+ Reply to Thread
Results 1 to 2 of 2

Thread: Biomed experiment - am I using unpaired t-test correctly?

  1. #1
    Points: 4, Level: 1
    Level completed: 7%, Points required for next Level: 46

    Posts
    1
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Biomed experiment - am I using unpaired t-test correctly?




    Iíve just completed a biomed experiment to examine the hypothesis that exposing follicles to a particular substance would affect their growth. However, Iím having trouble deciding whether or not the results are statistically significant, and would be grateful for confirmation that the test Iíve used (unpaired t-test) is appropriate and correctly applied.

    I had 2 groups (Test and Control) each comprising 8 follicles. I determined the average diameter (in micrometres) of each follicle 4 times (once at the start, then daily for three days). I then calculated the diameter of each follicle as a percentage of its starting size (so all normalised values for Day 0 were 100). Then I did an unpaired t-test on the 2 groups of 8 normalised measurements for Day 3, which indicated that my results were not statistically significant. My main question is Ė given the sample sizes and the methodology Iíve followed, have I applied the statistical test correctly, and am I safe in concluding that the results are not statistically significant?

    However, I was also advised that I should be looking at normalised daily averages for each group rather than individual samples. Plugging these 8 values (2 groups, 4 days) into an online paired t-test calculator again concluded that my results werenít statistically significant. But noting that that the first values in each group are identical (they are both 100), I wondered if they should reasonably be excluded, so I tried plugging in only six values (ie, those for Days 1, 2 and 3 for each group) and this time the paired t-test result was determined to be statistically significant. But Iím not confident that a paired t-test on the normalised daily averages gives me a usable results, given the limited number of data points. Any comments on this or on other aspects of my methodology would also be appreciated. Thanks for any help you can offer!

  2. #2
    TS Contributor
    Points: 18,889, Level: 87
    Level completed: 8%, Points required for next Level: 461
    CowboyBear's Avatar
    Location
    New Zealand
    Posts
    2,062
    Thanks
    121
    Thanked 427 Times in 328 Posts

    Re: Biomed experiment - am I using unpaired t-test correctly?


    Hi there, unfortunately your post was caught in our spam filter for some reason - I've approved it now. Sorry for the delay.

    Quick thoughts from me:

    Then I did an unpaired t-test on the 2 groups of 8 normalised measurements for Day 3, which indicated that my results were not statistically significant.
    This seems ok. Here you're comparing the normalised post-test measurements across the two groups.

    Plugging these 8 values (2 groups, 4 days) into an online paired t-test calculator again concluded that my results weren’t statistically significant. But noting that that the first values in each group are identical (they are both 100), I wondered if they should reasonably be excluded, so I tried plugging in only six values (ie, those for Days 1, 2 and 3 for each group)
    Neither of these strategies really make sense to me. A paired t-test is typically used for comparing before and after measurements for a single sample. Here you're comparing two different groups - the pairing by day doesn't make sense. Excluding data also doesn't seem like a great strategy.

    The simplest way to use all your data in this case is a mixed ANOVA with group as a between-subjects factor and day as a within-subjects/repeated measures factor.
    Matt aka CB | twitter.com/matthewmatix

+ 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