glad you posted this diary, because I made the same "WTF?" observation when I initially saw those earlier graphs that so conveniently placed the baseline where it was. Kudos. But you forgot this one...

Photobucket
by Rhizomorph 2008-05-25 05:00PM | 0 recs
Re: Fun with Graphs!

Best comment on this site in ages!!

by tibbs 2008-05-25 06:42PM | 0 recs
Re: Fun with Graphs!

Hey -- this is a great graph, but it's actually completely honest.  Though I suppose it gets deception points for not having much context (ie. the number of supers who have switched.)

Well done!

by chinapaulo 2008-05-25 07:20PM | 0 recs
Re: Fun with Graphs!

What a fun diary. Rec'ed.

by JENKINS 2008-05-25 08:26PM | 0 recs
Re: Fun with Graphs!

Excellent diary.

by Bobby Obama 2008-05-25 01:51PM | 0 recs
This should top the rec list.

I look forward to Jerome using each of these graphs in future front page analysis.

by Firewall 2008-05-25 01:53PM | 0 recs
Re: Fun with Graphs!

I didn't think I could enjoy anything more than juliewolf's courageous smart and simple takedown on the things that were posted in that regard but THAT was fabulously funny

by Benjaminomeara 2008-05-25 01:53PM | 0 recs
Re: Fun with Graphs!

I love having fun with graphs! Here's my contribution:

by ragekage 2008-05-25 01:54PM | 0 recs
Re: Fun with Graphs!

Genius!:D

by animated 2008-05-25 02:09PM | 0 recs
Re: Fun with Graphs!

Hold on - the yellow part which resembles a pacman ONLY resembles a pacman because of the purple part that doesn't.

My colonising imperial British brain can't work this out - but are you saying: Obama resembles Obama because Hillary doesn't?

Kind of true in a recursive kind of way. Maybe Obama supporters should be even more grateful for her candidacy, because she makes Obama look better. And to Jerome, for making you and others post such great graphics.

by duende 2008-05-25 02:39PM | 0 recs
"Pac-man"?

Is that some kind of game you old people play? I'll send you a bootleg of GTA IV.

by edg1 2008-05-25 06:07PM | 0 recs
Re: Fun with Graphs!

Bravo!

by Reaper0Bot0 2008-05-25 01:56PM | 0 recs
Re: Fun with Graphs!

Jerome is peddling his pink-sheet stock rec deception game again.

by neonplaque 2008-05-25 02:12PM | 0 recs
Re: Fun with Graphs!

These hip hop info graphics crack me up:

by animated 2008-05-25 02:14PM | 0 recs
Re: Fun with Graphs!

That's quite brilliant.

by minnesotaryan 2008-05-25 02:35PM | 0 recs
Priceless! n/t

by sricki 2008-05-25 03:53PM | 0 recs
If I remember my Chapelle right...

"Wu-Tang Clan ain't nothin to fuck with!"

by Student Guy 2008-05-25 06:49PM | 0 recs
You know what...

73.5 per cent of all statistics are made up on the spot.

Great diary. Fun graphics. So glad I've got my rating and rec'cing ability back to...

rec this!

by duende 2008-05-25 02:35PM | 0 recs
Re: You know what...

I want mine back :(

by igottheblues 2008-05-25 04:49PM | 0 recs
Re: You know what...

It's actually 84%. WHY ARE YOU SO DISHONEST???!11!?!!!?1!

by CrazyDrumGuy 2008-05-25 06:52PM | 0 recs
Re: You know what...

From Todd Snider's song, Statistician's Blues:

They say 3 percent of the people use 5 to 6 percent of their brain
97 percent use 3 percent and the rest goes down the drain
I'll never know which one I am but I'll bet you my last dime
99 percent think we're 3 percent 100 percent of the time

64 percent of all the world's statistics are made up right there on the spot
82.4 percent of people believe 'em whether they're accurate statistics or not
I don't know what you believe but I do know there's no doubt
I need another double shot of something 90 proof
I got too much to think about

by fogiv 2008-05-25 10:52PM | 0 recs
Re: Fun with Graphs!

Here's a good one.  Obviously its incredibly deceptive.  Sure, it might show chemistry students the actual temperature at which metallic elements boil with a fair amount of precision.  But if those poor students don't bother to look at the axis, they might think that nickel takes more than twice the heat to boil as tin!  

As a tin supporter, I feel that this is an outrage.

by Fuzzy Dunlop 2008-05-25 02:39PM | 0 recs
Re: Fun with Graphs!

