Re: Anova() in car not works after loading piecewiseSEM package

On 2026-07-19 13:56, Duncan Murdoch wrote:
> On 2026-07-19 7:39 a.m., Kamil Bartoń wrote:
>> The problem may be that `car:::model.matrix.lme` uses:
>>
>> NextMethod(formula(object), [...]
>>
>> while the first argument for `NextMethod` should be a name of a generic function. This is why
>> it is the lme object that is passed to `model.matrix.default` (which in turn looks for
>> `model.frame.lme`, which is not implemented in "nlme") instead of a formula.
>>
>> I suppose the line in  `car:::model.matrix.lme` should be changed to:
>>
>> model.frame(formula(object), [...])
>>
>> to work as intended.
> 
> You're right that NextMethod is not being called the way the docs say it should be, but it does 
> appear to dispatch correctly.  When given an "lme" object, NextMethod calls the "default" method.
> 

But the code suggests the authors' intention was to call the "default" method but with model formula 
as the first argument, and not the "lme" object.



> Duncan Murdoch
> 
>>
>> ~kB
>>
>>
>>
>>
>> On 2026-07-19 13:21, varin sacha wrote:
>>> Sorry,
>>>
>>> There is no nlme:::model.frame.lme. That makes an S3 method signature mismatch the most likely 
>>> explanation. The question of how best to resolve it is probably one for the MuMIn package 
>>> maintainer.
>>>
>>> Best
>>>
>>>
>>>> Le 19 juil. 2026 à 12:30, varin sacha via R-help <[email protected]> a écrit :
>>>>
>>>> Thanks Duncan. This seems like a promising workaround. You should try re-registering the 
>>>> multcomp method:
>>>>
>>>> registerS3method("model.frame", "lme", multcomp:::model.frame.lme)
>>>>
>>>> And then test:
>>>>
>>>> car::Anova(fm2)
>>>>
>>>> If this resolves the issue, it would provide strong evidence that the problem is caused by the 
>>>> incompatible MuMIn::model.frame.lme() S3 registration rather than by car::Anova() itself.
>>>>
>>>>
>>>>
>>>>
>>>>> Le 19 juil. 2026 à 12:14, Duncan Murdoch <[email protected]> a écrit :
>>>>>
>>>>> On 2026-07-19 4:55 a.m., Jinsong Zhao wrote:
>>>>>>> On 7/19/2026 3:52 PM, Jinsong Zhao wrote:
>>>>>>> On 7/19/2026 2:59 PM, Jinsong Zhao wrote:
>>>>>>>>
>>>>>>>> On 7/19/2026 2:38 PM, varin sacha wrote:
>>>>>>>>> Ok !
>>>>>>>>> According to what I see, the error is occurring inside
>>>>>>>>> car:::Anova.lme(), specifically during the construction of the model
>>>>>>>>> matrix.
>>>>>>>>>
>>>>>>>>> It therefore appears to be a genuine compatibility problem between
>>>>>>>>> the current CRAN versions of car (3.1-5), piecewiseSEM (2.3.1),
>>>>>>>>> and/or nlme (3.1-170).
>>>>>>>>>
>>>>>>>>> The fact that you reproduced the same behavior on both Windows and
>>>>>>>>> FreeBSD also suggests that the issue is not platform-specific.
>>>>>>>>>
>>>>>>>>> I think this would be worth reporting to the package maintainers
>>>>>>>>> (perhaps starting with car, since the traceback shows that the
>>>>>>>>> failure occurs inside Anova.lme(), while mentioning that the problem
>>>>>>>>> only arises after loading piecewiseSEM). The reproducible example
>>>>>>>>> you’ve provided should make it straightforward for them to investigate.
>>>>>>>>
>>>>>>>> The current maintainers of car and piecewiseSEM packages are also
>>>>>>>> copied on this thread. As a regular user, I am just wondering what's
>>>>>>>> behind the change in Anova()'s behavior—specifically, what gets
>>>>>>>> modified after attaching piecewiseSEM?
