MAT 2379, Introduction to biostatistics, Assignment 4 1

MAT 2379, Introduction to biostatistics

Solution to Assignment 4
(Total = 20 points)

[2] Problem 8.4 A 95% confidence interval for the mean PVR for patients with heart failure associated to left ventricular
dysfunction is
s
x t = [183.26; 285.94],
n
where t = t0,025;29 = 2.045, x = 234.6, s = 137.5 and n = 30.

Problem 8.6
(a) A 95% confidence interval for the mean is of the form x 1.96 s/ n. Subtracting the upper limit from the
lower limit, we get:
6.5921 5.3279 = 2 (1.96 s/ n) s/ n = 0.3225.
Therefore, the estimated standard error for the
estimate
of the mean is 0.3225 mg/100 cc.
(b) The sample standard deviation is s = n(s/ n) = 150(0.3225) = 3.9498 mg/100 cc.
(c) Since we want a larger percentage of the confidence intervals to contain the value of the population mean,
then it should be natural to expect that a 97% confidence interval will be
larger than a 95% confidence interval.
(d) A 97% confidence interval for the mean is of the form x z s/ n, where z satisfies 0.97 = P (z < Z < z)
or equivalently P (Z < z) = 0.985. From Table 18.3, we get z = 2.17. So the 97% confidence interval for is

5.96 2.17 (0.3225) = [5.26; 6.66].

Problem 9.2 We denote by p the proportion of patients treated with the new drug who will have a recurrent UTI.
We would like to test p = 0.1 against p < 0.1. A point estimate for p is p = 29/347 = 0.0836. The observed value of
the test statistic is:
p 0.10
z0 = p = 1.02
(0.1)(0.9)/347
From Table 18.2, we see that p-value= P (Z < 1.02) = 0.1539. Since the p-value is larger than = 0.05, we do not
reject H0 . There is not enough evidence that p is smaller than 0.10.

Problem 9.6 Let p be the true success rate of PN in treating kidney stones. p
[2] (a) A point estimate for p is p = 289/350 = 0.8257 and its estimated standard error is s{p} = p(1 p)/n = 0.02028.
[4] (b) We want to test H0 : p = 0.78 against H1 : p 6= 0.78. The observed value of the test statistic is

p 0.78
z0 = p = 2.06.
.78(1 .78)/350

The p-value is 2 P (Z > 2.06) = 2 (1 0.9803) = 0.0394. At a level of significance of 1%, the evidence that the success
rate of PN in treating kidney stones is different than the success rate of open surgery is not significant.
[2] (c) A 95% confidence interval for p is
r
p(1 p)
p 1.96 = [0.786; 0.865].
n
We are 95% confident that the success rate of PN in treating kidney stones is between 78.6% and 86.5%. We are
95% confident that PN is more successful for treating kidney stones than open surgery.
 MAT 2379, Introduction to biostatistics, Assignment 4 2

[4] Problem 9.10 We want to test H0 : = 14 versus H1 : > 14. The test statistic is:
x 0 14.4 14
t0 = =p = 5.656
s/ n 0.025/5

From Table 18.4, we see that the p-value= P (T4 > 5.565) is smaller than 0.005. This means that the p-value is also
smaller than 0.05. Hence, we reject H0 , and conclude that the concern is justified.

Additional question:
z
/2 1.645 0.5 2
2
n =( ) = 6765.062.
0.01 0.01
We round it up to n = 6766.

Part (II)
[2] 1. a)

> carbon= read.table(file.choose(),header=TRUE,sep="\t")

Here is the quantile-quantile plot. There is a linear tendency in the QQ-plot with slight deviances in the tails.
It is reasonable to assume that the carbon monoxide concentration is normally distributed.

[1] b) Let be the mean carbon monoxide concentration. We want to test H0 : = 100 against H1 : > 100.
 MAT 2379, Introduction to biostatistics, Assignment 4 3

[2] c)

> t.test(x,mu=100,alternative="greater",conf.level=0.9)

One Sample t-test

data: x
t = 0.8778, df = 24, p-value = 0.1944
alternative hypothesis: true mean is greater than 100
90 percent confidence interval:
99.49269 Inf
sample estimates:
mean of x
101.012

The p-value is 19.44% and at a level of significance of 10%, the evidence against the null hypothesis is not
significant. We cannot conclude that the mean monoxide concentration is larger than 100 ppm.

[1] d) We failed to reject the null hypothesis, this means that we committed an error of type II, if we did indeed
commit an error.