University of Wisconsin Digital Collections
Link to University of Wisconsin Digital Collections
Link to University of Wisconsin Digital Collections
Ecology and Natural Resources Collection

Page View

LeMasters, Gary S.; Doyle, Douglas J. / Grade A dairy farm well water quality survey
[DNR-058] (1989)

Results,   pp. 11-14 PDF (2.4 MB)

Page 12

The procedure estimates that atrazine could be found at detectable levels in
12% of the Grade A wells with a standard error of 1.4%. The confidence
interval for this estimate is:
[0.12 ± (1.96)(0.014)] or [0.12 ± (0.027)]
Comparison of District Proportions
Statistical Z tests (Zar, 1974) of District level proportions indicate that the
South Central District had a higher proportion of wells with any pesticide
detect than the West Central District. The test used is as follows:
(A  h       A     1/2]
z =   i- PJ / [(M/'rh +  p/'r ]
The actual district level proportion estimates for any detect, atrazine, atrazine
above the PAL and atrazine above the ES are in Table 5. The confidence
intervals for some of the estimates for some of the Districts are very wide.
Therefore, the observed geographic pattern is suggestive of the need for more
State Estimates of Mean Concentrations
The mean concentration for atrazine in wells on Wisconsin Grade A dairy
farms is estimated at 0.12 ug/l (Table 6). The standard error was estimated at
0.04 ug/l. The 95% confidence interval is:
[0.12 ± (1.96)(0.04)] or [0.12 ± (0.08)]
Comparisons of District Mean Concentrations
It is useful to compare the district level estimates. When interpreting
multiple comparisons, it is important to remember that failure to reject the
hypothesis that two or more means are equal should not lead to the conclusion
that the population means are in fact equal (SAS Institute, 1985). Multiple
comparisons were done with a Student's t-test on all possible pairs of district
level means. For the iih and jt district means you can reject the null hypothesis
that the means are equal if:
IY i--Yj /s(1/ni + I/nj) > t(a;df)
where y, and y  are the means, n  and nj are the number of observations in the
respective cells, s is the root mean square error based on the degrees of
freedom (dfj, and the significance level of a two tailed test from a student's t
distribution shown by cz (ibid.).

Go up to Top of Page