>>>>>>>>
>>>>>>>> Best,
>>>>>>>>
>>>>>>>> Jinsong
>>>>>>>>
>>>>>>> A small step forward toward the root of the issue: I've just
>>>>>>> discovered that the model.frame.lme() function defined in the MuMIn
>>>>>>> package is what caused this problem (so I've copied this email to the
>>>>>>> maintainer of MuMIn).
>>>>>>>
>>>>>>> However, I'm still unclear as to why Anova() calls model.frame.lme()
>>>>>>> in the first place, given that car does not depend on the MuMIn
>>>>>>> package. And directly invoking MuMIn:::model.frame.lme(fm2, random=
>>>>>>> TRUE) did not cause error.
>>>>>>>
>>>>>> I think I've found the root cause.
>>>>>> piecewiseSEM imports MuMIn, where model.frame.lme is registered as an S3
>>>>>> method:
>>>>>> S3method(model.frame, lme)
>>>>>> Its definition is:
>>>>>> model.frame.lme <- function(formula, random = FALSE, ...)
>>>>>> But Anova() calls it (per traceback()) with:
>>>>>> model.frame.lme(object, data, xlev = xlev)
>>>>>> This passes data to the random argument, causing a type mismatch and the
>>>>>> error:
>>>>>> Error in if (random) { : the condition has length > 1
>>>>>> I've diagnosed the issue, but I don't yet know how to fix it.
>>>>>
>>>>> This looks hard to fix.
>>>>>
>>>>> One problem is that there are two definitions for model.frame.lme, one from MuMIn and the other 
>>>>> from multcomp.  The one from MuMIn is being called.  If the one in multcomp was called, things 
>>>>> would be fine. Perhaps a fix could be for the MuMIn package to change its definition to 
>>>>> something compatible with the multcomp definition, but the two functions appear to do different 
>>>>> things.  I don't know if they can be made compatible.
>>>>>
>>>>> Another problem is in the stats package.  The stats:::model.matrix.default method makes a call to
>>>>>
>>>>> data <- model.frame(object, data, xlev = xlev)
>>>>>
>>>>> The definition of the generic model.frame() looks like
>>>>>
>>>>> function (formula, ...)
>>>>> UseMethod("model.frame")
>>>>>
>>>>> so stats:::model.matrix.default has no basis for assuming that the second argument is the 
>>>>> data.  Changing that call to
>>>>>
>>>>> data <- model.frame(object, data = data, xlev = xlev)
>>>>>
>>>>> would fix the issue of binding data to the "random" argument, but you'd still end up calling 
>>>>> the "wrong" method.
>>>>>
>>>>> Maybe someone else has an elegant idea to fix this?
>>>>>
>>>>> Duncan Murdoch
>>>>>
>>>>>> Best,
>>>>>> Jinsong
>>>>>>> Best,
>>>>>>>
>>>>>>> Jinsong
>>>>>>>
>>>>>>>> library(MuMIn)
>>>>>>>> library(nlme)
>>>>>>>> library(car)
>>>>>>> Loading required package: carData
>>>>>>>> fm2 <- lme(distance ~ age + Sex, data = Orthodont, random = ~ 1)
>>>>>>>> Anova(fm2)
>>>>>>> Error in if (random) { : the condition has length > 1
>>>>>>>
>>>>>>>
>>>>>>>>>
>>>>>>>>>> Le 19 juil. 2026 à 08:22, Jinsong Zhao <[email protected]> a écrit :
>>>>>>>>>>
>>>>>>>>>> Thank for the instruction. Here is the whole outputs:
>>>>>>>>>>
>>>>>>>>>>> library(piecewiseSEM)
>>>>>>>>>> Registered S3 method overwritten by 'lme4':
>>>>>>>>>>     method           from
>>>>>>>>>>     na.action.merMod car
>>>>>>>>>>
>>>>>>>>>>     This is piecewiseSEM version 2.3.0.2.
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>     Questions or bugs can be addressed to <[email protected]>.
