Home > Statistics Every Writer Should Know > The Stats Board > Discusssion
Standard T-test
Message posted by Kathy (via 199.93.107.66) on January 10, 2002 at 2:14 PM (ET)
I have forgotten how to do a standard t-test. The overall issue is to prove statistical significance. This is the only thing that I can't remember.
READERS RESPOND:
(In chronological order. Most recent at the bottom.)
Re: Standard T-test
Message posted by JG (via 128.8.22.77) on January 11, 2002 at 9:02 AM (ET)
When the mean and SD are estimated from a sample - n LT 30 - and you wish to test if the difference of the mean from some given number is significant, you do a t-test.
Re: Standard T-test
Message posted by Jos_Belgium (via 195.0.29.11) on January 14, 2002 at 3:49 AM (ET)
First of all, you have to check your data. Your data must be continuous, not discrete or attribute. T-tests can only be performed with continious data.You can do two things with T-tests:
first, to show a significant difference between the mean of two data groups
second, to compare against a known value. (could be the real mean (µ))
You must carefully analyse your data and results. T-values are in units of st.deviations. They are also associated with a probability. That is expressed as a p-value.
example: T-value = -1.97 and the P-value = 0.085
This says that if the effect associated with that T-value is really 0.00 (no effect), then the chance of getting a T-value of -1.97 is 8.5%.
Most off the time you will use 95 confidence level. Then the border will be 5%. P-value below 5% means that you will accept a difference. (reject the H0 hyphotesis)
I hope that this will help a little bit.
Your $5 contribution helps cover part the $500 annual cost of keeping this site online.