Details
| Original language | English |
|---|---|
| Pages (from-to) | 31-45 |
| Number of pages | 15 |
| Journal | EPPO Bulletin |
| Volume | 36 |
| Issue number | 1 |
| Publication status | Published - 30 Jun 2006 |
Abstract
The evaluation of plant protection product efficacy in field trials is a complex task. Several aspects of the statistical evaluation are discussed in this paper, along with a recommendation to modify EPPO Standard PP1/152. Confidence intervals should be preferred over significance tests to avoid contradictions between 'statistical significance' and 'biological relevance'. Efficacy should first be demonstrated, by comparison of a new product with a well chosen reference product, using a test of non-inferiority. When non-inferiority has been established, a test of superiority can follow. Both can simply be obtained by one-sided confidence intervals of μ Reference - μProduct or μReference/ μProduct, where μReference and μ Product are the means for the reference and the product, respectively. Confidence intervals for the ratio are easier to interpret because of the definition of percentage change. The evaluation that a trial is realistic and coherent should be performed according to the present EPPO guideline before any analysis of efficacy is made. Efficacy data from trial series should be preferred over individual trials because this represents different sites and increased sample size (number of sites and number of plots). The treatments-by-sites interaction should be modelled as a random effect in a mixed model. Software in R is available for these suggested analyses.
ASJC Scopus subject areas
- Agricultural and Biological Sciences(all)
- Agronomy and Crop Science
- Agricultural and Biological Sciences(all)
- Plant Science
- Agricultural and Biological Sciences(all)
- Horticulture
Cite this
- Standard
- Harvard
- Apa
- Vancouver
- BibTeX
- RIS
In: EPPO Bulletin, Vol. 36, No. 1, 30.06.2006, p. 31-45.
Research output: Contribution to journal › Article › Research › peer review
}
TY - JOUR
T1 - Statistical aspects of efficacy evaluation of plant protection products in field trials
T2 - A comment on EPPO Standard PP1/152
AU - Hothorn, L. A.
AU - Bleiholder, H.
PY - 2006/6/30
Y1 - 2006/6/30
N2 - The evaluation of plant protection product efficacy in field trials is a complex task. Several aspects of the statistical evaluation are discussed in this paper, along with a recommendation to modify EPPO Standard PP1/152. Confidence intervals should be preferred over significance tests to avoid contradictions between 'statistical significance' and 'biological relevance'. Efficacy should first be demonstrated, by comparison of a new product with a well chosen reference product, using a test of non-inferiority. When non-inferiority has been established, a test of superiority can follow. Both can simply be obtained by one-sided confidence intervals of μ Reference - μProduct or μReference/ μProduct, where μReference and μ Product are the means for the reference and the product, respectively. Confidence intervals for the ratio are easier to interpret because of the definition of percentage change. The evaluation that a trial is realistic and coherent should be performed according to the present EPPO guideline before any analysis of efficacy is made. Efficacy data from trial series should be preferred over individual trials because this represents different sites and increased sample size (number of sites and number of plots). The treatments-by-sites interaction should be modelled as a random effect in a mixed model. Software in R is available for these suggested analyses.
AB - The evaluation of plant protection product efficacy in field trials is a complex task. Several aspects of the statistical evaluation are discussed in this paper, along with a recommendation to modify EPPO Standard PP1/152. Confidence intervals should be preferred over significance tests to avoid contradictions between 'statistical significance' and 'biological relevance'. Efficacy should first be demonstrated, by comparison of a new product with a well chosen reference product, using a test of non-inferiority. When non-inferiority has been established, a test of superiority can follow. Both can simply be obtained by one-sided confidence intervals of μ Reference - μProduct or μReference/ μProduct, where μReference and μ Product are the means for the reference and the product, respectively. Confidence intervals for the ratio are easier to interpret because of the definition of percentage change. The evaluation that a trial is realistic and coherent should be performed according to the present EPPO guideline before any analysis of efficacy is made. Efficacy data from trial series should be preferred over individual trials because this represents different sites and increased sample size (number of sites and number of plots). The treatments-by-sites interaction should be modelled as a random effect in a mixed model. Software in R is available for these suggested analyses.
UR - http://www.scopus.com/inward/record.url?scp=33745617124&partnerID=8YFLogxK
U2 - 10.1111/j.1365-2338.2006.00934.x
DO - 10.1111/j.1365-2338.2006.00934.x
M3 - Article
AN - SCOPUS:33745617124
VL - 36
SP - 31
EP - 45
JO - EPPO Bulletin
JF - EPPO Bulletin
SN - 0250-8052
IS - 1
ER -