How much do you trust statistics/figures?

A lot of numbers get thrown around in here, and the old addage goes "Statistics don't lie, but liars use statistics." I had a whole class in college on the subject of statistical interpretation and understanding what the statistics are REALLY saying when people use them. Two clever people can look at the same data, and can make it say whatever they want it to.





How much stock to you put into new studies when they are released? Do you even bother trying to find out who sponsored/paid for the study?

How much do you trust statistics/figures?
It depends on the study - official is always better (government, JRF, UN, Euro-wide research) just so you know it's been done properly, ethically, etc and it's just not based on the life histories of about 8 people and nothing else.





Then you always have to look to see how it is counted - for example 'women with degrees earn more' - than... women without degrees? ...than men with degrees? ...than other immigrants with degrees? How is it measured / counted? Are any groups left out or collapsed into other groups?





Then - what is the research trying to prove? Does it have an agenda? Research that showed Christians were happier and earned more money than non Christians, could be very biased if it came from a Christian organisation.





Is the research backed up by other studies? Are the findings similar or are they the exception? How have the findings changed over time? For example, a common thing with fertility statistics is to show the decline since the 50s - but we all know the 50s were an aberration - so what is the long term picture, is the fact increasing, decreasing, changing anywhere else or in other societies, etc..





So yeah, I do look. And it's true that two clever people can make the same research say two different things, which is why it's best to go and check out the figures for yourself and make your own mind up :-)
Reply:We have all pretty steadily discounted government backed stats as being any more credible, so your post was a typical skeptic's with no real information to back it up. A poor answer. Report Abuse

Reply:Statistics are not a very credible source.





The reason why is because statistics can, and only do state so much of what is on record. For an example, criminal statistics only present information on crimes reported and where the suspect is apprehended. There are many unreported crimes that happen in America everyday, and almost as many crimes where there is not any sufficient evidence that is a credible enough to punish the suspect.





There are also many other factors included which make statistics inaccurate, Media attention on certain geographical areas. Some places received more unwanted attention than others do to unpopularity, popularity of the place, the racial percentage of certain ethnic groups, and the local governments ability to sweep certain issues under the rug to prevent bad publicity.





Statistics are also written by an author with a bias opinion on the subject that they are provided statistics on.





With that said, i ignore statistics, just as i ignore movie critics.
Reply:Statistically, the highest levels of exposure to second-hand smoke have not been associated with any substantial health risks. Statistically, cigarette smokers are an economic asset to government coffers. Statistically, the social security system is financially robust. Statistically, war is not good for a country's economy. Statistically, financial outlays for social welfare are minuscule compared to corporate welfare. Statistically, babies born addicted to cocaine, methamphetamine, or heroin have no more physical or psychological problems by the age of 5 than babies born without drugs in their system -- provided that the drug-born babies are adopted in infancy. Statistically, only a minority of physically abused children become abusive adults. Statistically, vaccinations are not linked with autism. Statistically, the trickle-down theory of economics hasn't worked since the Egyptian pharoahs conveniently deposited massive amounts of riches in one-stop-shopping caches for the grave robbers.





I could go on and on. These statements represent consistent statistical results from multiple studies. Do they represent popular convictions?





At one point, I cautioned colleagues that using questionable conclusions from other studies to justify legitimate research proposals would diminish the overall credibility of their own findings. I admit I was wrong. It doesn't matter what the actual numbers show. What matters is what people believe they should show.





And as far as determining the sponsors of studies, most research is funded by the government, and most researchers toe the company line.
Reply:stats can easily be pushed around and twisted to support what the writer is arguing. I don't trust stats unless the person making them has a strong background in math or statistics. when A person with a degree in psychology uses stats to support his/her point i usually assume its wrong and ignore it. I've seen the stats classes they are required to take for their degree and I assure you they are not qualified to be given any statical figures on their own. I don't mean to offend people I just get dissappointed when people try to sell uninformed audiences products by impressing them with so called statistical figures they imagine. Its dishonest and outright cheating people.








look up prosecutors fallacy on wikipedia, it gives a great example of how conditional probability is abused. If you can follow the math I think you will find it really interesting.





stats don't lie, its the people making them
Reply:When you trade the currency markets the way I do, you're forced to read government sponsored economic statistics.





