[R-sig-ME] raters
Iasonas Lamprianou
lamprianou at yahoo.com
Fri Jan 29 07:09:12 CET 2010
Good morning colleagues,
I have a dataset where 100 raters (teachers who mark student responses to examination questions) participated to a number of examinations. Not everybofy participated to each examination. For example, one rater may have participated to only 2 examinations, and another rater may have participated to 12 examinations. The examinations do not happen at regular intervals i.e. sometimes a month may pass between two examinations, sometimes the distance between two examinations may be nine months. So, we have a situation where diffrent people participated to a different number of examinations and the distance between the examinations is not constant. Every time a rater participates to one examination, he is awarded some statistical indications of his/her rating profile: accuracy, consistency, restriction of range (he only uses part of the rating scale) etc.
I need to investigate (a) if the are raters who follow specific patterns e.g. the more the exminations they participate, the more consistent they become, or the more accurate they become. (b)I want to check if there is statistically significant variability between the raters' statistics. In every case, I want to use the time that elaplses between exams as a covariate.
How can I do the above two points using lme?
Dr. Iasonas Lamprianou
Assistant Professor (Educational Research and Evaluation)
Department of Education Sciences
European University-Cyprus
P.O. Box 22006
1516 Nicosia
Cyprus
Tel.: +357-22-713178
Fax: +357-22-590539
Honorary Research Fellow
Department of Education
The University of Manchester
Oxford Road, Manchester M13 9PL, UK
Tel. 0044 161 275 3485
iasonas.lamprianou at manchester.ac.uk
--- On Thu, 28/1/10, r-sig-mixed-models-request at r-project.org <r-sig-mixed-models-request at r-project.org> wrote:
> From: r-sig-mixed-models-request at r-project.org <r-sig-mixed-models-request at r-project.org>
> Subject: R-sig-mixed-models Digest, Vol 37, Issue 20
> To: r-sig-mixed-models at r-project.org
> Date: Thursday, 28 January, 2010, 21:26
> Send R-sig-mixed-models mailing list
> submissions to
> r-sig-mixed-models at r-project.org
>
> To subscribe or unsubscribe via the World Wide Web, visit
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
> or, via email, send a message with subject or body 'help'
> to
> r-sig-mixed-models-request at r-project.org
>
> You can reach the person managing the list at
> r-sig-mixed-models-owner at r-project.org
>
> When replying, please edit your Subject line so it is more
> specific
> than "Re: Contents of R-sig-mixed-models digest..."
>
>
> Today's Topics:
>
> 1. Re: Trend in total number of animals
> (Jarrod Hadfield)
> 2. Re: Time series LME with nested fixed
> effects (Austen Thomas)
> 3. Re: Science Fair data (Douglas Bates)
> 4. Re: Science Fair data (Doug Adams)
> 5. Re: estimates of between sex
> correlations (Jarrod Hadfield)
> 6. Re: Trend in total number of animals
> (David Duffy)
>
>
> ----------------------------------------------------------------------
>
> Message: 1
> Date: Thu, 28 Jan 2010 12:57:04 +0000
> From: Jarrod Hadfield <j.hadfield at ed.ac.uk>
> To: "ONKELINX, Thierry" <Thierry.ONKELINX at inbo.be>
> Cc: r-sig-mixed-models at r-project.org
> Subject: Re: [R-sig-ME] Trend in total number of animals
> Message-ID: <20100128125704.hn6nx4hrks4osw48 at www.staffmail.ed.ac.uk>
> Content-Type: text/plain;
> charset=ISO-8859-1; DelSp="Yes";
> format="flowed"
>
> Dear Thierry,
>
> I THINK the fixed effect slope should be what you're after
> if you want
> to predict the change in log numbers, but simply
> exponentiating the
> prediction will not give you a true measure of the
> arithmetic increase.
>
> The arithmetic prediction for years 1:10 (for example) when
> the slope
> variance for the year|room term is zero would be:
>
> exp(b_year*1:10+0.5*(v1+v2))
>
> where b_year is your slope estimate, and v1 is the year
> intercept
> variance and v2 is the room intercept variance.
>
> When slope variance exists this becomes more difficult,
> because it
> implies the variance v2 changes as a function of year. In
> this case:
>
> v2=diag(Z%*%V2%*%t(Z))
>
> where
>
> Z<-cbind(rep(1,10), 1:10)
>
> and V2 is the covariance matrix of the room
> intercept-slopes.
