Help making sense of a paired t-test

Cowboy Bebop

Can anyone help me translate this gobbledygook(paired t-test results) into a human tongue, you'd be doing me a massive favor?


Proto 1 Proto 2
Mean 19.7 30.1
Variance 10.01111111 11.21111111
Observations 10 10
Pearson Correlation 0.129002205
Hypothesized Mean Difference 0
df 9
t Stat -7.64852927
P(T<=t) one-tail 1.58175E-05
t Critical one-tail 1.833112923
P(T<=t) two-tail 3.16351E-05
t Critical two-tail 2.262157158

Dark Moon

I'm guessing you're doing a two tailed test (you're not testing to see if one mean is different than the other in a certain direction). If this isn't the case, you use the one tailed p-value. Anyways, your p-value is already calculated for you (yay stats packages: P (T<=t) two-tail 3.16351E-05) and can be explained as such: There is a 0.0032% chance that your two means came from the same population and the difference can be explained by chance alone. You can say with greater than 99% certainty (alpha = 0.01, which is your type I error rate) that your two means are significantly different enough for you to reject your null hypothesis of no difference.

Cowboy Bebop

Dark Moon wrote: »

I'm guessing you're doing a two tailed test (you're not testing to see if one mean is different than the other in a certain direction). If this isn't the case, you use the one tailed p-value. Anyways, your p-value is already calculated for you (yay stats packages: P (T<=t) two-tail 3.16351E-05) and can be explained as such: There is a 0.0032% chance that your two means came from the same population and the difference can be explained by chance alone. You can say with greater than 99% certainty (alpha = 0.01, which is your type I error rate) that your two means are significantly different enough for you to reject your null hypothesis of no difference.

Thanks for the help