>>>>>>>>>>> library(nlme)
>>>>>>>>>>> library(car)
>>>>>>>>>> Loading required package: carData
>>>>>>>>>>> fm2 <- lme(distance ~ age + Sex, data = Orthodont, random = ~ 1)
>>>>>>>>>>> Anova(fm2)
>>>>>>>>>> Error in if (random) { : the condition has length > 1
>>>>>>>>>>> car::Anova(fm2)
>>>>>>>>>> Error in if (random) { : the condition has length > 1
>>>>>>>>>>> car:::Anova.lme(fm2)
>>>>>>>>>> Error in if (random) { : the condition has length > 1
>>>>>>>>>>> traceback()
>>>>>>>>>> 8: model.frame.lme(object, data, xlev = xlev)
>>>>>>>>>> 7: model.frame(object, data, xlev = xlev)
>>>>>>>>>> 6: model.matrix.default(mod, data = structure(list(distance = c(26,
>>>>>>>>>>      25, 29, 31, 21.5, 22.5, 23, 26.5, 23, 22.5, 24, 27.5, 25.5, 27.5,
>>>>>>>>>>      26.5, 27, 20, 23.5, 22.5, 26, 24.5, 25.5, 27, 28.5, 22, 22, 24.5,
>>>>>>>>>>      26.5, 24, 21.5, 24.5, 25.5, 23, 20.5, 31, 26, 27.5, 28, 31, 31.5,
>>>>>>>>>>      23, 23, 23.5, 25, 21.5, 23.5, 24, 28, 17, 24.5, 26, 29.5, 22.5,
>>>>>>>>>>      25.5, 25.5, 26, 23, 24.5, 26, 30, 22, 21.5, 23.5, 25, 21, 20,
>>>>>>>>>>      21.5, 23, 21, 21.5, 24, 25.5, 20.5, 24, 24.5, 26, 23.5, 24.5,
>>>>>>>>>>      25, 26.5, 21.5, 23, 22.5, 23.5, 20, 21, 21, 22.5, 21.5, 22.5,
>>>>>>>>>>      23, 25, 23, 23, 23.5, 24, 20, 21, 22, 21.5, 16.5, 19, 19, 19.5,
>>>>>>>>>>      24.5, 25, 28, 28), age = c(8, 10, 12, 14, 8, 10, 12, 14, 8, 10,
>>>>>>>>>>      12, 14, 8, 10, 12, 14, 8, 10, 12, 14, 8, 10, 12, 14, 8, 10, 12,
>>>>>>>>>>      14, 8, 10, 12, 14, 8, 10, 12, 14, 8, 10, 12, 14, 8, 10, 12, 14,
>>>>>>>>>>      8, 10, 12, 14, 8, 10, 12, 14, 8, 10, 12, 14, 8, 10, 12, 14, 8,
>>>>>>>>>>      10, 12, 14, 8, 10, 12, 14, 8, 10, 12, 14, 8, 10, 12, 14, 8, 10,
>>>>>>>>>>      12, 14, 8, 10, 12, 14, 8, 10, 12, 14, 8, 10, 12, 14, 8, 10, 12,
>>>>>>>>>>      14, 8, 10, 12, 14, 8, 10, 12, 14, 8, 10, 12, 14), Subject =
>>>>>>>>>> structure(c(15L,
>>>>>>>>>>      15L, 15L, 15L, 3L, 3L, 3L, 3L, 7L, 7L, 7L, 7L, 14L, 14L, 14L,
>>>>>>>>>>      14L, 2L, 2L, 2L, 2L, 13L, 13L, 13L, 13L, 5L, 5L, 5L, 5L, 6L,
>>>>>>>>>>      6L, 6L, 6L, 11L, 11L, 11L, 11L, 16L, 16L, 16L, 16L, 4L, 4L, 4L,
>>>>>>>>>>      4L, 8L, 8L, 8L, 8L, 9L, 9L, 9L, 9L, 10L, 10L, 10L, 10L, 12L,
>>>>>>>>>>      12L, 12L, 12L, 1L, 1L, 1L, 1L, 20L, 20L, 20L, 20L, 23L, 23L,
>>>>>>>>>>      23L, 23L, 25L, 25L, 25L, 25L, 26L, 26L, 26L, 26L, 21L, 21L, 21L,
>>>>>>>>>>      21L, 19L, 19L, 19L, 19L, 22L, 22L, 22L, 22L, 24L, 24L, 24L, 24L,
>>>>>>>>>>      18L, 18L, 18L, 18L, 17L, 17L, 17L, 17L, 27L, 27L, 27L, 27L),
>>>>>>>>>> levels = c("M16",
