Opinions, Context & Ideas from the TPM Editors TPM Editor's Blog

A Sight to Behold

As you can see, three successive mountains of boom and bust -- Bachmann, Perry and Cain. It's a veritable Sierra Nevadas or perhaps a mini-Himalayas. And if you look really close, Newt has just eeked into first place, presumably set for his own rollercoaster ride. Hover over the graph with your mouse and you'll get an even clearer look, with labels showing each one surge and collapsing.

Next, New Hampshire. Here the pattern is the same, with the big exception that Romney has always held a big, big lead. First, the Perry surge, then Cain. And now Newt moving into 2nd place.

And finally the national race.

Here again, Bachmann's move in the polls is more muted. She never really got very far except in Iowa, which is a neighboring state. But again, Perry, Cain and now Newt slipping into the lead.

The pattern is painfully obvious at this point. But I'm still not certain it will play out quite the same way for Newt. What do you think? And let us know what you think of the new graphs, what you do or don't like.

Late Update: First, thanks for the kind words and the feedback. The most common question/critique was, why do the colors for each candidate change from chart to chart?

Very good question. Lemme try to answer. First, certainly in a case like this where I've presented several charts to together that would make it easier to visually process the data. And we may create a feature that allows us or you to do this. But let me explain a little background on why this doesn't happen automatically.

These are not one-off charts we've assembled. TPMPolltracker is data application that contains virtually all political polling data above the state legislative level going back to 2006. The app automatically generates charts for all races, contest and questions it has data for. And it chooses colors based on the frequency of data for different candidates or answers. So since each candidates does a little differently in each of these contests (both in levels of support and frequency of being polled) the app assigns them different colors in different contests they're involved in -- that's to optimize for using the most primary/clearest colors. (I realize this gets a little complicated. But that's the answer.) And the application doesn't know that I've arranged them near each other like this.

Now, there are some cases where certain colors are assigned. So for instance whenever there's a major-party partisan match up, the Democrat is blue and the Republican is red. You can see that here with the chart for the Massachusetts senate race. If there's a third party challenger in a party race, the challenger is assigned yellow. Similarly, in a chart for approval or favorability, "good" is always dark grey and "bad" is always red -- as we have here with Obama Approval.

The way to get around the changing colors like in this post would be to create a function that would allow the user to 'force' the assignment of colors to different candidates. I think we'll likely do that. But probably not for the first release.

You can already do various kinds of filtering, by removing classes of pollster (phone, robopoll, online poll) from the set or by removing individual pollsters. Click the filter button for that and other customization features.