In: Statistics and Probability
A study was conducted to determine the long-term health eects of workplace
exposure to the process of manufacturing the herbicide (2,4,5 trichlorophenoxy)
acetic acid. The outcome of interest was fetal deaths. Suppose the investigators
expect, given general population pregnancy statistics at the tiem of the survey
that 1:5% of pregnancies will result in a fetal death. They found that among 586
pregnancies where an exposed worker was the father, 11 resulted in a fetal death. Is
this enough evidence to claim that exposure to the herbicide has an adverse eect
on survival of children born to men exposed to the herbicide?
Let be the true proportion of pregnancies, when the father is exposed to herbicide, that will result in fetal death. We can say that exposure to the herbicide has an adverse effect on on survival of children born to men exposed to the herbicide if the propotion of fatalities is more than 1.5% that occurs in general population, that is we want to test if
that is we want to test the following hypotheses
We have the following information from a sample of 586 pregnancies
The standard error of the proportion is
Since the value of n*p = 11, and is greater than 5 we use normal
distribution for testing the hypotheses
The test statistics is
This is a one tailed test (The alternative hypothesis is >0.015). The critical value of z for significance level alpha =0.05 is given by
Using standard normal tables we get
the sample test statistics of 0.80 is less than the critical value 1.64. That means we do not reject the null hypothesis
We conclude that there is no sufficient evidence to support the claim that exposure to the herbicide has an adverse effect on on survival of children born to men exposed to the herbicide