+ Reply to Thread
Results 1 to 5 of 5

Thread: Statistical plan in science fair

  1. #1

    Statistical plan in science fair




    Hi,

    I’d like comments concerning our statistical plan in a science fair project. The project tests device X against a control device. Using either device has a binary outcome of success or failure.

    First, we performed 100 trials with each device and found 66% success with X and 61% success with control. We used a Power table for Cohen’s Effect Size d = 0.2, two-tailed alpha = 0.05, and power of .80 to decide that we needed at least 393 trials with each device. We’ll try to do more trials, but hopefully we’re in the ballpark.

    Now, we don’t have the trials done yet but our next calculation would be an independent two-sample t-test for equal sample sizes, equal variance after http://en.wikipedia.org/wiki/Student's_t-test.

    Finally, with our t statistic and degrees of freedom, we’ll use a t-table to find a p-value.

    (I tried this post on another statistics site with no reply, so any expertise here is greatly appreciated.)

  2. #2
    ggplot2orBust
    Points: 71,220, Level: 100
    Level completed: 0%, Points required for next Level: 0
    Awards:
    User with most referrers
    trinker's Avatar
    Location
    Buffalo, NY
    Posts
    4,417
    Thanks
    1,811
    Thanked 928 Times in 809 Posts

    Re: Statistical plan in science fair

    I'm not sure if a t-test is the most appropriate test for a binary (yes no) outcome. Are you locked into using a t-test?
    "If you torture the data long enough it will eventually confess."
    -Ronald Harry Coase -

  3. #3
    Points: 12, Level: 1
    Level completed: 23%, Points required for next Level: 38

    Posts
    3
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: Statistical plan in science fair

    Hi Trinker,

    T-test seemed appropriate and easy enough, but we're not locked in. What would you suggest? If we stick with the t-test though, how do you like the rest of our plan?

    Thanks for asking.

  4. #4
    ggplot2orBust
    Points: 71,220, Level: 100
    Level completed: 0%, Points required for next Level: 0
    Awards:
    User with most referrers
    trinker's Avatar
    Location
    Buffalo, NY
    Posts
    4,417
    Thanks
    1,811
    Thanked 928 Times in 809 Posts

    Re: Statistical plan in science fair

    You're violating assumptions by using the t-test, one of which is that your data is drawn from a random population (You check this by looking at the residuals). Your distribution will likely binomial. Have a link at this LINK.
    "If you torture the data long enough it will eventually confess."
    -Ronald Harry Coase -

  5. #5
    Points: 12, Level: 1
    Level completed: 23%, Points required for next Level: 38

    Posts
    3
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: Statistical plan in science fair


    Hi Trinker,

    I see what you mean, but from the following link I'm getting that "If N is sufficiently large, the t probability distribution and the binomial distribution are approximated to the normal distribution." So, is my N (>393) large enough to keep it simple and follow my original plan? Or, is it 7.2 Comparing Two Proportions that I need?

    http://www.indiana.edu/~statmath/sta...st/ttest7.html

+ Reply to Thread

           




Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts






Advertise on Talk Stats