emmeans migration¶
These examples assume an existing data frame d and fitted models. Import the
named functions from rparity. For statsmodels examples, import
statsmodels.api as sm and statsmodels.formula.api as smf.
See numerical accuracy and limitations and the option-level coverage when reproducing an R analysis.
Control contrasts require an explicit supported adjustment. The R dunnettx
default is unavailable; Python defaults to none.
| Use | R | Python |
|---|---|---|
| Marginal means | emmeans(m,"a") |
emmeans(m,"a") |
| Two factors | emmeans(m,c("a","b")) |
emmeans(m,["a","b"]) |
| Conditioning | emmeans(m,"a",by="b") |
emmeans(m,"a",by="b") |
| Formula conditioning | emmeans(m,~a\|b) |
emmeans(m,"~a\|b") |
| At a numeric value | emmeans(m,"a",at=list(x=2)) |
emmeans(m,"a",at={"x":[2]}) |
| At several values | emmeans(m,"x",at=list(x=c(0,1,2))) |
emmeans(m,"x",at={"x":[0,1,2]}) |
| Equal weights | emmeans(m,"a",weights="equal") |
emmeans(m,"a",weights="equal") |
| Proportional weights | emmeans(m,"a",weights="proportional") |
emmeans(m,"a",weights="proportional") |
| Cell weights | emmeans(m,"a",weights="cells") |
emmeans(m,"a",weights="cells") |
| Outer weights | emmeans(m,"a",weights="outer") |
emmeans(m,"a",weights="outer") |
| Flat weights | emmeans(m,"a",weights="flat") |
emmeans(m,"a",weights="flat") |
| Pairwise | pairs(e) |
pairs(e) |
| Reverse pairs | pairs(e,reverse=TRUE) |
pairs(e,reverse=True) |
| Pairwise contrast | contrast(e,"pairwise") |
contrast(e,"pairwise") |
| Control contrasts | contrast(e,"trt.vs.ctrl",ref=1,adjust="bonferroni") |
contrast(e,"trt.vs.ctrl",ref=1,adjust="bonferroni") |
| Consecutive contrasts | contrast(e,"consec") |
contrast(e,"consec") |
| Polynomial contrasts | contrast(e,"poly") |
contrast(e,"poly") |
| Custom contrasts | contrast(e,list(diff=c(1,-1,0))) |
contrast(e,{"diff":[1,-1,0]}) |
| Tukey adjustment | pairs(e,adjust="tukey") |
pairs(e,adjust="tukey") |
| Bonferroni | pairs(e,adjust="bonferroni") |
pairs(e,adjust="bonferroni") |
| Holm | pairs(e,adjust="holm") |
pairs(e,adjust="holm") |
| Sidak | pairs(e,adjust="sidak") |
pairs(e,adjust="sidak") |
| No adjustment | pairs(e,adjust="none") |
pairs(e,adjust="none") |
| FDR | pairs(e,adjust="fdr") |
pairs(e,adjust="fdr") |
| Response scale | summary(e,type="response") |
e.summary(type="response") |
| Regrid | regrid(e,transform="response") |
regrid(e,transform="response") |
| KR degrees of freedom | emmeans(m,"a",lmer.df="kenward-roger") |
emmeans(m,"a",lmer_df="kenward-roger") |
| Satterthwaite | emmeans(m,"a",lmer.df="satterthwaite") |
emmeans(m,"a",lmer_df="satterthwaite") |
| Asymptotic | emmeans(m,"a",lmer.df="asymptotic") |
emmeans(m,"a",lmer_df="asymptotic") |
| Trends | emtrends(m,"a",var="x") |
emtrends(m,"a",var="x") |
| Joint tests | joint_tests(m) |
joint_tests(m) |
| Confidence limits | confint(e,level=.90) |
e.confint(level=.90) |
| Tests and intervals | summary(e,infer=c(TRUE,TRUE)) |
e.summary(infer=(True,True)) |