+ Reply to Thread
Results 1 to 6 of 6

Thread: God you people are clever!

  1. #1
    Points: 1,917, Level: 26
    Level completed: 17%, Points required for next Level: 83

    Posts
    2
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Exclamation God you people are clever!




    Hi,
    I'm collecting data from a single group of 15 participants showing their improvement in a 5 stage levelled test before and after an intervention. I will have a before and after level for each participant (and therefore an improvement for each) and the nationally expected improvement in level with and without the intervention. It's been a while since I had to do anything like this so I'm a bit confused! I want to show that the intervention significantly increases the rate of improvement.... but what test should I do? (The closest I've found is an Independent Measures T-Test comparing a single group with an exernal measure but I REALLY don't know what I'm talking about!)

    PLEASE HELP ME...

    Thanks x

  2. #2
    Super Moderator
    Points: 31,766, Level: 100
    Level completed: 0%, Points required for next Level: 0
    bugman's Avatar
    Posts
    2,255
    Thanks
    290
    Thanked 324 Times in 265 Posts
    I think a paired t-test is what you are after:

    http://udel.edu/~mcdonald/statpaired.html

    Phil

  3. #3
    Points: 1,917, Level: 26
    Level completed: 17%, Points required for next Level: 83

    Posts
    2
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Thumbs up

    Thank you so much.

    Is this too good to be true? I put in the improvement result compared to the national expectation of result and it says it's 'extremely statisticaly significant' (see cut and paste below)! Have I done it wrong or has my research actually worked?! Can I say somthing like... 'The Paired T-test shows that the rate of progress after the intervention is significantly better than the expected rate of progress with no intervention at a 95%confidnece level' ?

    I wish I'd remembered all this stuff!

    Thanks again


    Paired t test results
    P value and statistical significance:
    The two-tailed P value equals 0.0009
    By conventional criteria, this difference is considered to be extremely statistically significant.

    Confidence interval:
    The mean of improvement minus prediction equals 0.900
    95% confidence interval of this difference: From 0.441 to 1.359

    Intermediate values used in calculations:
    t = 4.2094
    df = 14
    standard error of difference = 0.214

    Learn more:

    GraphPad's web site includes portions of the manual for GraphPad Prism that can help you learn statistics. First, review the meaning of P values and confidence intervals. Next check whether you chose an appropriate test. Then learn how to interpret results from an unpaired or paired t test. These links include GraphPad's popular analysis checklists.

    Review your data:


    Group improvement prediction
    Mean 1.400 0.500
    SD 0.828 0.000
    SEM 0.214 0.000
    N 15 15

  4. #4
    Points: 3,490, Level: 36
    Level completed: 94%, Points required for next Level: 10

    Posts
    29
    Thanks
    0
    Thanked 0 Times in 0 Posts
    I do not wish to burst your bubble but I do not think you are interpreting this test correctly. The paired sample t-test compares too groups of data (both from the same sample) to determine if there is a difference. So the significant difference you are finding is that the intervention is leading to a significant improvement. However, it does not tell you whether your improvement was significantly better than the "expected" improvement. The best way to determine whether your intervention leads to a significantly better improvement is to get before and after data from two groups (one with your intervention and one without) and perform a repeated-measures 2 x 2 ANOVA.

  5. #5
    Points: 2,198, Level: 28
    Level completed: 32%, Points required for next Level: 102

    Posts
    275
    Thanks
    0
    Thanked 1 Time in 1 Post
    Assuming the test statistics at the bottom of your post are correct, then you have done the math correctly. The results are highly significant. In fact, since you were trying to show that the mean of the test group is higher (rather than just different) than the national average, you can cut the p-value in half... not that it matters when it is that small already.

    Congrats, your research shows what you wanted it to show.
    Last edited by squareandrare; 05-08-2010 at 01:00 AM.

  6. #6
    Points: 2,198, Level: 28
    Level completed: 32%, Points required for next Level: 102

    Posts
    275
    Thanks
    0
    Thanked 1 Time in 1 Post

    Quote Originally Posted by staryeyed View Post
    it does not tell you whether your improvement was significantly better than the "expected" improvement.
    You are completely able to test the hypothesis that the mean improvement is greater than some given value (in this case, a national average). Once you have the improvement scores, it is basically a one-sample t-test.

+ Reply to Thread

           




Similar Threads

  1. Probability for 20 people
    By ae5 in forum Probability
    Replies: 5
    Last Post: 12-16-2010, 06:55 PM
  2. Dividing 4 people into 4 groups
    By mobius0 in forum Probability
    Replies: 1
    Last Post: 07-12-2009, 06:37 AM
  3. Probability out of group people
    By littlequen in forum Probability
    Replies: 2
    Last Post: 12-26-2008, 03:04 PM
  4. Replies: 0
    Last Post: 12-14-2007, 09:48 AM
  5. Easy question for you people
    By martyncarr in forum Probability
    Replies: 2
    Last Post: 03-07-2006, 12:31 PM

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