Freeper XM does media bias survey of New York Times news articles...

LEXIS-NEXIS, my own pointy head | Today | xm177e2

[xm177e2]

New York Times National Desk Media Bias Survey, 1991-2000:

Note this search was done on LEXIS-NEXIS. ONLY news articles from the New York Times National Desk were searched—letters to the editor do not appear, nor do editorials or columns (unless they contain the term “national desk,” I have no idea how often this would occur, I assume it’s very rare). This is because LEXIS-NEXIS has no way of searching just within news articles, but every article from the National Desk contains the term “National Desk” in the LEXIS-NEXIS summary.

“Occurrences” is the number of articles in which the term appears. “Both” is the number of articles in which both terms appear. “Mut. Excl.” is the number of articles in which ONLY one of the two terms appears. Relative frequency is the number to watch, if the “Relative Frequency” of “right-wing” to “left-wing” is 5, that means the term “right-wing” is FIVE times as likely to appear in a news story! A relative frequency of "1" would occur if the words are used an equal number of times. A value over 1 might suggest liberal bias, while values under 1 suggest conservative bias.

#Div/0 denotes an Excel error because one cannot divide by zero, so certain relative frequencies are incalculable.

“w/5” means “within five words of,” so “Right-Wing! w/5 Radical!” would pick up phrases like “right-wing radical terrorists from the…” or “but certain radicals on the right-wing demand an end to…”

When a word ends in !, that means the LEXIS-NEXIS search will look for all words beginning with whatever is before the exclamation point. “Extrem!” thus could be “extreme,” “extremely,” “extremist,” “extremities,” etc.

These are the results, they have not been peer reviewed, it’s just a little research done by a college student on a lark, but I assume it’s all correct.

Terms/Phrases:	Occurrences	Both	Mut. Excl.
Number of uses of the word referring to
x
Right-Wing!	828	32	796
Left-Wing!	164		132
Relative Frequency:	5.04878		6.030303
x
Right-Wing	806	31	775
Left-Wing	156		125
Relative Frequency:	5.166667		6.2
x
Hard Right-Wing!	1	0	1
Hard Left-Wing!	0		0
Relative Frequency:	#DIV/0!		#DIV/0!
x
Right-Wing Extrem!	48	0	48
Left-Wing Extrem!	0		0
Relative Frequency:	#DIV/0!		#DIV/0!
x
Extreme Right!	62	1	61
Extreme Left!	6		5
Relative Frequency:	10.33333		12.2
x
Right-Wing Radical!	7	0	7
Left-Wing Radical!	4		4
Relative Frequency:	1.75		1.75
x
Radical Right!	48	1	47
Radical Left!	17		16
Relative Frequency:	2.823529		2.9375
x
Right-Wing Reactionar!	1	0	1
Left-Wing Reactionar!	0		0
Relative Frequency:	#DIV/0!		#DIV/0!
x
Reactionary Right!	0	0	0
Reactionary Left!	0		0
Relative Frequency:	#DIV/0!		#DIV/0!
x
Right-Winger!	29	0	29
Left-Winger!	8		8
Relative Frequency:	3.625		3.625
x
Right-Wing Conservative!	11	0	11
Left-Wing Liberal!	4		4
Relative Frequency:	2.75		2.75
x
Right-Wing! w/5 Extrem!	89	0	89
Left-Wing! w/5 Extrem!	0		0
Relative Frequency:	#DIV/0!		#DIV/0!
x
Right-Wing! w/5 Radical!	21	0	21
Left-Wing! w/5 Radical!	6		6
Relative Frequency:	3.5		3.5
x
Right-Wing! w/5 Reaction	2	0	2
Left-Wing! w/5 Reactiona	0		0
Relative Frequency:	#DIV/0!		#DIV/0!
x
Right-Wing! w/5 Mean-Spi	1	0	1
Left-Wing! w/5 Mean-Spir	0		0
Relative Frequency:	#DIV/0!		#DIV/0!
x
Right-Wing! w/5 Angry	3	0	3
Left-Wing! w/5 Angry	1		1
Relative Frequency:	3		3
x
Right-Wing! w/5 Racis!	2	0	2
Left-Wing! w/5 Racis!	0		0
Relative Frequency:	#DIV/0!		#DIV/0!
x
Right-Wing! w/5 Violen!	9	0	9
Left-Wing! w/5 Violen!	0		0
Relative Frequency:	#DIV/0!		#DIV/0!
x
Conserva! w/5 Extrem!	71	4	67
Liberal! w/5 Extrem!	29		25
Relative Frequency:	2.448276		2.68
x
Conserva! w/5 Radical!	25	2	23
Liberal! w/5 Radical!	26		24
Relative Frequency:	0.961538		0.958333
x
Conserva! w/5 Reactionar	6	0	6
Liberal! w/5 Reactionar!	1		1
Relative Frequency:	6		6
x
Conserva! w/5 Mean-Spir!	2	0	2
Liberal! w/5 Mean-Spir!	2		2
Relative Frequency:	1		1
x
Conserva! w/5 Angry	34	2	32
Liberal! w/5 Angry	9		7
Relative Frequency:	3.777778		4.571429
x
Conserva! w/5 Racis!	6	0	6
Liberal! w/5 Racis!	8	8
Relative Frequency:	0.75		0.75
x
Liberal-Leaning!	44	1	43
Conservative-Leaning!	28		27
Relative Frequency:	1.571429		1.592593

[xm177e2]

Part One of Two...