>
> Or if you like
>
> v2 = V2[1,1]+(1:10)*V2[1,2]*2+(1:10^2)*V2[2,2]
>
> Another difficulty is the possibility that your missing
> data are not
> "completely missing at random". By default lmer just seems
> to omit
> missing data rather than dealing with it properly, but
> perhaps there
> is an argument that can be passed to na.omit which
> suppresses this? If
> so, then the less strict assumption of "missing at random"
> can be
> made. In this latter case the missing data only have to be
> random
> conditional on the observed data - for example, if there
> were no bats
> in room A in year 1 which made the field workers less
> inclined to
> visit room A in year 2 based on their knowledge of the 1'st
> year's
> count.
>
> Cheers,
>
> Jarrod
>
> Quoting "ONKELINX, Thierry" <Thierry.ONKELINX at inbo.be>:
>
> > Dear all,
> >
> > We are modelling the total numbers of hibarnating bats
> in a fortress. We
> > have data of the number of bats per room spanning ten
> years. The main
> > problem is that not all rooms were visited each year.
> The fieldworkers
> > did not known or find all rooms and some rooms were
> not allways
> > accessible.
> >
> > Some of the rooms were not counted in the early years
> and they contain a
> > rather high number of bats in the more recent years.
> So a glm on the
> > total observed number would be very biased. Therefore
> we would use a
> > mixed model on the numbers of bats per room. The model
> looks like:
> > glmer(Number ~ Year + (1|Year) + (Year|Room), family =
> poisson). Year is
> > the long-term trend. (1|Year) allows for year-to-year
> variability (due
> > to weatherconditions) and (Year|Room) allows for a
> random intercept and
> > slope per room.
> >
> > Our main question about this model is the
> interpretation of the
> > long-term trend (fixed effect of Year). Given the
> model specification it
> > is the trend in an 'average' room from the population
> of rooms. Can we
> > assume that this trend equals the trend in the total
> number of bats in
> > the fortress. That would be the trend in to total
> observed numbers if we
> > could have investigated every room in every year.
> > Or is it better to use the model to simulate the total
> number of bats
> > and then model this simulated totals using a simple
> glm? Repeating the
> > simulations a large number of times would yield an
> average and
> > confidence intervals for the trend.
> >
> > Best regards,
> >
> > Thierry
> >
> >
> ------------------------------------------------------------------------
> > ----
> > ir. Thierry Onkelinx
> > Instituut voor natuur- en bosonderzoek
> > team Biometrie & Kwaliteitszorg
> > Gaverstraat 4
> > 9500 Geraardsbergen
> > Belgium
> >
> > Research Institute for Nature and Forest
> > team Biometrics & Quality Assurance
> > Gaverstraat 4
> > 9500 Geraardsbergen
> > Belgium
> >
> > tel. + 32 54/436 185
> > Thierry.Onkelinx at inbo.be
> > 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
> >
> >
> >
> >
> > Druk dit bericht a.u.b. niet onnodig af.
> > Please do not print this message unnecessarily.
> >
> > Dit bericht en eventuele bijlagen geven enkel de visie
> van de schrijver weer
> > en binden het INBO onder geen enkel beding, zolang dit
> bericht niet
> > bevestigd is
> > door een geldig ondertekend document. The views
> expressed in this message
> > and any annex are purely those of the writer and may
> not be regarded
> > as stating
> > an official position of INBO, as long as the message
> is not
> > confirmed by a duly
> > signed document.
> >
> > [[alternative HTML version
> deleted]]
> >
> > _______________________________________________
> > R-sig-mixed-models at r-project.org
> mailing list
> > https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
> >
> >
>
>
>
> --
> The University of Edinburgh is a charitable body,
> registered in
> Scotland, with registration number SC005336.
>
>
>
> ------------------------------
>
> Message: 2
> Date: Wed, 27 Jan 2010 17:27:37 -0800
> From: Austen Thomas <austen.thomas at gmail.com>
> To: Andrew Dolman <andydolman at gmail.com>
> Cc: R-SIG-Mixed-Models at r-project.org
> Subject: Re: [R-sig-ME] Time series LME with nested fixed
> effects
> Message-ID:
> <9d4eee51001271727i4edcc565l1a5d7999ade2fc2b at mail.gmail.com>
> Content-Type: text/plain
>
> Hi Andrew,
>
> Thank you so much for taking the time to look this over for
> us. Your
> assistance has been exceedingly helpful.
>
> -Austen
>
> On Wed, Jan 27, 2010 at 1:23 PM, Andrew Dolman <andydolman at gmail.com>
> wrote:
>
> > Hello Austin,
> >
> > I stand to be corrected but as no-one else has offered
> anything yet I'll
> > make a suggestion or two.
> >
> >
> > 1.