>>>>>>>>>>      "M05", "M02", "M11", "M07", "M08", "M03", "M12", "M13", "M14",
>>>>>>>>>>      "M09", "M15", "M06", "M04", "M01", "M10", "F10", "F09", "F06",
>>>>>>>>>>      "F01", "F05", "F07", "F02", "F08", "F03", "F04", "F11"), class
>>>>>>>>>> = c("ordered",
>>>>>>>>>>      "factor")), Sex = structure(c(1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L,
>>>>>>>>>>      1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L,
>>>>>>>>>>      1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L,
>>>>>>>>>>      1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L,
>>>>>>>>>>      1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L,
>>>>>>>>>>      2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L,
>>>>>>>>>>      2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L,
>>>>>>>>>>      2L, 2L, 2L, 2L), levels = c("Male", "Female"), class =
>>>>>>>>>> "factor")), row.names = c("1",
>>>>>>>>>>      "2", "3", "4", "5", "6", "7", "8", "9", "10", "11", "12", "13",
>>>>>>>>>>      "14", "15", "16", "17", "18", "19", "20", "21", "22", "23", "24",
>>>>>>>>>>      "25", "26", "27", "28", "29", "30", "31", "32", "33", "34", "35",
>>>>>>>>>>      "36", "37", "38", "39", "40", "41", "42", "43", "44", "45", "46",
>>>>>>>>>>      "47", "48", "49", "50", "51", "52", "53", "54", "55", "56", "57",
>>>>>>>>>>      "58", "59", "60", "61", "62", "63", "64", "65", "66", "67", "68",
>>>>>>>>>>      "69", "70", "71", "72", "73", "74", "75", "76", "77", "78", "79",
>>>>>>>>>>      "80", "81", "82", "83", "84", "85", "86", "87", "88", "89", "90",
>>>>>>>>>>      "91", "92", "93", "94", "95", "96", "97", "98", "99", "100",
>>>>>>>>>>      "101", "102", "103", "104", "105", "106", "107", "108"), outer
>>>>>>>>>> = ~Sex, class = c("nfnGroupedData",
>>>>>>>>>>      "nfGroupedData", "groupedData", "data.frame"), formula =
>>>>>>>>>> distance ~
>>>>>>>>>>          age | Subject, labels = list(x = "Age", y = "Distance from
>>>>>>>>>> pituitary to pterygomaxillary fissure"), units = list(
>>>>>>>>>>          x = "(yr)", y = "(mm)"), FUN = structure(function (x)
>>>>>>>>>>      max(x, na.rm = TRUE), source = "function (x) max(x, na.rm =
>>>>>>>>>> TRUE)"), order.groups = TRUE),
>>>>>>>>>>          contrasts.arg = list(Sex = structure(c(0, 1), dim = 2:1,
>>>>>>>>>> dimnames = list(
>>>>>>>>>>              c("Male", "Female"), "Female"))))
>>>>>>>>>> 5: NextMethod(formula(object), data = data, contrasts.arg =
>>>>>>>>>> object$contrasts)
>>>>>>>>>> 4: model.matrix.lme(mod)
>>>>>>>>>> 3: model.matrix(mod)
