January 30, 2020

The Trib's VERY Subtle Reporting Of The Poll Data

Let's take a break from the overt, slow-motion Senatorial coverup being perpetrated in DC and take a look, instead, at this piece by Deb Erdley over at the Tribune-Review.

Keep an eye out for the subtle skewing of the data.

The frame of the entire piece is found in the headline:

Trump support edges up in Pennsylvania poll as impeachment continues 

And the first paragraph:
President Trump has grown stronger with Pennsylvania voters despite ongoing impeachment proceedings, a poll finds.
The next few paragraphs answer the question of which poll:
A majority of the 628 registered Pennsylvania voters who responded to a Jan. 20-26 Franklin & Marshall Poll said it is time for a change in the White House. Still, 38% of them said President Trump is doing an “excellent” or “good” job as president. That’s a 3% increase from his 35% favorability rating three months earlier.

And 41% said he is doing well enough to be reelected, an increase of 4% from October.
So far all about Trump and how his support "edges up" and "grown stronger" despite impeachment, right?

Let's go take a look at the poll to see the numbers in context. What's the context of that first percentage (38%)?

The data is found in the "Job Performance" section of the poll (pg 7):
About two in five (38%) registered voters in Pennsylvania believes President Trump is doing an “excellent” or “good” job as president, which is slightly higher than his October Franklin & Marshall College Poll rating.
So far, so good. But what about the percentage who thought otherwise? Isn't that number just as important?

For that info we need to go to the "RatePres" table (pg 25) which breaks the 38% down to Excellent - 21% and Good - 17%. (21 + 17 = 38). On the other hand, that very same table tells us that 49% (which would be a full 11% more than 38%) of Pennsylvanians think that Trump is doing a poor job.

Where is that in Erdley's reporting?

If you add the "fair job" percentage (12%) to that you get a whopping 61% of polled Pennsylvanians who think Donald Trump is doing a fair to poor job.*

I realize the piece is about the change in approval ratings since the last poll. The last poll had Trump at 35% Good + Excellent rating against a 54% Poor rating. So while there is a shift up of 3% in the positive numbers and a shift down of 5% in the negative, it has to be seen in reference to the fact that half of Pennsylvania voters think that  Trump is doing a poor job.


Inching up to 38% from 35% doesn't mean much in comparison, does it?

Go back and read the headline and the first paragraph. It's all true but it hits you somewhat differently now, doesn't it?

Subtle skewing of the data.

Then there's this from the third paragraph:
And 41% said he is doing well enough to be reelected, an increase of 4% from October.
This is from the poll's REPRES table (pg 25). The very next line tells us the percentage of Pennsylvanians who think that "it's time for a change." And that's 57%.

Where is that in Erdley's reporting?

For that, you'll have to go way down in the ninth paragraph, nestled in right after a quotation from a fervid Trump supporter.

In the Franklin and Marshall report the info is in the same sentence:
Two in five (41%) registered voters believe President Trump has done a good enough job to deserve re-election, while three in five (57%) voters say it is time for a change.
In the Trib reporting of that report, those numbers are separated by 6 paragraphs.

In order to emphasize the notion that Trump's numbers are getting stronger in Pennsylvania, they de-emphasize the context of how unpopular he is in Pennsylvania.

Subtle skewing of the data.

See how it works?

* - I'm separating these out because I am not sure if Franklin and Marshal consider "Fair" to be a good thing or a neutral thing or a bad thing. I am thinking that as there are two "positive" ratings (Good and Excellent) there should be two "negative" ratings - but Fair as an adjective isn't really analogous to a "negative good", so I'm stumped.

No comments: