Results 1 to 13 of 13

Thread: A question on statistic t-test (Urgent)

  1. #1
    Member
    Join Date
    Jun 2007
    Location
    North-East
    Posts
    1,977

    Default A question on statistic t-test (Urgent)

    Hi Ppl,

    I need your help regarding the statistic t-test.

    I have conducted a t-test on comparing 2 sample means and the following are the results:
    Sample size 1 = 25, Mean 1 = 4.24, Standard Deviation 1 = 1.74.
    Sample size 2 = 25, Mean 2 = 4.08, Standard Deviation 2 = 2.10.
    Degree of Freedom = 48

    Based on formula, I got a calculated t as 0.29.
    From the t-distibution table based on degree of freedom = 48, level of significant = 0.05, two-tail, my table t-value is 2.011.

    Comparing calculated t-value of 0.29 to table t-value of 2.011,
    Does it mean that there is no significant difference in both sample? From the table am I able to find any critical upper and lower limit?

    Thanks ppl....
    In the eyes of the beholder!

  2. #2
    Member
    Join Date
    Dec 2005
    Location
    Singapore, Bedok
    Posts
    1,129

    Default Re: A question on statistic t-test (Urgent)

    hi..

    It's weird that such a question would pop up in a photography forum.. hehe.. but since it is kopitiam section, I'll make some wild guess and speculation..

    Since the calculated value is 0.29, it implies that there is 29% probability that these 2 sample "have no significant difference".

    Then, there'll be an hypothesis.. of the relationship between the 2 samples.

    If the hypothesis is the 2 samples "have no significant difference", and the level is 0.05, then the hypothesis is true since 0.29 is > 0.05

    If the hypothesis is the 2 samples "have significant difference", and the level is 0.05, then the hypothesis is false since 0.29 > 0.05

  3. #3
    Member
    Join Date
    Aug 2007
    Location
    Yishun, Singapore
    Posts
    590

    Default Re: A question on statistic t-test (Urgent)

    wish i could help.. but i returned all my statistics knowledge to my math teacher when i finished my a's last year. HAHA i hate statistics.

  4. #4

    Default Re: A question on statistic t-test (Urgent)

    Quote Originally Posted by Limsgp View Post
    hi..

    It's weird that such a question would pop up in a photography forum.. hehe.. but since it is kopitiam section, I'll make some wild guess and speculation..

    Since the calculated value is 0.29, it implies that there is 29% probability that these 2 sample "have no significant difference".

    Then, there'll be an hypothesis.. of the relationship between the 2 samples.

    If the hypothesis is the 2 samples "have no significant difference", and the level is 0.05, then the hypothesis is true since 0.29 is > 0.05

    If the hypothesis is the 2 samples "have significant difference", and the level is 0.05, then the hypothesis is false since 0.29 > 0.05
    er, t values have nothing to do with percentage, you need the read off from statistical tables to be able to tell what is the exact percentage, or estimations, anyways.

  5. #5

    Default Re: A question on statistic t-test (Urgent)

    Not sure how you got the value in bold

    You can use this website:
    http://www.graphpad.com/quickcalcs/ttest1.cfm

    This is what I had:

    Unpaired t test results

    P value and statistical significance:
    The two-tailed P value equals 0.7705
    By conventional criteria, this difference is considered to be not statistically significant.

    Confidence interval:
    The mean of Group One minus Group Two equals 0.1600
    95% confidence interval of this difference: From -0.9367 to 1.2567

    Intermediate values used in calculations:
    t = 0.2933
    df = 48
    standard error of difference = 0.545

    P = 0.775

    So it is greater than 0.05, hence the null hypothesis that the 2 samples are the same, stands.


    t is NOT = p so 0.29 is NOT "29% probability that these 2 sample "have no significant difference".

    Quote Originally Posted by undergrd View Post
    Hi Ppl,

    I need your help regarding the statistic t-test.

    I have conducted a t-test on comparing 2 sample means and the following are the results:
    Sample size 1 = 25, Mean 1 = 4.24, Standard Deviation 1 = 1.74.
    Sample size 2 = 25, Mean 2 = 4.08, Standard Deviation 2 = 2.10.
    Degree of Freedom = 48

    Based on formula, I got a calculated t as 0.29.
    From the t-distibution table based on degree of freedom = 48, level of significant = 0.05, two-tail, my table t-value is 2.011.

    Comparing calculated t-value of 0.29 to table t-value of 2.011,
    Does it mean that there is no significant difference in both sample? From the table am I able to find any critical upper and lower limit?

    Thanks ppl....
    Nikon D750; FM2; FG; 55mm Micro Nikkor; 28-300 VR; 70-200 VR; Nikon V1 + 10-30mm

  6. #6

    Default Re: A question on statistic t-test (Urgent)

    Quote Originally Posted by undergrd View Post

    I have conducted a t-test on comparing 2 sample means and the following are the results:
    Sample size 1 = 25, Mean 1 = 4.24, Standard Deviation 1 = 1.74.
    Sample size 2 = 25, Mean 2 = 4.08, Standard Deviation 2 = 2.10.
    Degree of Freedom = 48

    Based on formula, I got a calculated t as 0.29.
    From the t-distibution table based on degree of freedom = 48, level of significant = 0.05, two-tail, my table t-value is 2.011.

    Comparing calculated t-value of 0.29 to table t-value of 2.011,
    Does it mean that there is no significant difference in both sample? From the table am I able to find any critical upper and lower limit?

