Emails, Deen Freelon, associate professor of media and journalism, University of North Carolina-Chapel Hill, July 20, 2017

On 7/20/2017 2:42 PM, Selby, Gardner (CMG-Austin) wrote:

Professor Freelon:

Your colleague nudges me in your direction below, also offering her own thoughts.

I seek your focus on the accuracy of a recent claim by Mark Cuban. You can see my question (and Cuban’s claim) below Dr. Clark’s reply.

On-the-record analysis?

THANKS,

g.

From: Meredith Clark

Sent: Thursday, July 20, 2017 1:28 PM

Hi Gardner:

...For my own work, I usually refer to Pew's numbers.

However, I'm not a quantitative scholar. My colleague, Deen Freelon (dfreelon@gmail.com) is a Big Data expert. He's the person you want for a more scientific breakdown of the numbers.

I get where you're going with this analysis, but conceptually, it's a little more complicated. Cuban's claim rests on the the assumption that everyone has an equal probability of seeing any other user's tweets, which we don't, because of Twitter's algorithms and our own decisions on whom to follow, re-tweet, mute, block, etc.

And Twitter's numbers don't communicate "reach," either. No matter the number of "active" users on the platform, simply logging in to Twitter doesn't necessarily mean you're 1) an actual person and 2) reading anything posted there. The active word phrase here is "potential reach."

A better metric might be the average of unique impressions Trump's tweets get from users in the States. Perhaps the company

From: Deen Freelon

Sent: Thursday, July 20, 2017 1:55 PM

To: Selby, Gardner (CMG-Austin)

Subject: Re: Austin reporter, urgent inquiry for a fact-check about Twitter

Gardner,

I agree with Meredith's analysis. Here are some other points to consider:

• Pew, who has the best numbers on this, reports that 24% of American adults that use the internet regularly are also regular Twitter users. That's 21% of all American adults. (http://www.pewinternet.org/2016/11/11/social-media-update-2016/)
• Philosophically, your rating of Mr. Cuban's statement may depend on how the term "reach" is defined. One might first consider people who read tweets directly on Twitter, which is probably the easiest to measure of everything I'm about to mention. But also consider these types of exposure, which are much harder to tally:

• Prominent tweets quoted in online news stories and on TV and radio news (e.g. Trump)
• Tweets that run in interactive media programs that ask viewers/listeners to tweet at them (e.g. 1A on NPR)
• Tweets embedded in other web media such as Storify and reposted on other social media platforms like Instagram and Facebook
• Tweets that are screencapped and shared on other platforms

So while we have good numbers on how many American adults use Twitter, a liberal definition of the term "reach" pulls in many more people. Exactly many is probably impossible to say with any degree of certainty. /DEEN

On 7/20/2017 2:57 PM, Selby, Gardner (CMG-Austin) wrote:

Thank you for this breadth and speed. If 21 percent of American adults are regular Twitter users, then is Cuban’s 15 percent six points on the low side?

1:59 p.m.

If by "population" he means "American adults," yes. That wouldn't include American minors or non-Americans. "The population," like "reach," is a bit vague... /DEEN