Thursday, January 11, 2007

Statistician help wanted.

I need a statistician’s help.

But before any of you consider helping me, you deserve to know:

1) Trolls and some reporters and editors at the Raleigh News & Observer say I’m a pretty bad guy. I don’t believe them, but I could be fooling myself;

2) There’ll be no fee for your services

3) Be prepared. I’ll check anything you say with others. I’ll even question your credentials.

Now, are you still interested in “the job?”

You are?

OK, here’s what’s involved: As many of you know, back on Mar. 24 the N&O published on its front page the story it claims “broke the Duke lacrosse story.” (“DNA tests ordered for Duke athletes”)

In it, the N&O referred seven times to the False Accuser as “the victim” or with the possessive “victim’s.”

The N&O never once used standard journalism qualifiers such as “alleged” or “reported” which alert readers to the fact an accuser’s claim(s) is disputed and unproven.

By leaving out qualifiers in the story in which the public and media first learned of “the Duke lacrosse case,” the N&O cast the False Accuser as the victim and framed the Duke students as her victimizers.

Most of the blame for that travesty of honest and accurate journalism has fallen on Samiha Khanna and Anne Blythe, the two reporters bylined on the story.

However, a number of N&O editors worked on the story. Regardless of their particular editorial assignments, all of them were, as the N&O’s exec editor for news Melanie Sill often tells readers, responsible for identifying and correcting errors.

Now here’s where I need statistical help.

Journalists familiar with how a paper like the N&O operates have given me their best estimates of the number of editors who likely worked on the Mar. 24 story. The lowest number given was six; the highest was ten.

I'll use six as a "working number."

If six editors each fact-check seven items, each of which has an error, and none of the editors note any of the errors, that makes a total of forty-two errors that weren’t noted.

My question for statisticians: What is the probability that all six newspaper editors would miss by chance all seven instances in the N&O story where the N&O failed to use a qualifier such as “alleged” or “reported?”

Folks, I believe the N&O made a deliberate decision to not use qualifiers in its Mar. 24 story; and instead tell the public and media the False Accuser was a victim of sex crimes.

Whether reporters Khanna and Blythe were part of that decision we don’t know for certain. Editors often change reporters’ copy.

I doubt editors would “pencil out” qualifiers but with the N&O we can’t always be sure. Who would have thought the N&O would have withheld from its Mar. 25 story all the news it had of the players cooperation with police and instead promulgate the lie that the players were stonewalling and refusing to cooperate with police?

What we do know for sure is that editors, even N&O editors, are supposed to identify and correct errors. I believe forty-two or possibly more errors of the type we’re talking about here didn’t happen by chance or because of “deadline pressure.”

Exposing the N&O’s deliberate errors in its Duke lacrosse coverage is the essential first step to getting the N&O to eventually correct and apologize for what it did in its Mar. 24 story, “DNA tests ordered for Duke athletes.”

13 comments:

Anonymous said...

John,

I suggest you pose this question to the N&O publisher, Orage Quarles. All the staff ultimately report to him.

Anonymous said...

The statistical question isn't straightforward because I think you have to make an assumption on how often an editor misses an error. For argument sake, let's say an editor misses 1 in 10 errors (the number seems high but it makes the math easy).

This is the same error repeated, so let's say each editor has a 1 in 10 chance of missing it throughout the article. Each editor's review is an independent event so the chance would be 1 in 10 raised to the 6th power that all editors miss this error. That's the proverbial one in a million chance.

Anonymous said...

It might also be helpful to look at past N&O reporting on rape cases, and see how often they used the term victim vs alleged victim.

Anonymous said...

John,
I agree with the second anonymous(5:37):unless one knows the odds of a single fact checker missing an error,there's no way to calculate (1/x)to the nth exponent.And I'm not a statistician,but I do have a science backgound--MD-.Neither John von Neumann nor Isaac Newton could calculate this.
Best,Corwin
But from beginning calc;as the limit of a function approaches zero,it can be taken as zero.So no chance it was an error.

Anonymous said...

Quarles has been very quiet and has done nothing about apologizing or correcting the terrible and inflammatory March 24 and 25 reporting. Better to send letters to the McClatchy board of directors.

Anonymous said...

John,
You were the first blogger I found on this case. You've been relentless and inspiring! I know you will keep the pressure on!

This disaster was ignited by the N&O. How do they sleep at night?

Anonymous said...

I would say there is a problem with the question: there were not seven errors, there was at most one error committed seven times. You are not referring to independent events like misspellings, you are referring to dependent events. If one had been changed, all would have been changed.

Anonymous said...

Early on, I emailed the McClatchy people a few times and was simply told to send my complaints to the local editor. Much good that did!
Texas Mom

Anonymous said...

As noted above, the focus should not be on the odds that all these editors missed these oversights in one article. Rather, a sample of previous coverage in the N&O of criminal complaints should be gathered. If 20 previous articles are reviewed, and each one is quite careful to use the words "allege", etc. then there is a serious question as to whether the omission in this case could be construed as accidental. On the other hand, if in 20 previous articles the word "alleged" were left out 4-6 times, then you could chalk it up to sloppy editing.

Anonymous said...

John,

It would be 1/7 to the sixth power. It is 1/117,649 or 0.00000000849986. The last number is the probability that all of them would miss the error, or a 1 in 8.5 billion chances.

Hope this helps.

Anonymous said...

john: as a person who works with numbers--doing analysis, daily, i think you must step back and re-examine the bigger picture: 1 biased paper+ 1 group of pin headed editors + 1 righteous cause = 0 errors.
dang, that was EASY....
now, back to crunching numbers...
Lee

Anonymous said...

But John, in this society, you are a bad guy. You know that emotional comfort achieved through feigned outrage is far superior to facts and truth. Yet you insist on shining the light of truth on all those comfortable liars hiding in the dark recesses of politically correct high dudgeon and pretend, but believed, moral superiority.

Don't you know that "not true, but accurate" is superior to truth? Where were you when they held that class for the cognitive deficient?

Anonymous said...

Error!!

It is 0.000008499859. My mistake.

However, we still are dealing with some pretty interesting probabilities. The figure above is the probability that they WOULD NOT catch the referenced mistake.

A number like that tells me that the omission was deliberate, or at least -- ah, no, -- deliberate.