CitizenBBN
12-19-2012, 04:22 PM
This is a gun article, a poll about those who support either major restrictions or outright bans, but my point isn't the gun issue but rather how this story is carefully slanted.
Title of the article:
CNN Poll: Bare majority now support major gun restrictions
First paragraph:
Americans are reacting to the tragic events in Newtown, Connecticut with elevated levels of anger, shock and fear that have led a growing number of people to look for action from the government and society that can prevent future incidents, according to a new national survey.
So far so good. Now the actual poll data, deep in the article:
The number of Americans who favor major restrictions or an outright ban has typically hovered just under the 50% mark in recent years; now that number is just over 50%. According to the poll 52% say they favor major restrictions on guns or making all guns illegal. That's a five point rise from a CNN survey conducted in early August, ... The five point rise is within the poll's sampling error.
Here's the glitch. Statistics 101: if you sample something and it comes in at X then you sample it again and it comes in Y, and X and Y are within each other's sampling error, there is NO CHANGE in your result. You can't say it's up or down b/c you honestly don't know the values close enough to say. More properly you'd have reported that "in August between 45 and 55% support restrictions, and now it's between 47% and 57%." Like I said, within the effective ability of the poll, no change.
Now the story did report other changes that are in fact significant and some that aren't changes that also are telling. This isn't about the story though, it's about how facts are manipulated to create the desired "accepted fact". That's my term for something that may or may not be factually accurate but becomes an accepted truth.
In fact, in contradiction to the title, we don't in fact know if a majority support major gun restrictions. All we know is it is pretty close to August. Statistically it may even be lower. I don't personally doubt it is higher, I'm sure it is, but you can't claim you have a poll or study to support that conclusion as if it's a "fact". We can draw a conclusion but it's not a fact.
Sorry, this stuff bugs me to no end. No this one is not a massive mistelling, but we constantly tolerate the conversion of studies that say limited things into "facts" that aren't supported by the data.
(PS - while some of the numbers are large shifts, a 2-4 point swing in overall favoring of major restrictions is pretty small given the scale of this event. The thunderous change of 15 points is in being likely to support some kind of new law, but even with that spike it's still 43%. )
Title of the article:
CNN Poll: Bare majority now support major gun restrictions
First paragraph:
Americans are reacting to the tragic events in Newtown, Connecticut with elevated levels of anger, shock and fear that have led a growing number of people to look for action from the government and society that can prevent future incidents, according to a new national survey.
So far so good. Now the actual poll data, deep in the article:
The number of Americans who favor major restrictions or an outright ban has typically hovered just under the 50% mark in recent years; now that number is just over 50%. According to the poll 52% say they favor major restrictions on guns or making all guns illegal. That's a five point rise from a CNN survey conducted in early August, ... The five point rise is within the poll's sampling error.
Here's the glitch. Statistics 101: if you sample something and it comes in at X then you sample it again and it comes in Y, and X and Y are within each other's sampling error, there is NO CHANGE in your result. You can't say it's up or down b/c you honestly don't know the values close enough to say. More properly you'd have reported that "in August between 45 and 55% support restrictions, and now it's between 47% and 57%." Like I said, within the effective ability of the poll, no change.
Now the story did report other changes that are in fact significant and some that aren't changes that also are telling. This isn't about the story though, it's about how facts are manipulated to create the desired "accepted fact". That's my term for something that may or may not be factually accurate but becomes an accepted truth.
In fact, in contradiction to the title, we don't in fact know if a majority support major gun restrictions. All we know is it is pretty close to August. Statistically it may even be lower. I don't personally doubt it is higher, I'm sure it is, but you can't claim you have a poll or study to support that conclusion as if it's a "fact". We can draw a conclusion but it's not a fact.
Sorry, this stuff bugs me to no end. No this one is not a massive mistelling, but we constantly tolerate the conversion of studies that say limited things into "facts" that aren't supported by the data.
(PS - while some of the numbers are large shifts, a 2-4 point swing in overall favoring of major restrictions is pretty small given the scale of this event. The thunderous change of 15 points is in being likely to support some kind of new law, but even with that spike it's still 43%. )