>>>>>>>>>> 2: Anova_II_lme(mod, vcov., singular.ok = singular.ok)
>>>>>>>>>> 1: car:::Anova.lme(fm2)
>>>>>>>>>>> sessionInfo()
>>>>>>>>>> R version 4.6.1 (2026-06-24 ucrt)
>>>>>>>>>> Platform: x86_64-w64-mingw32/x64
>>>>>>>>>> Running under: Windows 10 x64 (build 19045)
>>>>>>>>>>
>>>>>>>>>> Matrix products: default
>>>>>>>>>>     LAPACK version 3.12.1
>>>>>>>>>>
>>>>>>>>>> locale:
>>>>>>>>>> [1] LC_COLLATE=Chinese (Simplified)_China.utf8
>>>>>>>>>> [2] LC_CTYPE=Chinese (Simplified)_China.utf8
>>>>>>>>>> [3] LC_MONETARY=Chinese (Simplified)_China.utf8
>>>>>>>>>> [4] LC_NUMERIC=C
>>>>>>>>>> [5] LC_TIME=Chinese (Simplified)_China.utf8
>>>>>>>>>>
>>>>>>>>>> time zone: Asia/Shanghai
>>>>>>>>>> tzcode source: internal
>>>>>>>>>>
>>>>>>>>>> attached base packages:
>>>>>>>>>> [1] stats     graphics  grDevices utils     datasets methods base
>>>>>>>>>>
>>>>>>>>>> other attached packages:
>>>>>>>>>> [1] car_3.1-5          carData_3.0-6      nlme_3.1-170
>>>>>>>>>> piecewiseSEM_2.3.1
>>>>>>>>>>
>>>>>>>>>> loaded via a namespace (and not attached):
>>>>>>>>>>    [1] Matrix_1.7-5       jsonlite_2.0.0     compiler_4.6.1 Rcpp_1.1.2
>>>>>>>>>>    [5] DiagrammeR_1.0.12  splines_4.6.1      boot_1.3-32 fastmap_1.2.0
>>>>>>>>>>    [9] lattice_0.22-9     TH.data_1.1-5      Formula_1.2-5
>>>>>>>>>> MuMIn_1.48.19
>>>>>>>>>> [13] rbibutils_2.4.1    htmlwidgets_1.6.4  MASS_7.3-66
>>>>>>>>>> visNetwork_2.1.4
>>>>>>>>>> [17] nloptr_2.2.1       insight_1.5.2      minqa_1.2.8
>>>>>>>>>> RColorBrewer_1.1-3
>>>>>>>>>> [21] rlang_1.3.0        multcomp_1.4-31 performance_0.17.1
>>>>>>>>>> estimability_2.0.0
>>>>>>>>>> [25] cli_3.6.6          magrittr_2.0.5     Rdpack_2.6.6 emmeans_2.0.4
>>>>>>>>>> [29] digest_0.6.39      grid_4.6.1         mvtnorm_1.4-2
>>>>>>>>>> sandwich_3.1-2
>>>>>>>>>> [33] lme4_2.0-6         reformulas_0.4.4   glue_1.8.1 codetools_0.2-20
>>>>>>>>>> [37] zoo_1.8-15         survival_3.8-9     abind_1.4-8 stats4_4.6.1
>>>>>>>>>> [41] tools_4.6.1        htmltools_0.5.9
>>>>>>>>>>> packageVersion("car")
>>>>>>>>>> [1] ‘3.1.5’
>>>>>>>>>>> packageVersion("piecewiseSEM")
>>>>>>>>>> [1] ‘2.3.1’
>>>>>>>>>>> packageVersion("nlme")
>>>>>>>>>> [1] ‘3.1.170’
>>>>>>>>>>
>>>>>>>>>> All the packages are installed from CRAN, and updated to the latest
>>>>>>>>>> version. I also run the codes on FreeBSD 15.1, the same output.
>>>>>>>>>>
>>>>>>>>>> Best,
>>>>>>>>>>
>>>>>>>>>> Jinsong
>>>>>>>>>>
>>>>>>>>>>> On 7/19/2026 2:02 PM, varin sacha wrote:
>>>>>>>>>>> Hi,
>>>>>>>>>>>
>>>>>>>>>>> Could this be a package compatibility bug?