> > I don't think you want your random effects nested; you
> want them crossed.
> > This you can only do with lmer, not nlme.
> >
> > So something like:
> >
> > lmer(MDB~Season*Period+(1|Seal)
> +(1|Season)+(1|Period), data=d,
> > REML=F)->fit10
> >
> > This is because the seasons and periods have meaning
> independently. i.e.
> > saying "spawn" means the same thing for each seal, and
> each period. Or in
> > other words, you might expect "spawn" data points to
> share similarities with
> > each other that they do not share with non-spawn,
> regardless of which time
> > period or seal.
> >
> > 2. But!
> >
> > You only have 2 categories for season. This is not
> enough to treat it as
> > random. 4 periods is not really enough either. 5 seals
> is possibly just
> > enough. So I would just have seal as the only random
> effect. Just treat
> > season and period as fixed.
> >
> > lmer(MDB~Season*Period+(1|Seal), data=d,
> REML=F)->fit10
> >
> >
> >
> > andydolman at gmail.com
> >
> >
> > 2010/1/27 Austen Thomas <austen.thomas at gmail.com>
> >
> >> We are using transmitting time/depth
> recorders to measure the diving
> >> behavior of 5 harbor seals in Puget Sound.
> For each seal, we have two
> >> seasons, while herring are spawning and after they
> spawn. Nested in each
> >> season, we have four times of day (labeled
> period). We are interested in
> >> whether there is an effect of “season” (spawn
> / non-spawn), “period”
> >> (morning, day, evening, night) on the “Modal
> Depth Bin” of harbor seals,
> >> or
> >> an interaction between these two factors.
> >>
> >> Below is the code that we think is correct.
> Does anyone think the model
> >> is
> >> specified incorrectly, and if so what do they
> recommend (and why)?
> >> Thanks very much for your time.
> >>
> >> -Austen Thomas
> >>
> >> Biology Department
> >> Western Washington University
> >> Biology Building BI 315
> >> Mail Stop 9160
> >> Bellingham, WA 98225-9160
> >>
> >>
> >>
> >> For lme4 package
> >> summary(d)
> >> lmer(MDB~Season*Period+(1|Seal/Season/Period),
> data=d, REML=F)->fit10
> >> summary(fit10)
> >> lmer(MDB~Season+Period+(1|Seal/Season/Period),
> data=d, REML=F)->fit11
> >> lmer(MDB~Season+(1|Seal/Season/Period), data=d,
> REML=F)->fit12
> >> lmer(MDB~1+(1|Seal/Season/Period), data=d,
> REML=F)->fit13
> >> anova(fit10, fit11, fit12, fit13)
> >>
> >> for the nlme package
> >> lme(MDB~Season*Period,
> random=~1|Seal/Season/Period, method="ML",
> >> data=d)->fit20
> >> summary(fit20)
> >> lme(MDB~Season+Period,
> random=~1|Seal/Season/Period, method="ML",
> >> data=d)->fit21
> >> lme(MDB~Season, random=~1|Seal/Season/Period,
> method="ML", data=d)->fit22
> >> lme(MDB~1, random=~1|Seal/Season/Period,
> method="ML", data=d)->fit23
> >> anova(fit20, fit21, fit22, fit23)
> >>
> >> [[alternative HTML
> version deleted]]
> >>
> >>
> >> _______________________________________________
> >> R-sig-mixed-models at r-project.org
> mailing list
> >> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
> >>
> >>
> >
>
>
> --
> Biology Department
> Western Washington University
> Biology Building BI 315
> Mail Stop 9160
> Bellingham, WA 98225-9160
>
> [[alternative HTML version deleted]]
>
>
>
> ------------------------------
>
> Message: 3
> Date: Thu, 28 Jan 2010 09:35:02 -0600
> From: Douglas Bates <bates at stat.wisc.edu>
> To: Doug Adams <fog0 at gmx.com>
> Cc: R sig-mixed-models <r-sig-mixed-models at r-project.org>
> Subject: Re: [R-sig-ME] Science Fair data
> Message-ID:
> <40e66e0b1001280735r5c90ad8arf2c73a101aadc8b2 at mail.gmail.com>
> Content-Type: text/plain; charset=ISO-8859-1
>
> I'm writing a book about lme4 and lmer. When it gets
> to the point
> where others can read it without too much frowning and
> scratching of
> heads, I will make chapter drafts available.
