Results 1 to 10 of 10

Thread: Actual statistical studies on ADISC

  1. #1

    Default Actual statistical studies on ADISC

    I think that its safe to say that most or all of us have voted in the poll threads that pop up now and again. I was just thinking that those are common response polls, meaning that you choose to be in the survey, not the other way around. From a statistics point of view, we cant trust these polls. Sure, they may give us some indication of the question at hand, but for some of these questions I'd like to know, with a reasonable degree of certainty, what the distributions are.

    For example, What percent of us like to wet and or mess our diapers. What percent of us consider AB/DLism sexual or emotional.

    If this is something people are interested in here's how I think we should do it:

    1. A margin of error should be established.

    2. from the margin of error (MOE) we determine the sample size

    3. participants in a study are selected by assigning every member a number (each number has to have the same number of digits so if there are 20,000 members then #1 will read 00001 and #20000 will read 20000)

    4. we use a random number generator to create what we call a simple random sample, a sample in which each sample of size 'n' has and equal probability of being selected.

    5. once the subjects of the study are established, the mods message them with a survey, which of course anyone may refuse to answer.

    After the results are in, we can really see what the distributions look like.

  2. #2

    Default

    Well, I think this could be really cool! We have a large enough user base that we could create some statistically valid polls, and possibly be a real source of information on the AB/DL community. (Being a bit of a research nerd, I'd love to see a project like this get off the ground).

    Of course, you'd have to put a lot of privacy protections in place if every member on ADISC could be polled. Most of us are very concerned with privacy. And you'd need to have a focused goal for any research done. But I think something like this would be very valuable (and who knows, we may actually see some legit research done on AB/DL).

    I'd suggest coming up with a plan for what you think this should look like, then make a thread in the Administrative Stuff forum to propose your idea. This could be very interesting and a real source of understanding for our community! Great idea!

  3. #3

    Default

    If this ever comes to fruition, I'll participate.

  4. #4

  5. #5

    Default



    Quote Originally Posted by Adventurer View Post
    Well, I think this could be really cool! We have a large enough user base that we could create some statistically valid polls, and possibly be a real source of information on the AB/DL community. (Being a bit of a research nerd, I'd love to see a project like this get off the ground).

    Of course, you'd have to put a lot of privacy protections in place if every member on ADISC could be polled. Most of us are very concerned with privacy. And you'd need to have a focused goal for any research done. But I think something like this would be very valuable (and who knows, we may actually see some legit research done on AB/DL).

    I'd suggest coming up with a plan for what you think this should look like, then make a thread in the Administrative Stuff forum to propose your idea. This could be very interesting and a real source of understanding for our community! Great idea!
    Its a breach of ethics to have names associated with a study therefore they should all be anonymous. Im sure some of the people who are good with html programming would be able to set it up so that NOBODY would know who answer which question, if necessary. Although at a bare minimum, I dont think that the usernames of anyone who participates should be released.

  6. #6

    Default

    There is a lot more that goes into this type of work. The question have to be written correctly and the subject bias has to be taken into account. then the polling style has to be set up correctly.

    This is why a lot of "general public" polls have to be taken with a grain of salt.

    I would rather use a standardized testing system so the statistical package is straight forward chi square or student T analysis.

  7. #7

    Default

    From a site administration POV that sounds like a nightmare...

    I appreciate the idea sounds appealing but, well, it also sounds like a ton of work.

  8. #8

    Default



    Quote Originally Posted by egor View Post
    There is a lot more that goes into this type of work. The question have to be written correctly and the subject bias has to be taken into account. then the polling style has to be set up correctly.

    This is why a lot of "general public" polls have to be taken with a grain of salt.

    I would rather use a standardized testing system so the statistical package is straight forward chi square or student T analysis.
    That's pretty much what what I was talking about. However chi square and T procedures are only necessary if the variables are numerical. Most of the questions I would like to ask would be categorical variables, although I guess if you are going to block by age you could perform T tests, chi square or ANOVA.

    The most important thing about these polls is that they would not be based upon common response. However, you do raise an excellent point. The questions have to be presented in a non biased way, though I dont think there is any way to remove the participants bias, except maybe to increase the sample size.

  9. #9

    Default

    If you looked at the number of people who viewed the poll, and compared it to the number of people who voted (time frame?)...would you get any better data for the results?
    X#-viewed, Y#-responded, Z% of the response chose "B" from a range of "A" through "D"...

    Can someone please help me here??
    Z% of Y#= typical result
    so...Y# is what % of X#, and how does that modify the Z%?

    Would that give us better data??

  10. #10

    Default



    Quote Originally Posted by Marka View Post
    If you looked at the number of people who viewed the poll, and compared it to the number of people who voted (time frame?)...would you get any better data for the results?
    X#-viewed, Y#-responded, Z% of the response chose "B" from a range of "A" through "D"...

    Can someone please help me here??
    Z% of Y#= typical result
    so...Y# is what % of X#, and how does that modify the Z%?

    Would that give us better data??
    Not really. The problem, as I see it, is that people are choosing to participate rather than being chosen. I know it might not sound like a big deal but this kind of thig ALWAYS, without fail screws up any reliable data.



    Quote Originally Posted by Near View Post
    From a site administration POV that sounds like a nightmare...

    I appreciate the idea sounds appealing but, well, it also sounds like a ton of work.
    Really the only work from an Admin point of view would be to number each member, easy with a program with only 1 loop, then private message them the results.

Similar Threads

  1. ADISC Book Circle!
    By Dawes in forum Off-topic
    Replies: 27
    Last Post: 23-Feb-2008, 01:47
  2. Uncle ADISC needs you...
    By Moo in forum News
    Replies: 0
    Last Post: 29-Jan-2008, 21:51

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •  
ADISC.org - the Adult Baby / Diaper Lover / Incontinence Support Community.
ADISC.org is designed to be viewed in Firefox, with a resolution of at least 1280 x 1024.