PDA

View Full Version : Student's t-Test -- just looking for confirmation



JoynerCN
06-21-2010, 11:39 PM
For this particular t-Test, I'm comparing a sample with 16 values, mean 5.34, stdev .28 to a population whose mean is 5.27, stdev .08.

I'm certain that there's not a statistically-significant value, but for this question we need to actually provide the probability.

I'd just like confirmation that I'm doing this right -- I'm a little confused that we were given the population standard deviation when I don't think it's necessary:

Calculate the t-statistic: (5.34 - 5.27) / (.28 / sqrt(16) ) = 1
Calculate the probability: .3332

(To go from t-statistic to probability, I'm using: http://www.graphpad.com/quickcalcs/pvalue1.cfm -- not sure how to do it with OpenOffice or Excel)

Is that the correct way to do this and the correct answer?

Dason
06-21-2010, 11:59 PM
Why are you using .28 in the denominator? I don't believe you're using a correct quantity for the standard deviation. Depending on if you're assuming equal variances or not there are slightly different formulas for the standard deviation to use but just picking one of the standard deviations from the groups isn't the right way to go about it.

JoynerCN
06-22-2010, 12:55 AM
That's the thing, it's a one-sample T-test, not a two-sample T-test -- I'm comparing the population mean to the sample mean. My understanding had been that you use the sample standard deviation for that formula, as per: http://en.wikipedia.org/wiki/Student%27s_t-test#Independent_one-sample_t-test