>
> For the time being you may find the slides from a short
> course given
> last summer informative. They are at
> http://lme4.R-forge.R-project.org/slides/2009-07-21-Seewiesen/
>
> On Wed, Jan 27, 2010 at 2:53 PM, Doug Adams <fog0 at gmx.com>
> wrote:
> > Hi, and thanks again. ?That makes sense with the 3
> levels of division:
> >
> > (Intercept) divisionJunior divisionSenior
> > ? ? 80.306526 ? ? ?-3.252372 ? ? ?-2.694055
> >
> > ...so that the Junior and Senior levels are both
> slightly lower than
> > the Elementary level.
> >
> > I'd love to really understand the summary of lmer and
> what ranef,
> > fixef, coef and fitted are extracting - so that
> probably means I don't
> > understand the basics and nomenclature of HLMing as I
> thought I might
> > have. ?I took a 1-week class on HLM, and I have the
> book you (Douglas
> > Bates) wrote... ?Maybe I just need to study up on
> things a little
> > better!
> >
> > Anyway, I appreciate your help very much ? : )
> >
> > Doug
> >
> >
> >
> > On Wed, Jan 27, 2010 at 8:07 AM, Douglas Bates <bates at stat.wisc.edu>
> wrote:
> >> On Tue, Jan 26, 2010 at 10:10 PM, Doug Adams
> <fog0 at gmx.com>
> wrote:
> >>> I appreciate that, both of you (& that's
> ok for the mistake Christopher) ? :)
> >>
> >>> So fixed factors as simply listed by
> themselves (no 1| notation) and
> >>> random effects are listed with appropriate
> nesting... ?I do want to
> >>> consider schools as random effects; that will
> give me the information
> >>> I'd like to have about the variability (and
> reliability too?) of the
> >>> schools as they fit into the big picture.
> >>
> >>> When I use fixef & ranef to extract
> estimates for division and schools
> >>> (& maybe districts eventually too now), am
> I right in thinking that
> >>> the 3 numbers given for each level of division
> (fixef) are the
> >>> intercepts for each level -- as if there were
> individual OLS
> >>> regressions performed for each?
> >>
> >> Not quite. ?They should be labeled "(Intercept)"
> and something like
> >> division2 and division 3. ?(By the way, it helps
> if you quote the
> >> output when you want to discussion what particular
> values mean.) ?The
> >> (Intercept) coefficient represents the prediction
> at the first level
> >> of the division factor. ?The next two coefficients
> are the change from
> >> the first to the second level and from the first
> to the third level.
> >>
> >>> And are the random effects for the
> >>> schools (ranef) are the slopes associated with
> those regression lines?
> >
> > _______________________________________________
> > R-sig-mixed-models at r-project.org
> mailing list
> > https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
> >
>
>
>
> ------------------------------
>
> Message: 4
> Date: Thu, 28 Jan 2010 09:48:06 -0700
> From: Doug Adams <fog0 at gmx.com>
> To: Douglas Bates <bates at stat.wisc.edu>,
> R sig-mixed-models
> <r-sig-mixed-models at r-project.org>
> Subject: Re: [R-sig-ME] Science Fair data
> Message-ID:
> <74cfd9161001280848u7192230ey1cc450405cf3c579 at mail.gmail.com>
> Content-Type: text/plain; charset=ISO-8859-1
>
> That's great! Thanks!
>
>
>
> On Thu, Jan 28, 2010 at 8:35 AM, Douglas Bates <bates at stat.wisc.edu>
> wrote:
> > I'm writing a book about lme4 and lmer. ?When it gets
> to the point
> > where others can read it without too much frowning and
> scratching of
> > heads, I will make chapter drafts available.
> >
> > For the time being you may find the slides from a
> short course given
> > last summer informative. ?They are at
> > http://lme4.R-forge.R-project.org/slides/2009-07-21-Seewiesen/
> >
> > On Wed, Jan 27, 2010 at 2:53 PM, Doug Adams <fog0 at gmx.com>
> wrote:
> >> Hi, and thanks again. ?That makes sense with the 3
> levels of division:
> >>
> >> (Intercept) divisionJunior divisionSenior
> >> ? ? 80.306526 ? ? ?-3.252372 ? ? ?-2.694055
> >>
> >> ...so that the Junior and Senior levels are both
> slightly lower than
> >> the Elementary level.
> >>
> >> I'd love to really understand the summary of lmer
> and what ranef,
> >> fixef, coef and fitted are extracting - so that
> probably means I don't
> >> understand the basics and nomenclature of HLMing
> as I thought I might
> >> have. ?I took a 1-week class on HLM, and I have
> the book you (Douglas
> >> Bates) wrote... ?Maybe I just need to study up on
> things a little
> >> better!
