This function is used when comparing at least two strains. It tests whether the mortality-dose regressions are similar for different strains, using a likelihood ratio test (LRT). If there are more than two strains, it also computes pairwise tests, using sequential Bonferroni correction (Hommel, 1988) to account for multiple testing.
Value
a list with model outputs: a chi-square test if there are only two strains or if there are more than two strains, first an overall model assessment (i.e. one strain vs. all) and given overall model is significant, then a bonferroni test of significance from a pairwise model comparison.
Details
A global LRT test assesses a strain’s effect, by comparing two models, one with and one without this effect (i.e. comparing a model with several strains to a model where all the data originate from a single strain). If there are more than two strains, pairwise tests are computed, and p-values of significance are assessed using sequential Bonferroni correction (Hommel, 1988) to account for multiple testing.
Warning: We strongly encourage users to not use this function when the dose-mortality response for at least one strain significantly deviates from linearity (see resist.ratio() function for more details): in such cases the test cannot be interpreted.
Examples
data(bioassay)
transd<-probit.trans(bioassay$assay2)
data<-transd$tr.data
model.signif(data)
#> Analysis of Deviance Table
#>
#> Model 1: mortality ~ log10(data$dose)
#> Model 2: mortality ~ log10(data$dose) * data$strain
#> Resid. Df Resid. Dev Df Deviance Pr(>Chi)
#> 1 32 936.46
#> 2 28 70.69 4 865.77 < 2.2e-16 ***
#> ---
#> Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
#> complete model is significant against a NULL model
#> continueing to pair-wise comparison
#> Output details
#> model.pval - significance value of ANOVA on the binomial GLM test of the strain pair
#> bonferroni - significance of the model.pval with bonferroni correction
#> res.Dv - residual deviance
#> thr - threshold for the significance of the pvalue
#> str - values for the strains
#> int - values for the interaction between the strain and the dose
#>
#> $model
#> strain1 strain2 model.pval bonferroni res.Dv.Null res.Dv.str res.Dv.int
#> 1 DZOU DZOU2 0.26402 non-sig NA NA NA
#> 2 DZOU KIS-ref 0 sig 1150.41 87.892 49.747
#> 3 DZOU2 KIS-ref 0 sig 1064.79 95.043 53.537
#> str.pval str.thr int.pval int.thr
#> 1 NA NA NA NA
#> 2 0 0.01250 0.001 0.025
#> 3 0 0.01667 0.001 0.050
#>