Interesting Pew Survey (2016 vs 2018)

Interesting data for political stats wonks.


(Update: Also see, “How Broad and How Happy Is the Trump Coalition?” Nate Cohn’s article mentioned below.)


Here are some fascinating Pew Research survey findings in the run-up to the “mid-year” 2018 election and backward look at 2016.

Why is it interesting?

The 2018 findings are from a tracking poll of the same individuals who participated in the 2016 Pew American Trends Panel. In other words, the same people who participated in 2016 also participated in this year’s panel. (Let me try again: It’s not a random sampling. It’s a survey of the same people who participated in 2016.)

While the write-up of the Pew findings is comprehensive, today’s NYT “The Daily” podcast has guest Nate Cohn of @UpshotNYT diving deep into the role of educated suburban women in both the 2016 and 2018 races.

Bottomline

Before this survey, the conventional wisdom has been that a core of “uneducated white rural male voters” is the key to Trump’s 2016 victory. While the Pew survey concurs that that demographic was a key member of the Trump “core,” it also reveals that educated women in the suburbs who were against Hillary Clinton were what sealed the deal for Trump. And, as this Pew chart shows, this cohort is the most likely voter to have become disenchanted with Trump.

“Still” is the new “on the other hand”

Many times, when I post one of these rants about the way reporters write about statistics (especially statistics about the economy), I refer (negatively) to the “on-the-other-hand” crutch nearly all business-related stories now include. History buffs will recognize this as a reference to a Harry Truman quote: “Give me a one-handed economist! All my economists say, On the one hand on the other.” Here’s today’s mini-rant:

This morning, the NYTimes.com is reporting on unemployment (or is it employment?) numbers that came out this morning. Here are the first three paragraphs:

The nation’s employers eliminated 51,000 jobs in July, the seventh consecutive contraction in the labor market, as the unemployment rate reached a four-year high, signs that the pressure on business owners and consumers is likely to continue.

Still, the decline in the job market has softened since the spring. The number of layoffs was less than the 75,000 that economists had expected, and the government said that businesses cut fewer jobs in June and May than previously reported.

Still, the nation’s unemployment rate has steadily moved higher. In July, it rose to 5.7 percent from 5.5 percent in June, its highest level since March 2004.

I’m a reader and I would like the New York Times to help me understand what those unemployment numbers mean. The article says unemployment has reached a four-year high — that sounds bad. But that’s not as bad a economists predicted — that sounds good. The nation’s unemployment rate has steadily grown — that sounds bad. But the government says that the previous two months were not as bad as they had earlier reported — that sounds good.

The only thing for certain is this: If you’re unemployed, it’s bad.

Still, I’m confused.

A trillion here and a trillion there

Google is oogling over its numbers today. Not the numbers related to financial performance, but the numbers that math majors get goose bumps about. Granted, I may be misinterpreting what exactly the Google blog post is reporting, but I think it says that when Google started 10 years ago, they could index 26,000,000 web pages over the course of about 2 hours. But now, they can simultaneously index 1,000,000,000,000 unique URLs on the web at once.

As I was not a math major, at first I was impressed with that 1,000,000,000,000 number. But then I read that the U.S. House of Representatives today passed a bill that enables a potential government rescue of Fannie Mae and Freddie Mac, who together own or guarantee about $5,200,000,000,000 of the nation’s $12,000,000,000,000 in mortgages. To accommodate that potential rescue, the bill raises the national debt limit to $10,600,000,000,000, an increase of $800,000,000,000.

While you may think all of those 0s make for some big number, I’ll remind you that Google’s name is actually a play on the number, googol, which is the number 1 followed by 100 zeroes. So, when you start wanting to be impressed by how many URLs Google can index or how much potential debt Congress can instantly commit you and me to, just remember: Compared with a Googol, a trillion is like, well, 1.0 × 10 to the negative 91st power, give or take a few rounding errors.

Or, for those of you who were better at art than math:

Here’s a trillion: 1,000,000,000,000

Here’s a googol: 10,000,000,000,000,000,000,
000,000,000,000,000,000,000,000,000,000,000,
000,000,000,000,000,000,000,000,000,000,000,
000,000,000,000,000.

When Google can google a googol URLs at once, then I’ll be impressed.

As for Congress, they may as well go ahead a raise the debt limit to a googol, as that’s where they seem to be heading anyway.

Zemanta Pixie

It’s 2008. Why hasn’t Internet advertising surpassed magazine advertising?

Four years ago, an article in the Wall Street Journal suggested Internet advertising would match magazine advertising by 2007 and blow past it in 2008. What happened?

The very short version: During 2007, almost $60 billion was spent on advertising that appeared in print while $11.31 billion was spent on advertising that appeared on the Internet.

The very long version: I don’t expect any readers of this weblog to remember a four-year-old rant I wrote (and here) about a Wall Street Journal article appearing in July, 2004. Screen grabbed on the left, the WSJ story carried the headline “Online Ad Dollars Set to Match, Then Go Ahead of Magazines (sub. required).” The article was based on a Jupiter Research report predicting that in 2007, Internet advertising spending would grow to $13.8 billion which, claimed the Wall Street Journal, “would match magazine advertising.”

My rant, which later became an article appearing in Folio: Magazine, was directed more at the Wall Street Journal reporter’s mis-interpretation of the research than it was at the Jupiter Research report. Their prediction was not really a comparison of Internet advertising to magazine advertising, merely their estimate of online advertising spending through 2007 and beyond. It was the Journal reporter who decided to mashup a comparison of future Internet advertising (based on Jupiter’s numbers) and its magazine number estimate.

However — and this was a major focus of my rant — the reporter (and Jupiter) failed to recognize that the Internet advertising prediction included all online advertising while the magazine advertising prediction excluded all business-to-business magazine advertising.

In my response to the article, I suggested that a better prediction of 2007 magazine ad spending would be the 2004 estimate by Veronis Suhler that $28.3 billion would be spent on magazine advertising (consumer and B-to-B) in 2007.

Fast-forward four years. Today, Advertising Age issued a report that included the actual ad spending (split by media) in 2007. As you can see in the Advertising Age pie chart below, $11.31 billion was spent on Internet advertising and $30.33 billion was spent on magazine advertising. Throw in the $28.22 billion spent on newspaper advertising and there was nearly $60 billion spent on print advertising last year.

Let’s break this down a bit. Let’s look at a comparison of the 2004 predictions vs. actual performance from Jupiter Research and Veronis Suhler regarding Internet and magazine advertising. As you can see on my comparison below, Jupiter over-shot their Internet advertising prediction while Veronis-Suhler undershot their magazine advertising prediction.

(Granted, Jupiter Research’s prediction during the most recent four-year span was dramatically better than their 1999-2003 prediction. In 1999, they predicted that online advertising in 2003 would total $11.5 billion compared to the $6.6 billion it actually hit.)

What does this mean? First, it means, (to quote a wonderful headline I saw this morning) “90% of all statistics can be made to say anything 50% of the time.” No doubt, the statistics in today’s report can be spun any way you want. I’ve spun them one way. Most bloggers would spin them in a way that suggests they are another nail in the coffin of the print medium. Frankly, the way headlines and intro paragraphs will be written can make most any statistics imply whatever you want — at least 50% of the time.

As for me, personally: I love Internet advertising. Without a doubt, it’s growing faster than any other form of advertising and I, personally, am benefiting from that. In 15 years, it has grown from zero to $11.3 billion, an amazing feat. However, my complaint is with the misuse of numbers by reporters and tech-oriented analysts — and, to be honest, just about everybody I know — to support a narrative that can be summed up in three words: Print is dead. As much as I love the Internet and all things digital, that narrative is probably not going to be true in the lifetime of anyone making that prediction.

Today’s narrative — as it was back in 2004 and 1999 and 1954 when TV was going to kill print and radio and movies — is that the Internet is going to bury all other forms of media one day. Today’s narrative is that Internet advertising is growing at a far larger percentage (which even a middle-schooler should understand is easier to do when you have a lower base on which to grow). Today’s narrative is that newspapers are going to be dead in, what, a year of so? Certainly, they won’t last for an entire decade, goes the narrative. According to Steve Ballmer, “…there will be no media consumption left in 10 years that is not delivered over an IP network. There will be no newspapers, no magazines that are delivered in paper form. Everything gets delivered in an electronic form.” (He later said he could be off on the number of years, claiming, “…If it’s 14 or if it’s 8, it’s immaterial to my fundamental point . . . “

Of course, he also said the iPhone would flop.

Personally, I am doubtful about the longterm viability of the kind of print product the national chains of newspapers produce. Outside the sports section, I find little of value or interest to me in my hometown daily churned out by one of those national chains. And as I’ve said many times on this blog, I think many business-to-business print publications that focus merely on the transactions of their industries will be replaced by online properties that can provide a better, more timely flow of such information.

So, yes, I do think print will constrict while the Internet grows — over time. But die? Not likely.

Designing the impossible – a simple chart that interprets a mountain of data

Artists Shan Carter and Amanda Cox created the interactive chart in the video I’ve embedded below. (Added later: There is now a permanent spot for the graphic on NYTimes.com: here.) It is appearing this afternoon on the front page of NYTimes.com and interprets exit polling data from the entire Democratic Party presidential primary campaign. My video is a quick screencast (a video screengrab) of me clicking through the tabs of data that are displayed in the graphic. (What I didn’t show was how a cursor-hover reveals data related to each state box.)

I believe it’s an impressive (dazzling) use of subtle interactive-animation and information design that effectively translates a mountain of incomprehensible data into an understandable statistical narrative. I’m especially struck by the way the NYTimes artists eschewed colors and relied on the animation and the relative placement of data to interpret the statistics. For a chart junkie (I confess), I’m impressed that it is strongly visual, yet at the same time, is very “un” info-graphic. This is serious information design porn, in other words. (For future readers, today is the last day of the marathon 2008 Democratic Party presidential primary campaign.)

Credits that appeared with the graphic: Source: Edison/Mitofsky exit poll Design: Shan Carter and Amanda Cox