TRAN-STAT statistics for environmental studies, issue No. 24, August 1983. Field sampling designs, simple random and stratified random sampling
Several methods are discussed by which sampling locations and times for environmental studies may be selected including haphazard, judgment, probability, and systematic methods. In practice, a given sampling plan may be a composite of several methods (excluding haphazard sampling which should always be avoided). The probability methods such as simple and stratified random sampling have the nice statistical property of being statistically unbiased, but systematic sampling may give a more accurate estimate of means and totals. Judgment must be used in defining the target population of units to be sampled, and the patterns of variation over time and space that should be considered when designing the study. But biased estimates of parameters can result if an individual chooses subjectively the exact location and times a sample or measurement is taken. The allocation of samples to different regions or strata (in time or space) can be determined using formulas presented here if there is advance information available on costs of sampling and the variances for each strata. Some suggestions are made for how to treat less-than and negative data when computing means and variances. The effects on anti x and s/sup 2/(anti x) when measurement errors, perhaps correlated, are present is also discussed. The method of interpenetrating subsamples is given as a way of estimating Var(anti x) when measurement errors within subsamples are correlated. 40 references.
- Research Organization:
- Pacific Northwest Lab., Richland, WA (USA)
- DOE Contract Number:
- AC06-76RL01830
- OSTI ID:
- 5982901
- Report Number(s):
- PNL-SA-11551; ON: DE83016826
- Country of Publication:
- United States
- Language:
- English
Similar Records
TRAN-STAT, Issue No. 2, November 1977. Topics discussed: some statistical aspects in the evaluation of environmental transuranic contamination for possible remedial action
Spectroscopic Needs for Imaging Dark Energy Experiments