Thursday, September 8, 2011

Truth versus MSNBC (about Ron Paul)

Last night after the Republican presidential debate, MSNBC conducted a poll online to see who viewers thought won the debate. It was a blow away victory for Ron Paul, below is a bar chart showing the Ron Paul win and also the bar chart that was run by MSNBC. Notice the incorrect proportions in the MSNBC chart showing Rick Perry and Mitt Romney much closer to Ron Paul, in votes, than they actually were. By the way the MSNBC chart was even more distorted early on. Bob Murphy caught a screenshot of the Big Lie.


(Thanks 2 John T Foster for the correctly proportioned chart.)

33 comments:

  1. I work with programmers and websites. Believe me when I say, the vast majority of people in the field are incompetent idiots. I can pretty much guarantee you it's just shoddy coding, not some crazy conspiracy.

    ReplyDelete
  2. Has anyone compiled a highlights reel?

    ReplyDelete
  3. @Wobbles,
    That may be so, but it seems that Ron Paul's numbers are disproportionatly skewed moreso than all the other candidates.

    @Bob,
    There is one one at DailyPaul.com.

    ReplyDelete
  4. WSJ's loaded poll...establishment shills:- http://online.wsj.com/community/groups/election-day-684/topics/following-debate-do-you-now?commentid=3019902#identifier

    ReplyDelete
  5. @Bob English, I hope so, I missed all but the tail end of the debate and would like to see how Ron did earlier on. I was pretty shocked at the final question they threw at Ron; they were practically lobbing softballs at everyone, then they tell a story that intentionally makes Ron look like the Grinch right as the time runs out. Could they be any more obvious with the blatant anti Paul bias?

    ReplyDelete
  6. Bob:

    http://ronpaulnews.net/2011/09/ron-paul-debate-highlights.html

    ReplyDelete
  7. http://www.youtube.com/watch?v=hxiPHP3-VOE&feature=player_embedded

    Check that out. Paul humiliated some little goon O Reilly sent to interview Paul, but he turned the tables on him by saying O Reilly isn't really a journalist. LOL

    ReplyDelete
  8. I don't know. It's far easier to get a poll correct in proportion than it is to skew a poll. In algorithms, it's a simple one-to-one ratio. To get a poll graph skewed would require intentional manipulation of the ratio. However, It is possible, but would still require manipulation, that they put a cap on any one candidates potential votes and that Ron Paul reached that cap. That said, it just wouldn't make sense to do that either.

    ReplyDelete
  9. MSNBC hates them all.

    ReplyDelete
  10. I also work in the industry, it's no uncommon to be told to edit certain bits of code from time to time to make it look fair.

    ReplyDelete
  11. @ wobbles

    I beg to differ. I have worked in the graphic arts field for 20+ years. I have done a lot of work for fortune 100 companies. Manipulating bar chart visuals is just the first attempt at obfuscation.

    ReplyDelete
  12. I am a programmer, and the first comment is absolutely incorrect. The same function would be used to create each bar in the chart, and as you can see, all of the other bars are correctly proportioned. A completely separate formula would had to have been used to generate the Ron Paul bar, as apposed to all of the others. It is obvious that this is intentional. The real question is, was it done by an overly opinionated programmer, or at the direction of someone in charge.

    ReplyDelete
  13. I'm a programmer also. It's possible it's incompetence, but bar-chart stuff is so mundane that you'd usually use an off-the-shelf object. If the numbers are displayed correctly, the bars should be correct.

    Of course, the chart could have been done by the climate "scientists" that Jon Huntsman so loves. After all, when their paleo tree-ring data diverged from the known, measured, data, they just dropped it from the temperature graph. In some versions of the graph they even added the satellite temperature data to the end and mathematically smoothed the boundaries to cover the evidence.

    However, 98% of climate scientists, who haven't double-checked the data -- and for many papers you can't even GET the data to double-check it -- agree that AGW is real.

    So are the bar charts....

    ReplyDelete
  14. And here I thought FOX owned the "Fair and Balanced" slogan!

    ReplyDelete
  15. I'm also a programmer and I agree with the last comment that it is possible that it is incompetence, but not likely.

    Last night I looked at the same poll (and the distribution of votes was similar to what is shown above), but the bar for Ron Paul was the same length as the next closest person's even though Paul had twice as many votes. Obviously they've changed it since then due to complaints, but not enough to make it entirely accurate. That doesn't seem like incompetence to me.

