Reply to thread

Message

<blockquote data-quote="Rechan" data-source="post: 4900277" data-attributes="member: 54846">Now, I have to go back on something I said. Looking at <a href="http://www.sciencebuddies.org/science-fair-projects/project_ideas/Soc_participants.shtml" target="_blank">this page</a>, a sample size of 100 isn't good enough for a true blue study to say something with confidence. You need 500 participants. Because 500 participants grant a margin of error a little below 5%, meaning you are 95% certain the numbers are correct.If you look at the page, you'll note that 10,000 participants only get you 1% margin of error. The difference between 1% and 5% margin of error really isn't that much. 30K participants is, well, overkill.WITH THAT SAID, do you have any idea how much that would cost to conduct using 30k participants? Not to mention teh TIME to enter all that data? Not to mention getting that many people to participate in the first place. Let alone the amount of cash you're forking out for the printing of 4 or so pages, at minimum, per participant... The only people who could conduct studies then would be corporate polling companies and governments. As a rough guestimation, I'd say your typical study is conducted by 1-3 professors who have at most 3 grad students helping them. That's it. To give you an idea, you might have 500 people sign UP for a study, but you might wind up throwing out 30-40 purely because they didn't fill out the questionaires entirely. You can't use a half-finished questionaire.Bingo. Most studies I've looked at in school (or did myself, or helped do), usually the author looks at ONE SPECIFIC LITTLE TINY THING. And often he'll get some interesting results he didn't expect and say, &quot;Yeah, we need more research to look at this funny thing I noticed to get more details about it&quot;. Or often, &quot;There's very little research looking at this specific group, or addressing this variable (like gender, race, age, etc). We need more.&quot;</blockquote>

[QUOTE="Rechan, post: 4900277, member: 54846"] Now, I have to go back on something I said. Looking at [URL="http://www.sciencebuddies.org/science-fair-projects/project_ideas/Soc_participants.shtml"]this page[/URL], a sample size of 100 isn't good enough for a true blue study to say something with confidence. You need 500 participants. Because 500 participants grant a margin of error a little below 5%, meaning you are 95% certain the numbers are correct. If you look at the page, you'll note that 10,000 participants only get you 1% margin of error. The difference between 1% and 5% margin of error really isn't that much. 30K participants is, well, overkill. WITH THAT SAID, do you have any idea how much that would cost to conduct using 30k participants? Not to mention teh TIME to enter all that data? Not to mention getting that many people to participate in the first place. Let alone the amount of cash you're forking out for the printing of 4 or so pages, at minimum, per participant... The only people who could conduct studies then would be corporate polling companies and governments. As a rough guestimation, I'd say your typical study is conducted by 1-3 professors who have at most 3 grad students helping them. That's it. To give you an idea, you might have 500 people sign UP for a study, but you might wind up throwing out 30-40 purely because they didn't fill out the questionaires entirely. You can't use a half-finished questionaire. Bingo. Most studies I've looked at in school (or did myself, or helped do), usually the author looks at ONE SPECIFIC LITTLE TINY THING. And often he'll get some interesting results he didn't expect and say, "Yeah, we need more research to look at this funny thing I noticed to get more details about it". Or often, "There's very little research looking at this specific group, or addressing this variable (like gender, race, age, etc). We need more." [/QUOTE]

Verification