![[Experimental]](figures/lifecycle-experimental.svg)
A copy from VCA::VCAinference in VCA package
Arguments
- ...
Arguments passed on to
VCA::VCAinferenceobj(object) of class 'VCA' or, alternatively, a list of 'VCA' objects, where all other argument can be specified as vectors, where the i-th vector element applies to the i-th element of 'obj' (see examples)
alpha(numeric) value specifying the significance level for \(100*(1-alpha)\)% confidence intervals.
total.claim(numeric) value specifying the claim-value for the Chi-Squared test for the total variance (SD or CV, see
claim.type).error.claim(numeric) value specifying the claim-value for the Chi-Squared test for the error variance (SD or CV, see
claim.type).claim.type(character) one of "VC", "SD", "CV" specifying how claim-values have to be interpreted:
"VC" (Default) = claim-value(s) specified in terms of variance(s),
"SD" = claim-values specified in terms of standard deviations (SD),
"CV" = claim-values specified in terms of coefficient(s) of variation (CV) and are specified as percentages.
If set to "SD" or "CV", claim-values will be converted to variances before applying the Chi-Squared test (see examples).VarVC(logical) TRUE = if element "Matrices" exists (see
anovaVCA), the covariance matrix of the estimated VCs will be computed (seevcovVC, which is used in CIs for intermediate VCs if 'method.ci="sas"'. Note, this might take very long for larger datasets, since there are many matrix operations involved. FALSE (Default) = computing covariance matrix of VCs is omitted, as well as CIs for intermediate VCs.excludeNeg(logical) TRUE = confidence intervals of negative variance estimates will not be reported.
FALSE = confidence intervals for all VCs will be reported including those with negative VCs.
See the details section for a thorough explanation.constrainCI(logical) TRUE = CI-limits for all variance components are constrained to be >= 0.
FALSE = unconstrained CIs with potentially negative CI-limits will be reported.
which will preserve the original width of CIs. See the details section for a thorough explanation.ci.method(character) string or abbreviation specifying which approach to use for computing confidence intervals of variance components (VC). "sas" (default) uses Chi-Squared based CIs for total and error and normal approximation for all other VCs (Wald-limits, option "NOBOUND" in SAS PROC MIXED); "satterthwaite" will approximate DFs for each VC using the Satterthwaite approach (see
SattDFfor models fitted by ANOVA) and all Cis are based on the Chi-Squared distribution. This approach is conservative but avoids negative values for the lower bounds.quiet(logical) TRUE = will suppress any warning, which will be issued otherwise
Examples
data(glucose)
fit <- anovaVCA(value ~ day / run, glucose)
VCAinference(fit)
#>
#>
#>
#> Inference from (V)ariance (C)omponent (A)nalysis
#> ------------------------------------------------
#>
#> > VCA Result:
#> -------------
#>
#> Name DF SS MS VC %Total SD CV[%]
#> 1 total 64.7773 12.9336 100 3.5963 1.4727
#> 2 day 19 415.8 21.8842 1.9586 15.1432 1.3995 0.5731
#> 3 day:run 20 281 14.05 3.075 23.7754 1.7536 0.7181
#> 4 error 40 316 7.9 7.9 61.0814 2.8107 1.151
#>
#> Mean: 244.2 (N = 80)
#>
#> Experimental Design: balanced | Method: ANOVA
#>
#>
#> > VC:
#> -----
#> Estimate CI LCL CI UCL One-Sided LCL One-Sided UCL
#> total 12.9336 9.4224 18.8614 9.9071 17.7278
#> day 1.9586
#> day:run 3.0750
#> error 7.9000 5.3251 12.9333 5.6673 11.9203
#>
#> > SD:
#> -----
#> Estimate CI LCL CI UCL One-Sided LCL One-Sided UCL
#> total 3.5963 3.0696 4.3430 3.1476 4.2104
#> day 1.3995
#> day:run 1.7536
#> error 2.8107 2.3076 3.5963 2.3806 3.4526
#>
#> > CV[%]:
#> --------
#> Estimate CI LCL CI UCL One-Sided LCL One-Sided UCL
#> total 1.4727 1.257 1.7785 1.2889 1.7242
#> day 0.5731
#> day:run 0.7181
#> error 1.1510 0.945 1.4727 0.9749 1.4138
#>
#>
#> 95% Confidence Level
#> SAS PROC MIXED method used for computing CIs
#>