And I can tell you that from the day the Bush Admin entered office, economic and inflation stats have gone from being questionable to dubious and finally, at this point, they're just flat out falsified.





I take no stock in any numbers coming either from the White house, nor anything widely distributed in the corporate media.





These people are lying through their teeth. In fact, they've been lying for so long and without any penalty, I don't think they know how to tell the truth anymore. They sure as hell haven't got any reason to.





Ask yourself, why are these people telling me xyz? Do they really like me and want me to know the truth?


Highly doubtful.


You're just a commodity to them that they trade to advertisers.
Reply:There is a great saying, I forget where I heard it and it goes like this: "In God we trust, all others bring data!"





Sounds like you already know the answer to your question. It's all about the data. Where did it come from? How was the data pulled, etc? If you answer those types of questions for each statistical result set you review, you will know whether or not to trust it. Do the sources for which the data was collect and the methods used for data collection meet my level of expectations? And the real kicker ... do I know enough about the subject matter to really evaluate or to have comfort in the data collection methods for which the analysis is based off of?
Reply:Personally I don't trust statistics. Statistics are good to get an over all idea about the chances of something. Now to actually go by it you must be well eduacted in STAT. Im taking a college course on it and it gets pretty confusing, in general I use them to get an idea of something. I tend to want to beat the odds to prove ceartin stat theorys wrong
Reply:My mom worked for the CDC (Center for Disease Control) who is a commonly quoted source of many statistics. The stats are most definitely political, with much data suppressed when data is unfavorable by the powers that be.





I most definitely do not believe in polls, as most self-respecting people don't even participate in polls.
Reply:I dislike stats and figures just as I dislike small range studies. Studies are usually made with a bias behind it. If the result being seeked is not proved, then it always requires "More research". I don't like stats because..seriously, I refuse to believe examples with headlines like "Out of a study of 25 women, 23 women say that they have worthless husbands. This proves that men are blah blah blah" Even though it is fake, I guarentee that people who don't question stats will believe it and make stereotypes.
Reply:In cases of surveys and politics, not very much. We all know how easily those can be twisted to suit those in power, and how difficult it is to get the real numbers.





When stats are published about an issue that is political, like domestic violence, NEVER take it at face value. Take it with TWO pinches of salt. The link in my profile should give you a very good idea of what I mean.
Reply:I generally try to read those boring methodology sections. Those tell me a lot.





Since I research and publish I know how the study is conducted as well as the methods involved are as important as the researchers 'findings'. Thats why I ask for cites here (and get lambasted for it). Does it take a phd to read some of those- yes. And they even put me to sleep.
Reply:I try to find a few studies and usually somewhere in the middle makes sense.





It is almost imossible to find any study without any agenda. there is always money going on somewhere behind the scenes.





And a lot of times common sense tells when you when something is out of whack.
Reply:I don't trust half of the statistics around and I doubt the other half. You're absolutely right - statistics beg for manipulation beginning with those who sponsor the work - and studies are always sponsored.
Reply:Statistics lie. You can use statistic so that almost anything can come out with the results you want - it's just a matter of how you apply it.
Reply:I take each study with a grain of salt, especially when it comes from a political or social organization with an axe to grind.
Reply:Statistics don't always reveal anything solid.
Reply:I toss and coin and see if I believe it or not. After manipulating variables with SPSS....I seem evil trying to manipulate numbers trying to go along with what I say lol.
Reply:90% -- 50% of the time !!
Reply:I trust them about 20% of the time...do you believe me?