REFERENCES

The surprising thing about this data set is how closely the ratio
estimates do agree as compared to the 235u data, despite the apparent
similarity in structure of the two data sets.

Aitchison, J., and J.A.C. Brown.

1969.

The Lognormal Distribution.

Cambridge University Press, London, 176 pp.

CONCLUSIONS

Cochran, W. G. 1963,

Sampling Techniques.

Conover, W. J. 1971.

Praetical Nonparametric Statistics.

Wiley and Sons, New York.
We have discussed the assumptions underlying some ratio estimates,
thelr standard errors, and confidence intervals. Their behavior was
compared for three sets of data from NAEG studies. As a consequence
of this, there are several points that should be considered before
calculating an average ratio. These are
1.

Wiley and Sons, New York.

John

Creasy, M. A. 1956. "Confidence Limits for the Gradient in the
Linear Functional Relationship". Journal of the Royal Statistical
Soetety.
Series B(18):65-69,

Plot the data for each variable against a spatial or time
coordinate to see if there is a pattern.
Calculating an
overall ratio when there is a trend with location or time

Emery, R. M., D. C. Klopfer, T. R. Garland, and W. C. Weimer.
1976.
"The Ecological Behavior of Plutonium and Americium in a Freshwater

data sets are possible examples.

ecology Symposium. C. E. Cushing, Ed., Dowden, Hutchison and Ross,
Inc., Stroudsburg, PA. pp. 74-85.

may obscure an important relationship.

Pond".

The 235 and 238y

2.

Plot the data (Y versus X) to see if a linear relationship
through the origin seems reasonable.
If not, an average
ratio may not be the appropriate quantity to describe the
phenomena.

3.

Order the ratios to get an indication of their distribution,
particularly with respect to skewness.
If the distribution
of the ratios is skewed,

uranics in Desert Environment.
Report NVO-153.

618

1976.

"Distribution of Ratios

Street, N.W., Washington, D.C. 20005.

The sample median of the ratios is a "safe" estimate of the

average ratio in the sense that no assumption on the type
of distribution is required. However, the length of the
confidence interval may be large.

EY-76-C-06-1830.

M. G. White and P. B. Dunaway, Eds.

for Some Selected Bivariate Probability Functions". In: Proc. Social
Stattatice Seetion of American Statistical Association. 806 15th

Y,/X-

\

pp. 339-448.

Mielke, P. W., Jr., and J. A. Flueck.

will be unduly influenced by extraordinarly large ratios

This paper was funded in part by the Nevada Applied Ecology Group
(NAEG), ERDA/NV, and in part by DBER, ERDA/HQ under Contract No.

Fourth National Radio-

Gilbert, R. 0., L. L. Eberhardt, E. B. Fowler, £. M. Romney, E. H.
Essington and J. E. Kinnear.
1975.
"Statistical Analysis of
239-240py and 24am Contamination of Soil and Vegetation on NAEG
Study Sites". In: The Radtoecology of Plutoniwn and Other Trans-

>>

ACKNOWLEDGMENT

In: Radicecology and Energy Resources.

Gilbert, R. 0. and L. L. Eberhardt.
1974.
"Statistical Analysis of
Pu in Soil at the Nevada Test Site--Some Results’. In: The Dynamics
of Plutonium in Desert Envtronments. P. B. Dunaway and M. G. White,
Eds. Report NVO-142. pp. 43-89.

=
Y,/X
n fi ii

4.

Second Edition, John

Ricker, W. E. 1973.

"Linear Regression Fisheries Research". Journal

of the Fisheries Research Board of Canada 30:409-434.
10.

Snedecor, G. W., and W. G. Cochran.
1967. Statistical Methods.
Sixth Edition, Iowa State University Press, Ames, Lowa.

lt.

Workshop on Environmental Research for Transuranic Elements.
1976.
Proceedings of the Workshop November 12-14, 1975.
Energy Research
and Development Administration, Division of Biomedical and Environ-

mental Research, ERDA~76/134.

Select target paragraph3