    ReplyDelete
  16. In my experience the Daily WTF is a pretty accurate picture of how competent an average programmer is. (http://thedailywtf.com/)

    @Doug: "It is obvious that this is intentional." – No offense but you can't possibly be a very good programmer if you can't even conceive of a single reason this might have happened in error. Either that or you're a flawless programmer who has never, ever had a bug in anything you've written, or seen the buggy code of an amateur.

    I understand the desire to believe we're being intentionally cheated, in this case it just doesn't seem very likely; especially if you look at the original version of the graph that showed Romney and Paul's bars exactly the same size. I would buy that when they fixed it they didn't push the programmers too hard to make sure it was 100% accurate.

    ReplyDelete
  17. Perhaps the chart is logarithmic.

    ReplyDelete
  18. I think the wrong question is being asked. The question should be - How real are any of these Online poll numbers anyway? I do not know about this one in particular, but on most of them you can just do a "Quick Clean" and vote as many times as you want. My hunch is that Ron Paul's continuously good showings in the Online Polls, reflect the intensity of his supporters and their computer ability, as much as anything. It does produce a "Feel Good" effect , however.

    ReplyDelete
  19. It doesn't take a programmer to develop an accurate chart. that is absolutely absurd. the data can go into Excel, produce a chart, convert to an image. viola. Accurate chart for a website. I'm no rocket surgeon but I work in Excel all day, every day for an asset management firm. If I can do it, anyone can do it. Even a programmer or an MBA.

    ReplyDelete
  20. Are you really shocked? ...Or are you just pretending you're shocked?

    Did you ever have a conversation with these journalist majors at the university you went to? They were all lying and moronic statist brats...Mostly from public grade schools.

    ReplyDelete
  21. This is not a coincidence. If it was, the skew would not be against RonPaul 100.00000% of the time. That's just obvious.

    And yes, it requires extra effort and human intervention to distort the graph than to present it accurately.

    ReplyDelete
  22. Ron Paul only candidate to provide substance, says Washington Times piece:

    http://communities.washingtontimes.com/neighborhood/political-pro-con/2011/sep/8/ron-paul-rick-perry-debate/

    ReplyDelete
  23. @wobbles, f'getaboutit. Whoever did the graphics would have used off the shelf charting software and edited it to get that lying graph. Lordy, wake up.

    ReplyDelete
  24. I can guess exactly what happened. An editor said, "Look at that! It looks like Ron Paul is running away!" and arbitrarily cut his bar off. Certainly not a programming error, because the number is correct ... a flat out (mind-numbingly misleading) editorial decision.

    ReplyDelete
  25. The MSNBC Bar charts are just a simple HTML error.

    http://www.whosplayin.com/xoops/modules/myalbum/photo.php?lid=909&com_id=87931&com_rootid=87931&#comment87931

    ReplyDelete
  26. Also a programmer and it's painfully obvious that the number used to display Ron's row is exactly half of his displayed total vote count. Very simple and intentional manipulation. As further proof of this, the Daily Paul had a screenshot of the poll when Ron was at 40% and Mittens was at 20%... their bars were the same length.

    ReplyDelete
  27. St. Clair: No, it wasn't. The number used to set Ron Paul's bar was "100%". That was the correct number to use for the longest bar in the chart. The problem was that the stylesheet limited every bar to a maximum width of 300, but failed to take into account that the percentage widths applied to the width of the entire table, including the labels and numbers.

    ReplyDelete
  28. They should ask if you watched the debate who do you think won. I think people who already liked Ron Paul more than the others voted and said he won. That said, I like Ron Paul and his ideas too. If I had watched I would have voted the same way. I think the poll is more a measure of the popularity of his ideas, who people like best, which is still a big deal, and a significant result to consider.

    ReplyDelete
  29. Anyone ever consider that a designer or editor looked at that and said "The top bar is way too wide for the content, too much white space, shrink it!" and a lazy programmer simply chopped off the top line to make it fit? As a programmer, I see this kind of crap all the time, and it's no conspiracy, just laziness and not thinking.

    ReplyDelete
  30. Lies, damn lies, statistics, and...CHARTS!

    ReplyDelete
  31. If you look at the right margin, it's clear what happened. Basically, the bar was truncated to fit the chart size. Would have been just as easy to normalize it, though, so I'm not sure what the score is. It's probably a little bit of mischief and a heaping helping for incompetence.

    ReplyDelete
  32. This is a software problem because --

    1. You have to truncate the top bar or it won't fit the others fit fine.
    2. No one would bother to fudge the graph and show the correct numbers, they would just fudge the numbers which cannot be checked at all.
    3. The charts display perfectly on Android (my phone).
    4. If you look closely it's using off the shelf object that pre-dates the poll.



    If that's not enough reason for you then I think you're just being stubborn.

    ReplyDelete