Polls are a matter of science and are mostly correct.
Done right they involve a scientific understanding of statistics, but they can use faulty methodology.
And if they start with incorrect assumptions the rigid application of proper methodology can cause them to go more askew. You can just randomly grab people for the study and your selection might be off. You can try to adjust for certain "known facts" about the general population and make your sample match your prior information, but if your prior information is incorrect you don't correct for your pool being non-representative, you actively create a less representative pool.
Then you can correctly, accurately and without bias, apply proper statistical principles to get your results, but to an extent even though the process is correct, the action as a whole is garbage in/garbage out.
Then there are potential issues about the wording or ordering of the questions.
Polls are a matter of science and are mostly correct.
I'm not sure they lie any less than anyone else, but I'm not accusing of them of lying, and I don't think the poll results are lies (or at least most of them aren't). My point would apply to a completely honest person, who just has a mistaken assumption. Take a faulty, but honestly believed, assumption, and push in in to a rigorous and proper process, and you are still likely to get an incorrect result.
------
Do you really think that those identifying with Democrats, outnumber those who self identify as Republicans, to a greater extent now then they did when Obama was elected? Some polls start with that assumption, others assume almost as much of a lead for the Democrats. |
