Can statistics be truthful? (America, Article)

[Sunstone]

There have been some recent posts that seem to claim statistics are either almost never truthful, or are never truthful. Do you believe that about statistics? Why or why not?

How well do you feel you understand statistics?

[Bouncing Ball]

Statistics gives a general answer. An avarage. This is a proof between a smaller part of, for example, the population of america. The more people used in your statistics, the closer it will get to the "real" answer. If you bring that number to the real total of the population, only then will you have the exact answer.
In statistics you have to watch out for what kind of people you use. If you want to know the percentage of people playing lego, it would not be fair to interview only 8 year old boys. And if you interview 50 year old people instead, the outcome will be entirelly different.
Therefor the reliabillity of statistics differ a lot.

My opinion however, is that the correct statistics will get you close enough to the truth, depending on how close you would like to get. There are institutes specially for statistics and they will have close to correct investigations. Even though it is possible that statistics are way off, most of the time they will come pretty close..

[9-10ths_Penguin]

Originally Posted by Sunstone

There have been some recent posts that seem to claim statistics are either almost never truthful, or are never truthful. Do you believe that about statistics? Why or why not?

I'd say that, absent any outright fraud or misrepresentation, statistics are almost always factual, at least within the parameters of accuracy and precision of the study. Whether they're truthful depends on how honestly the statistics are applied.

I worry about what's going on behind the scenes in some statistics that get quoted in the news. Professional survey houses will go to great lengths to get a random sample of people to survey and an impartial set of questions... they even do things like change the order of responses to get rid of any effects that might cause. It's not usually clear how good the survey was when you read an article about it, even with the "+/- 3.2% 19 times out of 20" note.

Bad pollsters will do things, either intentionally or unintentionally, to sway the vote in one direction or another. The answer to the question "would you vote for incumbent 'A' or candidate 'B'?" can be very different depending whether it comes before or after the questions "how do you rate the job incumbent 'A' has done on education?" and "of 'A' and 'B', who do you think has the better environmental platform?"

[RevOxley_501]

75.982345% of statistics are made up on the spot

[Mike182]

depends what you are looking at. statistics for how much tax was collected/spent etc will be very truthful, however when using statistics to chart human behavior i think they loose value.

when i studied sociology, the methodology for studying macro society vs studying micro society is very different, you get different results. i never really felt that 'generally speaking' said anything of real value when compared to the qualitative data collected by other means - and statistics charting human behavior can only say "generally speaking".

[Engyo]

The old saw says:

There are lies, there are damned lies, and then there are statistics.

[9-10ths_Penguin]

Originally Posted by Mike182

i never really felt that 'generally speaking' said anything of real value when compared to the qualitative data collected by other means - and statistics charting human behavior can only say "generally speaking".

I agree.

I doubt I know anyone who lives at the mean, median or mode for anything. If you don't capture the variation, you usually don't get the real picture.

[Willamena]

Originally Posted by Sunstone

There have been some recent posts that seem to claim statistics are either almost never truthful, or are never truthful. Do you believe that about statistics? Why or why not?

How well do you feel you understand statistics?

I'm a bit sensitive to misuse of prediction. Statistics are frequently, almost universally, incorrectly applied, even by people who know better, to predict the behavior of individuals within the group rather than the behavior of the group. You see this all the time in astrology. Each person's circumstances are entirely unique. I have no problem with stats if they're used sensibly, but the moment someone suggests that something inherently personal and particular to an individual can be predicted by them, my shackles will rise.

Granted, that wasn't the case the other night. The statistics showed a decline in abortions within a group, and I had no right to argue against that. I was fixated on the individual decisions, as if the statistics had a magical capacity to make them suddenly choose other than they would have chosen.
Do I owe someone an apology? Probably so.

Of course, the more significant moral of the story is never get into a debate about abortion, especially, but not limited to, if one's been drinking.