And where do those values fall within the range of boiling points of metals overall? Even more importantly, where do those average values fall within the range of values for each sub-population of metals being shown? If we pull a random sample of  nickel and a random sample of tin, is it likely that the tin will boil at a higher temperature than the sample of nickel, contradicting this chart? If we use this chart as a visual substitute for running an ANOVA analysis, will we be correct to deduce that tin and nickel do indeed melt at significantly different temperatures?

by letterc 2008-05-25 02:47PM | 0 recs
Re: Fun with Graphs!

"And where do those values fall within the range of boiling points of metals overall?"

Could you better answer this question with a graph scaled 0-6000?  No.

"Even more importantly, where do those average values fall within the range of values for each sub-population of metals being shown?"

They are not average values.  They are properties of the elements.  

"If we pull a random sample of  nickel and a random sample of tin, is it likely that the tin will boil at a higher temperature than the sample of nickel, contradicting this chart?"

No.  But you shouldn't need a graph to tell you that.

by Fuzzy Dunlop 2008-05-25 03:09PM | 0 recs
Re: Fun with Graphs!

Yes, I am aware that including zero doesn't make the graph clearer. I have pointed that out in several comments. That doesn't mean that you can scale your axes how ever you please.

On the second two points, you prove my point. This graph is not misleading because it correctly captures key features of the data. The graph that Jerome highlighted did not. The variation in how many primary voters each delegate represents is much larger than the difference between the average number of voters each Clinton delegate represents and the average number of voters each Obama delegate represents. If you pick a delegate at random, checking whether they are an Obama delegate or a Clinton delegate will not tell you whether they represent a larger or smaller number of voters. This makes them completely unlike tin and nickel, where checking whether the sample is tin or nickel will tell you its melting point.

It is possible that the voters per delegate for Clinton and the voters per delegate for Obama are statistically different using ANOVA, which is what the chart Jerome used implies, but it is not a certain thing. The chart Jerome used strongly implies that it is a certain thing.

by letterc 2008-05-25 04:12PM | 0 recs
Re: Fun with Graphs!

"It is possible that the voters per delegate for Clinton and the voters per delegate for Obama are statistically different using ANOVA, which is what the chart Jerome used implies, but it is not a certain thing. The chart Jerome used strongly implies that it is a certain thing."

I think this is where our disagreement lies.  

One would use ANOVA or some other kind of difference of proportions test to see whether you might reasonably conclude that the means of two subgroups generated by random sampling were different.  I agree with that wholeheartedly.

But this data is not generated by random sampling from a larger population.  We didn't select a sample of delegates, each corresponding to different numbers of votes, as a way to estimate the mean number of votes per delegate for the two candidates.  Instead, we are looking at the entire population.  The question of whether differences in the mean number of votes per delegate among the two candidates is an artifact of chance thus makes no sense.  It is a certain thing.  

The most important question thus becomes how big that difference might be.  In my opinion, using a graph for that was probably the wrong choice to begin with.  But if we want to display that information in graphical form, we probably want a strategy that effectively conveys the size of that difference using units which are readable.

by Fuzzy Dunlop 2008-05-25 05:02PM | 0 recs
Re: Fun with Graphs!

You are right that we can conclusively state that the means are different, since we have the whole population, but without running an ANOVA, we can't state whether or not the difference is one that would be likely to occur if we simply split the delegates into two groups at random.

But we did divide a population of delegates into two sub-populations based on a particular characteristic (are they pledged to vote for Clinton or Obama?). We are now trying to judge whether another characteristic of those delegates (how many primary/caucus voters do they represent?) is significantly different between the two sub-populations. If we divided the full population up into two groups at random repeatedly, how often would we see differences in the average number of voters per delegate that were larger than the difference that we see in this case. If it is very rare, then we can conclude that there is probably a connection between the reason we divided the groups up along these lines and the fact that there is a difference in how many voters each delegate represents. If it is common for a random division to produce a larger difference, then we can't conclude there is any connection.

A single bar graph can't represent this well, no matter what the axis. A whisker plot or a histogram of the two groups would probably be better.

Now if the point of interest is that there is some difference in voters per delegate, then a histogram of voters per delegate would be more effective.

by letterc 2008-05-25 06:18PM | 0 recs
Re: Fun with Graphs!

It's an example of something better expressed in a table.  

by kellogg 2008-05-25 03:17PM | 0 recs
Re: Fun with Graphs!

I agree with that.  Graphs are overused in general and tables are usually much better at conveying information.

I don't see hip hop tables getting much viral internet traction though.

by Fuzzy Dunlop 2008-05-25 04:01PM | 0 recs
Thanks for the chart.

Next time my wife asks for gold jewelry, I'll use it to demonstrate that nickel jewelry is a hotter choice.

by edg1 2008-05-25 06:14PM | 0 recs
Re: Fun with Graphs!

