Abusing the Use of Statistics

From newspapers to blogs to forums, statistics has been used widely and became a mind altering element in the media. We however, statistics have many conditions that determines the validity of a claim. Most of the studies that I have seen from newspapers and espicially the Associated Press have been inappropriately done as if they were made in the hands of toddlers. Statistics is a complex subject that requires mathematical skills, verbal and rhetorical skills, and espicially logic.
All of those "20% of people, blah blah blah can't locate the US on the map" or those "20% believes" in geocentrism or the genesis or "2% of the people said they wre abducted by aliens" are baseless. How were the studies done, did they sample 10 people?, 5 people? or did they ask a class of five yearolds. Most of the time, these informations are not given, and apparently many people believe these baseless statistics. When a study has bias, it cannot be applied to the whole population or everyone.
Apparently, the media and some amateurs decide to fool the general population and most of the time they succeed in doing so. Statistics can dramatically alter personal ideas and create mass amounts of bias.
I decided to make this thread and spread the message: Not all experiments or studies are trustworthy, DO NOT BELIEVE IN THEM UNLESS BACKGROUND INFORMATION AND THE METHODOLOGY ARE GIVEN. These are natural rules of statistics that everyone should know about; take a statistics class if you want to know more. I have so much to say, but not everyone will understand the technical information.


"""""Logic"""" people


[edit on 2/1/2008 by die_another_day]

I agree to a certain extent, although without the many statistics i doubt many of the same people who cant find the u.s. on the map would not know the state of education in the u.s.

Explain your vast knowledge of Stats, Ap student.

I am a call center manager in my "real" life. In my career i have been a trainer, supervisor, training manager, operations manager, and workforce manager. I have had lots of opportunity to touch several key pieces of the call center business.

One thing that has always been used as a "sort of" joke is "Figures don't lie, but liars can figure."

Of course statistics are abused. And if you are creative in how you calculate them (claiming statistical ignorance when caught) you can make chicken crap look a lot more like chicken salad.

Do the media misuse statistics? I tend to agree with the sentiments expessed by the OP and I offer a quote from Aaron Levenstein: "Statistics are like bikinis. What they reveal is suggestive, but what they conceal is vital."

Vic

My favorite is when someone tries to use the "percent to goal" facade. For example, if the goal is 85%, and they are at 75%, they will try to say "But i am 88% to goal". They completely ignore the contents of that 10% shortfall that kept them from being 100% to goal (or, God forbid, over delivering at 110% of goal).

80% of my brain thinks 40% of it is not using at least 20% of its full potential. However 100% of my friends thinks I'm 90% insane.
Just kidding of course and I agree that statistics are way over used and way too many folks believe what they hear on CNN, FOX, CNBC & MSNBC these days and that is very alarming.

i remember this since i was about 14 so----
u know how toothpaste commercials always have that part---"4 out of 5 dentists recoomend" colgate---aqua fresh, crest etc etc

well they could ask 10 dentists and 4 may recommend and 6 could say it's poison----yet they can choose to use the 4 yes and 1 no to get a great sounding 4 out 5 recomend it

so essentially they can choose which responses they want to use which makes them sound great and really may be far from the truth

although they're lieing, they're telling the truth at the same time

so---yes i can definitely see where your coming from

It is not necessarily that the statistics lie, but framing can influence the statistics. In the examples of surveys, the way a question is framed can skew results. For example if I asked a large group of people, "Do you believes some higher power controls the universe?" a large number of people may respond yes. If I asked "Do you believe in God?" a smaller, yet substantial number of people will respond yes. If I asked, "Do you believe in Jesus Christ, lord and savior, who will come to Earth any minute and judge all of mankind and cast the heathen sinners who voted for Hillary Clinton into Hell?" a small number of people will respond yes.

Once the statistics are gathered, the way the results are framed can skey how people perceive the results. Let us suppose a surveyer asked a large group of people, "Do you believe in Jesus Christ, lord and savior, who will come to Earth any minude and judge all of mankind and cast the heathen sinners who voted for Hillary Clinton into hell?" and came up with a very small percenage of yes respondants. Someone could frame that result as meaning a very small percentage of people are Christians, or a very small percentage of people believe in God, or a very small number of people believe a higher power runs the universe. Conversely if the surveyor asked people "Do you believe a higher power controls the universe?" and a large per centage of people respond yes, people could frame the results in all sorts of ways ranging from many people are agnostic to many people believe Jesus is coming to Earth to thrown all of Hilary Clinton's supportersa into hell.

Statistics show that 90% of statistics are incorrect
Two lines!!

I am 95% confident that the true proportion of people that had really seen an alien or a ghost out of all people who claims to have seen an alien or a ghosts is between 0.0005% and 0.0010%


[edit on 2/2/2008 by die_another_day]

It's called lying with statistics. There is actually a book that may still be in print on the practice. Called "Lying with Statistics"...many moons ago.

Statistics are the tool of those who want to manipulate data to their advantage.

If a variable correlates to another variable, it does not always mean that one will cause another. Also, you cannot find correlation among categorical or non-quantitative variables, you can only find association which is just what do you see, a curve? a line? but that barely means much.

reply to post by Valhall


That is not true. It is more of a "let the buyer beware" scenario than that.

I use statistical analysis to determine call arrival trending, and performance trending. Without it, i would be left guessing where my performance is heading, and my decisions would amount to frantic stabbing in the dark and hoping that i will blindly stumble into the right direction.

Any knowledge can be misused.