This site is a curated archive of material originally developed for, and presented at www.setac.org. This archive was prepared at the request of SETAC. Information presented represents best available consensus scientific opinion at the time originally published and may no longer reflect status quo or regulatory positions. This site is dedicated to those government, business and academic scientists that committed so much work to this effort.

What are the advantages and disadvantages of hypothesis testing and point-estimates?

WET compliance decisions are based upon statistical methods which are influenced by the experimental design and have inherent strengths and weaknesses. A comprehensive discussion of the advantages and disadvantages of hypothesis testing and point estimates as they relate to WET testing can be found in the SETAC publication on Whole Effluent Toxicity Testing (Chapman et al., 1996) and are summarized below.

Advantages of Hypothesis Testing

1. Well suited for comparing a treatment with the control.

2. Relatively simple to calculate.

Disadvantages of Hypothesis Testing

1. Dependent on concentrations tested.

2. Statistical power is influenced by variability.

3. Inability to calculate confidence intervals.

4. Confounded by hormesis or poorly behaved data.

5. Frequently need to use non-parametric statistical methods.

Advantages of Point Estimates

1. Establishes a concentration response relationship using data from all treatments.

2. Point estimates do not have to be a tested concentration.

3. Precision estimates and confidence intervals are provided.

4. Many model choices.

5. Can be used for all types of data.

6. Capability for advanced applications.

Disadvantages of Point Estimates

1. Level of effect must be selected.

2. Accuracy compromised with lack of partial responses.

3. Is model-fit dependent.

4. Construction and behavior of confidence intervals for low effect levels.

5. More sophisticated computations required.

6. Requires greater knowledge of statistical tools.

7. Is not widely used for chronic toxicity tests in the WET program.

Grothe et al (1996) also discuss ways to address or investigate these disadvantages, however, it must be recognized that the connection between biological and statistical significance is based more on how the tests are being performed (consideration of exposure variables) than the statistical analysis employed.

Chapman GA, Anderson BS, Bailer AJ, Baird RB, Berger R, Burton DT, Denton DL, Goodfellow WL, Heber MA, McDonald LL, Norberg-King TJ, Ruffier PJ, 1996. Discussion Synopsis. In: Grothe, D. R., K. L. Dickson, and D. K. Reed-Judkins, eds.1996. Whole Effluent Toxicity Testing: An Evaluation of Methods and Prediction of Receiving System Impacts, SETAC Press, Pensacola, FL, USA. 51-78 p.

Grothe, D. R., K. L. Dickson, and D. K. Reed-Judkins, eds.1996. Whole Effluent Toxicity Testing: An Evaluation of Methods and Prediction of Receiving System Impacts, SETAC Press, Pensacola, FL, USA. 340 p.

Comments are closed.