I'm actually in complete agreement with the sentiment of your post.  It's not always deceptive to show data in a graph where the y-axis doesn't start at 0.  It's really the context in which the data is presented -- hence my comments in quotes after each of my accurate graphs.

In Jerome's case, under the graph, he wrote "Is there any doubt that the nominating process needs a serious overhaul for 2012?"  This implied that the apparently-huge discrepancy in size between the Obama and Clinton bars was prima facie evidence of the obvious inequity of the Democratic Party's nomination process.  In his defense, I can't prove his intent to deceive, but at first glance, the graph fooled me -- and I'm at least not a total idiot.

by chinapaulo 2008-05-25 08:34PM | 0 recs
Re: Fun with Graphs!

You, Mr. Diarist, win teh internets.

by DeskHack 2008-05-25 02:45PM | 0 recs
Re: Fun with Graphs!

OMFG, this means so much to me.  I accept this award on behalf of all aspiring winners of the internets, the in+erw3bs, or the google.  I promise to take seriously the task of bestowing this award on the next deserving recipient, should I decide to give it up.  This is the best day of my life.  It's all mine! mine! mine!

by chinapaulo 2008-05-25 02:58PM | 0 recs
Re: Fun with Graphs!

The hip-hop graphs were my favorite thing on the internet for MONTHS.

by ihaveseenenough 2008-05-25 02:54PM | 0 recs
Re: Fun with Graphs!

I missed those!  Damn, they're all fantastic.

by minnesotaryan 2008-05-25 03:00PM | 0 recs
Re: Fun with Graphs!

This is officially the best diary in the history of the internets!

by Veteran75 2008-05-25 07:09PM | 0 recs
His Noodly Appendages guide us all

On a slightly related note, Obama just gained 1,000,000 new votes from my vast ocean of multiple personalities.  The DNC originally ruled that my brain vote would not "count" (they say popular vote is nice, but irrelevant), but I'll be outside on May 31st picketing them for disenfranchising my other selves. As long as no one around here appears to care about rules, let's go ahead and add a seventh version of the popular vote metric!

Popular Vote Total:
16,685,941    49.1%    16,227,514    47.7%
Obama +458,427    +1.4%

Estimate w/IA, NV, ME, WA
17,020,025    49.1%    16,451,376    47.5%
Obama +568,649    +1.6%

Popular Vote (w/FL)
17,262,155    48.3%    17,098,500    47.8%
Obama +163,655    +0.5%

Estimate w/IA, NV, ME, WA
17,596,239    48.3%    17,322,362    47.6%
Obama +273,877    +0.7%

Popular Vote (w/FL & MI)
17,262,155    47.5%    17,426,809    47.9%
Clinton +164,654    +0.45%

Estimate w/IA, NV, ME, WA
17,596,239    47.6%    17,650,671    47.7%
Clinton +54,432    +0.15%

Popular Vote (w/FL & MI & Ryan's selves)
18,596,239      49.0%   17,650,671      46.6%
Obama +945,568         +2.4%

WOW!  A landslide!  Supers!  Someone email them, because I obviously don't know how.

by minnesotaryan 2008-05-25 02:57PM | 0 recs
Re: His Noodly Appendages guide us all

Holy crap, this is a great new metric.  I can;t believe that Hillary Clinton would willfully disenfranchise 999,999 of your personalities.  I'm literally sick.  Someone ought to let RCP and a certain highly-rec'd Clinton supporter know about this metric!

I'd mojo this post a million times if I could.

by chinapaulo 2008-05-25 03:02PM | 0 recs
Re: His Noodly Appendages guide us all

I just read this post again and couldn't stop cracking up.  Totally boss.

by chinapaulo 2008-05-25 03:05PM | 0 recs
Re: Fun with Graphs!

Reasons to use numbers

1. The math is definite and it determines the outcome.

2. The math is definite and the outcome benefits you.

3. The outcome does not benefit you and #1 is fungible.

4. The math is indefinite and the outcome is in question.

Obama wins on #1 and #2.  #3 is meaningless because of #1.  Clinton and her supporters are relentlessly attempting to argue that #4 is in play.

If you want to seat FL and MI, you have conceded that the math is definite; therefore #4 is not a legitimate argument.  

The MI delegation cannot be sat, because virtually all the candidates did not put their names on the ballot. It is almost a certainty that the MI delegation will be split 50:50.  

The FL might have a chance at being sat but its unlikely that FL will be sat with any better results than MI.  So don't count on a big  improvement for Clinton if it is done.

If you think the total votes metric will work on Super Delegates, you'd better concede that caucuses are legitimate (to date the Clinton campaign has complained endlessly about this nominating devise because it has been a loser for her).  