    Thanks ppl....
    i cannot remember the formula for this particular scenario, i remember what it is roughly like, but assuming you applied the formula correctly,

    and that t statstic is 0.29, for TWO TAILED TEST, with df = 48, significance level = 5%

    then your t stat should lie between -2.011 and 2.011 for null hypothesis to NOT be rejected. null hypothesis is that there is no significant difference between the two sample means.
    Last edited by night86mare; 10th May 2009 at 08:05 PM.

  7. #7
    Senior Member Pablo's Avatar
    Join Date
    Sep 2004
    Location
    Blue/Green Planet
    Posts
    1,854

    Default Re: A question on statistic t-test (Urgent)

    Quote Originally Posted by undergrd View Post
    Hi Ppl,

    I need your help regarding the statistic t-test.

    I have conducted a t-test on comparing 2 sample means and the following are the results:
    Sample size 1 = 25, Mean 1 = 4.24, Standard Deviation 1 = 1.74.
    Sample size 2 = 25, Mean 2 = 4.08, Standard Deviation 2 = 2.10.
    Degree of Freedom = 48

    Based on formula, I got a calculated t as 0.29.
    From the t-distibution table based on degree of freedom = 48, level of significant = 0.05, two-tail, my table t-value is 2.011.

    Comparing calculated t-value of 0.29 to table t-value of 2.011,
    Does it mean that there is no significant difference in both sample? From the table am I able to find any critical upper and lower limit?

    Thanks ppl....
    Personally,

    I would take a rough guess.

    What are you talking about anyway
    Time, is an effortless construction :)

  8. #8
    Member
    Join Date
    Dec 2005
    Location
    Singapore, Bedok
    Posts
    1,129

    Default Re: A question on statistic t-test (Urgent)

    Oops.. oh.. I had mistaken the T value for the P value.. hehe..


    Quote Originally Posted by night86mare View Post
    er, t values have nothing to do with percentage, you need the read off from statistical tables to be able to tell what is the exact percentage, or estimations, anyways.

  9. #9
    Member
    Join Date
    Jun 2007
    Location
    North-East
    Posts
    1,977

    Default Re: A question on statistic t-test (Urgent)

    Hi Ppl, thanks for cracking your brains to help...

    Sorry that this kind of qn was asked in a photography website..hehe.

    Im able to understand bit by bit.Really appreciate all your help. Thanks thanks.
    In the eyes of the beholder!

  10. #10
    Member terryansimon's Avatar
    Join Date
    Dec 2007
    Location
    Quahog, Rhode Island
    Posts
    95

    Default Re: A question on statistic t-test (Urgent)

    what is your t? 0.29?

    what is your limit? 2.011?


    thus, if you draw a bell curve with two lines on either side to mark out -2.011 and +2.011, 0.29 would fall in between the two lines. thus, H0 is rejected simply because it is not significant.

    why is it not significant? because it does not fall outside the boundaries set out by your testing limits.

    thus, you will report it as (t=0.29, p>0.05 or whatever alpha value you were told to use)





    this proves that I have used SPSS so much for stats, I have completely forgotten all the theory behind the formulas/equations used by SPSS.
    Last edited by terryansimon; 11th May 2009 at 08:55 AM.
    chicken fight!

  11. #11
    Member
    Join Date
    Jun 2007
    Location
    North-East
    Posts
    1,977

    Default Re: A question on statistic t-test (Urgent)

    Quote Originally Posted by terryansimon View Post

    thus, if you draw a bell curve with two lines on either side to mark out -2.011 and +2.011, 0.29 would fall in between the two lines.
    The 2.011 is obtained from the t-distribution table at 48 degree of freedom and 0.05 significant level. Just to ask, is it really true that the upper and lower critical value is -2.011 & 2.011? or do I have to manually calculate?
    In the eyes of the beholder!

  12. #12

    Default Re: A question on statistic t-test (Urgent)

    Calculate for you with PHStat software:

    Data
    Hypothesized Difference 0
    Level of Significance 0.05
    Population 1 Sample
    Sample Size 25
    Sample Mean 4.24
    Sample Standard Deviation 1.74
    Population 2 Sample
    Sample Size 25
    Sample Mean 4.08
    Sample Standard Deviation 2.1

    Intermediate Calculations
    Population 1 Sample Degrees of Freedom 24
    Population 2 Sample Degrees of Freedom 24
    Total Degrees of Freedom 48
    Pooled Variance 3.7188
    Difference in Sample Means 0.16
    t Test Statistic 0.293341547

    Two-Tail Test
    Lower Critical Value -2.010634722
    Upper Critical Value 2.010634722
    p-Value 0.77052566
    Do not reject the null hypothesis
    i.e. There is no significant difference between the 2 means. Null hypothesis: X1 = X2

    Confidence Interval Estimate
    for the Difference Between Two Means

    Data
    Confidence Level 95%

    Intermediate Calculations
    Degrees of Freedom 48
    t Value 2.010634722
    Interval Half Width 1.09667914

    Confidence Interval
    Interval Lower Limit -0.93667914
    Interval Upper Limit 1.25667914

  13. #13
    Member terryansimon's Avatar
    Join Date
    Dec 2007
    Location
    Quahog, Rhode Island
    Posts
    95

    Default Re: A question on statistic t-test (Urgent)

    Quote Originally Posted by undergrd View Post
    The 2.011 is obtained from the t-distribution table at 48 degree of freedom and 0.05 significant level. Just to ask, is it really true that the upper and lower critical value is -2.011 & 2.011? or do I have to manually calculate?
    well based on the information given by you in your initial post, I assume that the upper and lower limits of your two tailed test is +/- 2.011

    how these numbers are arrived at, I have no idea. if I remember correctly, the intervals or something has got something to do with it... 1.96 or something. I'm not sure.
    chicken fight!

Bookmarks

Posting Permissions

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