OptionalcontrastsA contrast per factor of the fixed-effects part, as in lmer (see LmerContrastName); the random-effects terms always use treatment contrasts.
OptionalfamilyThe family: "binomial", "poisson", "Gamma", "inverse.gaussian" or "gaussian" (default, as R's; with the
identity link it is a linear mixed model, a RangeError telling to use lmer). The quasi families are the
error "quasi" families cannot be used in glmer.
OptionallevelsThe order of the levels of a factor ({ A: ["c", "a", "b"] }), as factor(x, levels = ...) in R. For a binomial
response given as strings, the first level is failure and the others success.
OptionallinkThe link: any of "logit", "probit", "cauchit", "cloglog", "identity", "log", "sqrt", "1/mu^2",
"inverse" (default: the family's, as in R: binomial logit, poisson log, Gamma inverse, inverse.gaussian
1/mu^2, gaussian identity).
Optionaln1 (default): the Laplace approximation, the deviance minimised over θ and β; 0: over θ only, with β from the
penalised IRLS (faster, less exact); 2 to 100: adaptive Gauss-Hermite quadrature with that many points, more
exact than the Laplace approximation. A non-integer is truncated as in R (−0.5 is 0); negative values after that
are a RangeError (R uses 0), above 100 R's error. Above 1 the model must have a single random-effects term with one column, such as
(1 | g) (otherwise R's errors nAGQ > 1 is only available for models with a single, scalar random-effects term or AGQ only defined for a single scalar random-effects term), and the family must be binomial or
poisson: for Gamma, inverse.gaussian and gaussian it is a RangeError, since lme4 holds the dispersion at 1 there
and its estimates are not maximum likelihood (R fits them).
OptionaloffsetOffset added to the linear predictor: one per row of data (null or NaN drops the row), or a column name.
OptionalstartWhere the search begins (it does not change the estimate unless the deviance has several minima): theta, the
covariance parameters (the length of θ, diagonal elements not negative), and fixef, the fixed effects (nAGQ = 1
only), each with R's error messages.
Optionalfixef?: readonly number[]Starting values of the fixed effects (R's start = list(fixef = ...); not with nAGQ = 0, as in R).
Optionaltheta?: readonly number[]Starting values of θ (R's start = list(theta = ...)).
OptionalweightsPrior weights (non-negative): one per row of data (null or NaN is missing and drops the row), or the name of
a numeric column. For the binomial family with proportions as the response, the numbers of trials.
Options of glmer. Only the keys below are accepted: R's
subset,naAction,na.action,control,verbose,devFunOnly,mustartandetastartgive the Errorglmer: argument '<key>' is not supported, any other key a RangeError (a key whose value isundefinedincluded).