[R-sig-ME] Course: Introduction to Linear Mixed Effects Models and GLMM with R-INLA

Fri Jun 12 16:07:03 CEST 2020

The flyer mentions nested _data_. IMHO that refers to multiple
_observations_ nested in a group (the structure within a group). The
example is correct from that perspective.

Mollie seems to refer to nested and crossed _random effects_, which is a
statement about the structure among the groups. Note that the second page
of the flyer does mention nested and crossed _random effects_. So the
distinction between nested and crossed random effects is handled in the
course.

The discussion becomes somewhat irrelevant if the grouping variables are
defined properly. Give each student, class, school, ... an ID that is
unique within the dataset. Then you only need a single variable to define a
specific group. Specify a random effect for every relevant grouping
variable as if they are crossed. The penalisation term in the likelihood
will take care of any implicit nesting.

Just my €0.02

Best regards,

ir. Thierry Onkelinx
Statisticus / Statistician

Vlaamse Overheid / Government of Flanders
INSTITUUT VOOR NATUUR- EN BOSONDERZOEK / RESEARCH INSTITUTE FOR NATURE AND
FOREST
Team Biometrie & Kwaliteitszorg / Team Biometrics & Quality Assurance
thierry.onkelinx using inbo.be
Havenlaan 88 bus 73, 1000 Brussel
www.inbo.be

///////////////////////////////////////////////////////////////////////////////////////////
To call in the statistician after the experiment is done may be no more
than asking him to perform a post-mortem examination: he may be able to say
what the experiment died of. ~ Sir Ronald Aylmer Fisher
The plural of anecdote is not data. ~ Roger Brinner
The combination of some data and an aching desire for an answer does not
ensure that a reasonable answer can be extracted from a given body of data.
~ John Tukey
///////////////////////////////////////////////////////////////////////////////////////////

<https://www.inbo.be>

Op vr 12 jun. 2020 om 14:50 schreef Highland Statistics Ltd <
highstat using highstat.com>:

