Posted by: Anonymous Coward
on April 19, 2005 11:28 PM
"Why won't Info-Tech reveal the questions?
You should take his word for it! It is true.
The questions are pretty apparent, though, from the results themselves. "
I have wondered for a while if it might be fun to design some software that would let us do statistically valid surveys with doctored questions. Then we could reveal the results while hiding the questions (and setup) until we had a storm blowing.
My thinking is that by obviously mis-using survey methods and getting outrageous results, we may help educate people about the possibilities.
For instance:
---------------
"You are a sex predator who kidnaps and abuses children under ten years old. If you are caught and convicted, do you think:
1. You should be put to death for your crimes? 2. You should be put in jail for life with no chance of parole? 3. You should be put in jail for life? 4. You are sick and should be given therapy and then set free? 5. You are innocent.
Then report the results something like this:
500 people were surveyed concerning sex crimes agains children.
1. 32% said they should be put to death for their sex crimes. 2. 50% said they should get life without parole for their crimes. 3. 10% said they should get life in jail. 4. 6% said they should get therapy and be released. 5. 2% said they were innocent of all sex crimes against children.
Conclusion: Of 500 random people polled, only 2% claimed to be innocent of sex crimes against children under 10 years old. We have do something about this.
---------------
Granted, that example is way over the top, but I think it illustrates what we can do. Will such a tactic help to educate people as to the ways figures and polls can be bent?
Learn about Linux
Download Linux
Get Linux help
Re:Questions apparent
Posted by: Anonymous Coward on April 19, 2005 11:28 PMYou should take his word for it! It is true.
The questions are pretty apparent, though, from the results themselves. "
I have wondered for a while if it might be fun to design some software that would let us do statistically valid surveys with doctored questions. Then we could reveal the results while hiding the questions (and setup) until we had a storm blowing.
My thinking is that by obviously mis-using survey methods and getting outrageous results, we may help educate people about the possibilities.
For instance:
---------------
"You are a sex predator who kidnaps and abuses children under ten years old. If you are caught and convicted, do you think:
1. You should be put to death for your crimes?
2. You should be put in jail for life with no chance of parole?
3. You should be put in jail for life?
4. You are sick and should be given therapy and then set free?
5. You are innocent.
Then report the results something like this:
500 people were surveyed concerning sex crimes agains children.
1. 32% said they should be put to death for their sex crimes.
2. 50% said they should get life without parole for their crimes.
3. 10% said they should get life in jail.
4. 6% said they should get therapy and be released.
5. 2% said they were innocent of all sex crimes against children.
Conclusion: Of 500 random people polled, only 2% claimed to be innocent of sex crimes against children under 10 years old. We have do something about this.
---------------
Granted, that example is way over the top, but I think it illustrates what we can do. Will such a tactic help to educate people as to the ways figures and polls can be bent?
#