You might also want to have a pithy response for your pathetic 4-12-1* against Obama in open primaries.  After all, in the general election; everyone votes.

Its not an easy road; its uphill into the wind and there are potholes everywhere

*Clinton won Arkansas, Indiana, Tennessee, and Texas.  Michigan does not count at this point. Obama won the remaining 12.

by kmwray 2008-05-25 03:46PM | 0 recs
ziiiiiiiing!!!!

n/t

by hootie4170 2008-05-25 04:27PM | 0 recs
Holy crap...

I would so love to rec this diary for too many reasons to mention.  Please, accept my phantom rec and mojo!

by igottheblues 2008-05-25 04:46PM | 0 recs
Re: Fun with Graphs!
92primary

If you missed that, Clinton only had 310 votes per delegate won, compared to Brown's 6830


by proseandpromise 2008-05-25 06:25PM | 0 recs
Re: Fun with Graphs!

Hah, Jerome doesn't need graphs to be dishonest.  He's dishonest each and every time he questions Obama's position as the front leader rather than declaring him the presumptive nominee.

by obscurant 2008-05-25 06:42PM | 0 recs
no flames please....

i am keeping in the spirit of the diary

by canadian gal 2008-05-25 08:02PM | 0 recs
Re: no flames please....

canadian gal -- as promised, please accept my double phantom mojo!

Your graph doesn't quite hit at the geeky math humor I was going for, but in all fairness, as an Obama supporter I recognize in myself all of the sentiments lampooned therein (as ugly as some of them may be).  Let it not be said that I'm unwilling to accept a little kick in the ribs.  :)

So cheers!

by chinapaulo 2008-05-25 08:22PM | 0 recs
Re: no flames please....

I'm an Obamaniac and I've been in all those places.  But it's really just because you Clintonistas are so mind-boggling.  :P

by proseandpromise 2008-05-25 08:44PM | 0 recs
Re: no flames please....

I support Obama, but mojo for you for venting in a humorous and creative way while observing the spirit of this fabulous diary.

by Roberta 2008-05-25 08:56PM | 0 recs
To quote Sasha Baron Cohen in

character, Very Niiiice!

mojo for you.

by Student Guy 2008-05-25 09:46PM | 0 recs
After careful analysis....

How to be a Clintonista:
Photobucket

:) no offence.

by grass 2008-05-25 10:49PM | 0 recs
Re: After careful analysis....

lol!

I love how after steps 1-4, it starts over again -- as if those steps will change the outcome.  Humor is, perhaps, the best way to convey the truths that really sting.

by chinapaulo 2008-05-26 09:35AM | 0 recs
Re: After careful analysis....

Remove the "Yes" option and it will be perfect:D

by animated 2008-05-26 12:25PM | 0 recs
Re: After careful analysis....

Oh visio, how workflows have made me hate thee...

by minnesotaryan 2008-05-26 03:40PM | 0 recs
Re: no flames please....

Wow.  I'm an Obama fan, and that's damn funny.  I'm LMAO and at the same time, just, ouch.  

Thanks for putting it out there in hostile territory, as it were.  I respect that.  No flames here.

by Matt Smith 2008-05-25 11:32PM | 0 recs
Re: no flames please....

"I'm LMAO and at the same time, just, ouch."

My sentiments exactly.  As a 110% Obama supporter, I can definitely say this one hits a little too close to home.

by chinapaulo 2008-05-26 09:23AM | 0 recs
Phantom Recs all around!

Thanks cg and everyone else.  This is fun.

-chris

by chrisblask 2008-05-26 06:42AM | 0 recs
thanks for this!

I needed a good laugh!

by obsessed 2008-05-25 10:58PM | 0 recs
Re: Fun with Graphs!

Brilliant diary, as was juliewolf's, and this is coming from a social scientist who lives and breathes statistics!

Jerome's claim that it's our fault that we can't read his perfectly innocent front page graph is horrifying.  I can read graphs fine; but graphs make visual arguments out of numbers, and the intention to deceive with that original front page graph was completely transparent, worthy of Hillaryis44 or TaylorMarsh, and a disgrace on MyDD.

Thanks, and rated while I still can, because I am sure I am not long for this world on MyDD, or "My Deep Denial," as another poster called it a while back.  

by fightbull 2008-05-26 05:39AM | 0 recs
If I could rec this 1,000,000 times

I would spend my Memorial Day doing so.

Jerome's petulant little mockery of juliewolf was exactly the sort of juvenile response that is causing this site to lose any shred of credibility it had remaining.

You've done a magnificent job of turning it right back on him.

by bookish 2008-05-26 07:13AM | 0 recs

Diaries

Advertise Blogads