> On 12/06/2020 10:34, Mollie Brooks wrote:
> > Yes, I haven’t seen any of the course material other than the flyer. I’m
> just surprised by how many people think they know what "nesting" means and
> are wrong (e.g. professors who have been using the methods for years and
> PhD students who just learned the methods).
>
> Multiple monkeys in an enclosure, with multiple enclosures in a zoo,
> with multiple zoos....that would be nested...wouldn't it? And if some
> monkeys decide to take the bus and move to another zoo because the
> bananas are better, then it is crossed (like your patients going to two
> hospitals).
>
> If don't think that a flyer is the right place to confuse people with
> "what-if" scenarios. Otherwise, I might as well attached a book.
>
>
> > I’m tired of having to correct them because it makes my work harder
> >   and I’m not really paid to do that work, so it would be helpful to me
> if instructors who are paid to do it could be more pedantic.
>
> The art (or skill) of teaching/communicating statistics to a
> non-statistical audience requires switching regularly between informal
> language (which may not be 100% statistically correct) and formal
> statistical language. If everything is pedantic, then only a few will
> enjoy statistics and the majority will hate it (speaking from 20 years
> experience of teaching it).
>
>
> > I realize that this is off topic for the list,
> Is it? If instructors teach it properly, then you get fewer questions on
> this mailing list.
>
> Alain
>
> >   so I’ll drop it.
>
>
>
>
> >
> >
> >> On 11Jun 2020, at 17:04, John Poe <jdpoe223 using gmail.com> wrote:
> >>
> >> I think it's a little bit pedantic to criticize the flyer here since I
> >> assume that the distinction between nested and cross classified grouping
> >> structures is made pretty clear in the class itself.
> >>
> >> I haven't seen the course material obviously so that's an assumption on
> my
> >> part given how I teach it. I do spend a lot of time on it whenever I
> teach
> >> mixed effects models because it is an important point of confusion for
> >> people. I tend to use either "clustered" or the term "grouping
> structure"
> >> as generic and differentiate between nested, crossed, and multiple
> >> membership personally. So i understand the point Mollie and Henrik are
> >> making but it might be unfair to expect that level of nuance in a flyer?
> >>
> >>
> >>
> >>
> >>
> >>
> >>
> >> On Thu, Jun 11, 2020, 10:51 AM Henrik Singmann <singmann using gmail.com>
> wrote:
> >>
> >>> But isn't that exactly Mollie's point? You write "Nested data means
> >>> multiple observations from the same [unit of observation]". And then
> she
> >>> gave an example where you can have multiple observations from the same
> unit
> >>> of observation without the data being nested.
> >>>
> >>> I also completely agree with her criticism that this terminology is
> >>> critical to get right. When I teach mixed models one of the things that
> >>> always comes up is that people misunderstand the concept of nested
> factors:
> >>> A factor A is nested in another factor B if certain levels of A only
> appear
> >>> with certain levels of B and not with all levels of B (the latter
> would be
> >>> called crossed). In other words, whether or not we have repeated
> measures
> >>> or multiple observations is unrelated to whether or not there exists
> >>> nesting in the data.
> >>>
> >>> Maybe it would make more sense to use "clustered" in that context
> instead
> >>> of "nested".
> >>>
> >>>
> >>> Am Do., 11. Juni 2020 um 16:20 Uhr schrieb Highland Statistics Ltd <
> >>> highstat using highstat.com>:
> >>>
> >>>> On 11/06/2020 14:58, Mollie Brooks wrote:
> >>>>> The flyer says "Nested data means multiple observations from the same
> >>>>> animal, site, area, nest, patient, hospital, vessel,
> >>>>> lake, hive, transect, etc.", but this doesn’t agree with my
> >>>>> understanding
> >>>>> (
> >>>
> https://bbolker.github.io/mixedmodels-misc/glmmFAQ.html#nested-or-crossed
> ).
> >>> What
> >>>>> if animals move from one site to another, or patients visit multiple
> >>>>> hospitals.
> >>>> Then it is not nested anymore.
> >>>>
> >>>>
> >>>>> I encounter a lot of scientists who have a misconception of the
> >>>>> meaning of nested data, so it would be good to be careful when
> >>>>> teaching the terminology. Does R-INLA require random effects to be
> >>>>> nested?
> >>>> No. They can even be spatially correlated....or temporally correlated.
> >>>>
> >>>> Alain
> >>>>
> >>>>> Kind regards,
> >>>>> Mollie
> >>>>>
> >>>>>> On 11Jun 2020, at 14:40, Highland Statistics Ltd
> >>>>>> <highstat using highstat.com <mailto:highstat using highstat.com>> wrote:
> >>>>>>
> >>>>>> We would like to announce the following online statistics course:
> >>>>>>
> >>>>>> Introduction to Linear Mixed Effects Models and GLMM with R-INLA
> >>>>>>
> >>>>>>
> >>>>>> This is an on-demand course with around 35-40 videos (each is 15-60
> >>>>>> minutes) with live (optional) Zoom summary sessions scheduled in 2
> >>>>>> different time zones:
> >>>>>>
> >>>>>> * Time zone 1: 09.00-11.00 British Summer Time.
> >>>>>> * Time zone 2: 19.00-21.00 British Summer Time.
> >>>>>>
> >>>>>> The course represents around 40 hours of work.
> >>>>>>
> >>>>>> The course fee includes an (optional) 1-hour face-to-face video chat
> >>>>>> with one or both instructors (you can discuss your own data).
> >>>>>>
> >>>>>> Starting date: 22 June
> >>>>>>
> >>>>>> Flyer:
> >>>>>>
> >>>>
> http://highstat.com/Courses/Flyers/2020/Flyer2020_06_GLMMINLA_Online.pdf
> >>>>>> Website: http://highstat.com/index.php/courses-upcoming
> >>>>>>
> >>>>>>
> >>>>>> Kind regards,
> >>>>>>
> >>>>>>
> >>>>>> Alain Zuur
> >>>>>>
> >>>>>> --
> >>>>>>
> >>>>>> Dr. Alain F. Zuur
> >>>>>> Highland Statistics Ltd.
> >>>>>> 9 St Clair Wynd
> >>>>>> AB41 6DZ Newburgh, UK
> >>>>>> Email:highstat using highstat.com
> >>>>>> URL:www.highstat.com
> >>>>>>
> >>>>>>
> >>>>>> [[alternative HTML version deleted]]
> >>>>>>
> >>>>>> _______________________________________________
> >>>>>> R-sig-mixed-models using r-project.org mailing list
> >>>>>> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
> >>>> --
> >>>>
> >>>> Dr. Alain F. Zuur
> >>>> Highland Statistics Ltd.
> >>>> 9 St Clair Wynd
> >>>> AB41 6DZ Newburgh, UK
> >>>> Email: highstat using highstat.com
> >>>> URL:   www.highstat.com
> >>>>
> >>>>
> >>>>         [[alternative HTML version deleted]]
> >>>>
> >>>> _______________________________________________
> >>>> R-sig-mixed-models using r-project.org mailing list
> >>>> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
> >>>>
> >>>
> >>> --
> >>> Dr. Henrik Singmann
> >>> Assistant Professor, Department of Psychology
> >>> University of Warwick, UK
> >>> http://singmann.org
> >>>
> >>>         [[alternative HTML version deleted]]
> >>>
> >>> _______________________________________________
> >>> R-sig-mixed-models using r-project.org mailing list
> >>> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
> >>>
> >>      [[alternative HTML version deleted]]
> >>
> >> _______________________________________________
> >> R-sig-mixed-models using r-project.org mailing list
> >> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>
> --
>
> Dr. Alain F. Zuur
> Highland Statistics Ltd.
> 9 St Clair Wynd
> AB41 6DZ Newburgh, UK
> Email: highstat using highstat.com
> URL:   www.highstat.com
>
> _______________________________________________
> R-sig-mixed-models using r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>

	[[alternative HTML version deleted]]