The pooled estimate of the population SD is sqrt{ [10.672 × 9 + 9.302 × 4] / 13 } = 10.27.
The estimated SE of the difference between the sample means is 10.27 × sqrt{ 1/10 + 1/5 } = 5.62
The 97.5 percentile of the t distribution
with 13 degrees of freedom is 2.16. (In R, use
qt(0.975,13)
.)
Thus the 95% confidence interval is (103-67) ± 2.16 × 5.62 = 36 ± 12 = (24, 48).
In R, if x
is the data for
the sample from strain A and y
is the data for the
sample from strain B, type
t.test(x,y,var.equal=TRUE)
to get the above
confidence interval.
The 97.5 percentile of the t distribution
with 5 degrees of freedom = qt(0.975, 5)
=
2.57.
Thus the 95% confidence interval for the population mean is 100.8 ± 2.57 × 8.28 / sqrt{6} = 100.8 ± 8.7 = (92.1, 109.5).
The 95% confidence interval for the population SD is (8.28 × sqrt{5/12.8}, 8.28 × sqrt{5/0.831}) = (5.2, 20.3).
The difference between the sample means is 96.58-92.33=4.25. Our estimate of the standard error of the difference between the sample means is sqrt{29.092/12 + 12.172/9} = 9.33.
The "degrees of freedom" parameter we use for the t-distribution is [29.092/12 + 12.172/9]2 / {[29.092/12]2/11 + [12.172/9]2/8} = 15.56.
The 97.5 percentile of the t distribution
with 15.56 degrees of freedom = qt(0.975, 15.56)
= 2.12.
Finally, the 95% confidence interval for the difference between the population means is 4.25 ± 2.12 × 9.33 = 4.3 ± 19.8 = (-15.6, 24.1).
Last modified: Wed Feb 22 09:43:23 EST 2006 |