Obviously, it's impossible to mathematically "prove" something as ethereal as "media bias," but I'm giving it my best shot. I forget if I posted this before, I honestly don't remember.

[xm177e2]

Ping

[fhayek]

Great job XM!

[xm177e2]

Also see Patrick Ruffini's article here, from 3/1/02. His views on the media are basically the same as mine... (his methodology is also similar, although he doesn't go as far as I did (but he searched more than just the NYT), I'm not accusing him of copying off me or anything, because I don't think I've ever posted this before)

[Enlightiator]

Excellent work, very good proof of liberal bias at the NYT. This ought to be read by everyone at FR and cited by conservative columnists!

[Pokey78]

Bravo!

[xm177e2]

The second part I've promised will take a bit longer, I'll probably have it up in two to three weeks, it looks at how they label (and fail to label!) interest groups like Hertige and Emily's List, etc.

I did a lot of research on it before, but I forgot the methods I used, so the data is basically worthless.

[WOSG]

great research!!!!!!

Send this to Brent Bozell!! Publishable material!!!

[Drango]

Yesterday or the day before on NPR, Terry Gross's show, some dude from Stanford did a similar study with 20-30 newspapers from around the US.

His conclusion was, contrary to Goldberg's view, liberals were 2-3 times more likely to have "labels" associated with thier names! I'll see if I can find it on-line and post the link. You and Terry and the Stanford Geek can duke it out :-)

[xm177e2]

Send this to Brent Bozell!!

Done...

[xm177e2]

Yes, but he was only looking at the words "liberal" and "conservative," he was ignoring "right-" and "left-wing," which are far more pernicious labels.

[WOSG]

btw, it is great to do this stuff on NYT, they are masters at using adjectives for everything. it defines their world-view .

try " Republican ... mean-spirited " versus "Democrat ... mean-spirited" and also other 'harsh' adjectives close to Repub or Dem ("extremist" or "harsh" or "hard-line" or "controlled" or "regime" or "forced" or "threatened").

on the "threatened" for example, i have notice the NYT/LibPress would say "Clinton promises veto" but would say "Bush threatens veto". So you may get: "Extremist Republicans, taking a hard line, threatened to overturn LovelyDemPres promised veto."

[xm177e2]

And he also looks at organizations, I know that conservative organizations are smeared with a label (at least in the NYT national desk news pages) far more often, because that's the other research I did.

In fact, it was because of his research that I decided to post this (I saw something about it over at the "morethanzero" Blog...

[xm177e2]

I should acknowledge I'm not the first person to think this up, I read in Harry Stein's book "How I Accidentally Joined the Vast Right-Wing Conspiracy (and Found Inner Peace)" about a few searches he did with LEXIS-NEXIS, so I decided to try my hand at it...

[WOSG]

Yesterday or the day before on NPR, Terry Gross's show, some dude from Stanford did a similar study with 20-30 newspapers from around the US.

ROFLMAO! NPR looking at Liberal Media bias ... and they are completely lost trying to figure out what it is! They have no idea! :-)

[Graewoulf]

Thank you sir! It is high-time to note that "the media" does not use left-wing and extreme to describe itself, but we all know that is exactly what they are!

[Drango]

OK, but here's the NPR link...I'll let the EXPERTS argue this one...

Linguist Geoff Nunberg
Linguist Geoff Nunberg does some of his own checking on liberal bias in the media. Expanded Coverage

[xm177e2]

Okay, I'll do a search now, from 1992 to 2001.

Here are the number of times these things appear (in the NYT national desk news pages):

Democrats Promise: 2
Republicans Promise: 39

Democrats threaten: 3
Republicans threaten: 6

Every time I look for bias, I don't always find conclusive results...

[xm177e2]

Okay, I'll do a search now, from 1992 to 2001.

Here are the number of times these things appear (in the NYT national desk news pages):

Democrats w/10 mean-spi!: 22
Republicans w/10 mean-spi!: 40

This is also not proof of media bias, because these might just be quotes from the article--the NYT isn't guilty of bias just because it quotes a shrill harpy left-wing activist and a reasonable-sounding conservative activist. Also, this just means the words are in proximity, it doesn't necessarily mean they describe each other (for instance, "Democrats say mean-spirited conservatives are..." would show up under "democrats w/10 mean-spi!" It's important to remember the limitations of this method.

[xm177e2]

Another problem with his research is he expects Cato to be labeled "Conservative," when it's not! It's libertarian! If newspapers are labelling it "conservative," they're making a mistake!