|
1. Normal distribution test for reproduction and growth data;
2. Binomial distribution test for survival/mortality data;
3. If the data didn’t follow the normality or binomially, the data could be transformed with either log of sqr or …..It should be tried until the data follow the distribution, otherwise the data can only test with non parametric methods;
4. Perform one way ANOVA to test the significancy. Normally, if p<0.5, then the results are considered to be significantly different in 0.05 level and multiple comparisons should be performed afterwards.
5. Many choices within multiple comparison should be used for different purpose. For instance, Dunnett’s test for NOEC/LOEC level; LSD/Duncan/Turkey test for comparing every two of treatments (The data set decide the test method).
6. About EC 50, Dose-response curve should be formulated to detect the EC value and confident intervals.
Archiver|手机版|科学网 ( 京ICP备07017567号-12 )
GMT+8, 2024-5-19 18:09
Powered by ScienceNet.cn
Copyright © 2007- 中国科学报社