>>>>>>>>>>>
>>>>>>>>>>> Since Anova(fm2), car::Anova(fm2), and even car:::Anova.lme(fm2)
>>>>>>>>>>> all produce the same error, it doesn’t appear to be a simple
>>>>>>>>>>> namespace masking issue.
>>>>>>>>>>>
>>>>>>>>>>> Could you post the output of:
>>>>>>>>>>>
>>>>>>>>>>> traceback()
>>>>>>>>>>> sessionInfo()
>>>>>>>>>>> packageVersion("car")
>>>>>>>>>>> packageVersion("piecewiseSEM")
>>>>>>>>>>> packageVersion("nlme")
>>>>>>>>>>>
>>>>>>>>>>> That should help identify the exact call that’s failing and
>>>>>>>>>>> whether the problem lies in car, piecewiseSEM, or an
>>>>>>>>>>> incompatibility between the two packages.
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>>>>> Le 19 juil. 2026 à 04:58, Jinsong Zhao <[email protected]> a écrit :
>>>>>>>>>>>>> I have tried each solution in a new R session, and neither works.
>>>>>>>>>>>>>
>>>>>>>>>>>>>> library(piecewiseSEM)
>>>>>>>>>>>>> Registered S3 method overwritten by 'lme4':
>>>>>>>>>>>>>     method           from
>>>>>>>>>>>>>     na.action.merMod car
>>>>>>>>>>>>>
>>>>>>>>>>>>>     This is piecewiseSEM version 2.3.0.2.
>>>>>>>>>>>>>
>>>>>>>>>>>>>
>>>>>>>>>>>>>     Questions or bugs can be addressed to <[email protected]>.
>>>>>>>>>>>>>> library(nlme)
>>>>>>>>>>>>>> library(car)
>>>>>>>>>>>>> Loading required package: carData
>>>>>>>>>>>>>> fm2 <- lme(distance ~ age + Sex, data = Orthodont, random = ~ 1)
>>>>>>>>>>>>>> Anova(fm2)
>>>>>>>>>>>>> Error in if (random) { : the condition has length > 1
>>>>>>>>>>>>>> car::Anova(fm2)
>>>>>>>>>>>>> Error in if (random) { : the condition has length > 1
>>>>>>>>>>>>>> car:::Anova.lme(fm2)
>>>>>>>>>>>>> Error in if (random) { : the condition has length > 1
>>>>>>>>>
>>>>>>>>> ______________________________________________
>>>>>>>>> [email protected] mailing list -- To UNSUBSCRIBE and more, see
>>>>>>>>> https://stat.ethz.ch/mailman/listinfo/r-help
>>>>>>>>> PLEASE do read the posting guide
>>>>>>>>> https://www.R-project.org/posting-guide.html
>>>>>>>>> and provide commented, minimal, self-contained, reproducible code.
>>>>>>>>
>>>>>>>> ______________________________________________
>>>>>>>> [email protected] mailing list -- To UNSUBSCRIBE and more, see
>>>>>>>> https://stat.ethz.ch/mailman/listinfo/r-help
>>>>>>>> PLEASE do read the posting guide
>>>>>>>> https://www.R-project.org/posting-guide.html
>>>>>>>> and provide commented, minimal, self-contained, reproducible code.
>>>>>> ______________________________________________
>>>>>> [email protected] mailing list -- To UNSUBSCRIBE and more, see
>>>>>> https://stat.ethz.ch/mailman/listinfo/r-help
>>>>>> PLEASE do read the posting guide https://www.R-project.org/posting-guide.html
>>>>>> and provide commented, minimal, self-contained, reproducible code.
>>>>>
>>>>
>>>> ______________________________________________
>>>> [email protected] mailing list -- To UNSUBSCRIBE and more, see
>>>> https://stat.ethz.ch/mailman/listinfo/r-help
>>>> PLEASE do read the posting guide https://www.R-project.org/posting-guide.html
>>>> and provide commented, minimal, self-contained, reproducible code.
>>>
>>
>