> >>
> >> Anyway, I appreciate your help very much ? : )
> >>
> >> Doug
>
>
>
> ------------------------------
>
> Message: 5
> Date: Thu, 28 Jan 2010 17:23:44 +0000
> From: Jarrod Hadfield <j.hadfield at ed.ac.uk>
> To: r-sig-mixed-models at r-project.org
> Subject: Re: [R-sig-ME] estimates of between sex
> correlations
> Message-ID: <20100128172344.1nykjdyvwg0sg4wg at www.staffmail.ed.ac.uk>
> Content-Type: text/plain;
> charset=ISO-8859-1; DelSp="Yes";
> format="flowed"
>
> Hi Celine,
>
> Unfortunately I don't think the model you want to fit can
> be fitted in
> MCMCglmm. Currently only (conditional) diagonal and
> (conditional)
> block residual covariance matrices are implemented, and
> what you want
> is block diagonal. ASReml is probably you best way forward
> at the
> moment. If you are prepared to assume that the residual
> covariances
> for the 3 traits are equivalent in the two sexes then you
> can use
> rcov=~us(trait):sex:units.
>
> With respect to the permanent environment effects
> you can fit a
> block diagonal using
>
> us(at.level(sex,1):trait):id+us(at.level(sex,2):trait):id
>
> where id are the individual identifiers.
>
> I may allow this type of syntax in the residual structure
> in the near future.
>
> Cheers,
>
> Jarrod
>
>
> Quoting Celine Teplitsky <teplitsky at mnhn.fr>:
>
> > Dear all,
> >
> > I would like to fit a model to estimate between sex
> genetic
> > correlations for 3 traits (3 same traits expressed
> both in males and
> > females) using MCMCglmm. My problem is I don't
> understand how I can
> > manage to fix the covariances between male and female
> traits to 0 for
> > residual variance and permanent environment. Could
> anyone help me with
> > this?
> >
> > Many thanks in advance
> >
> > All the best
> >
> > Celine
> >
> > --
> >
> > Celine Teplitsky
> > D?partement Ecologie et Gestion de la Biodiversit? UMR
> 7204
> > Unit? Conservation des Esp?ces, Restauration et Suivi
> des Populations
> > Case Postale 51
> > 55 rue Buffon 75005 Paris
> >
> > Webpage :http://www2.mnhn.fr/cersp/spip.php?rubrique96
> > Fax : (33-1)-4079-3835
> > Phone: (33-1)-4079-3443
> >
> > _______________________________________________
> > R-sig-mixed-models at r-project.org
> mailing list
> > https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>
>
>
> --
> The University of Edinburgh is a charitable body,
> registered in
> Scotland, with registration number SC005336.
>
>
>
> ------------------------------
>
> Message: 6
> Date: Fri, 29 Jan 2010 07:25:27 +1000 (EST)
> From: David Duffy <David.Duffy at qimr.edu.au>
> To: "ONKELINX, Thierry" <Thierry.ONKELINX at inbo.be>
> Cc: r-sig-mixed-models at r-project.org
> Subject: Re: [R-sig-ME] Trend in total number of animals
> Message-ID: <Pine.LNX.4.64.1001290717440.16167 at orpheus.qimr.edu.au>
> Content-Type: TEXT/PLAIN; charset=US-ASCII; format=flowed
>
> On Thu, 28 Jan 2010, ONKELINX, Thierry wrote:
>
> > We are modelling the total numbers of hibernating bats
> in a fortress. We
> > have data of the number of bats per room spanning ten
> years. The main
> > problem is that not all rooms were visited each year.
> The fieldworkers
> > did not known or find all rooms and some rooms were
> not allways
> > accessible.
> >
> > Some of the rooms were not counted in the early years
> and they contain a
> > rather high number of bats in the more recent years.
>
> I would try (fixed-effects) log-linear models for
> incomplete tables as
> well, to deal with the sampling (I have written an R
> package "gllm", that
> may be suitable). Then you could compare the
> estimates of total counts to
> the GLMM results. If you are going to simulate, you
> might might as well
> go to BUGS.
>
> Cheers, David.
>
> --
> | David Duffy (MBBS PhD)
>
>
> ,-_|\
> | email: davidD at qimr.edu.au
> ph: INT+61+7+3362-0217 fax: -0101 /
> *
> | Epidemiology Unit, Queensland Institute of Medical
> Research \_,-._/
> | 300 Herston Rd, Brisbane, Queensland 4029,
> Australia GPG 4D0B994A v
>
>
>
> ------------------------------
>
> _______________________________________________
> R-sig-mixed-models mailing list
> R-sig-mixed-models at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>
>
> End of R-sig-mixed-models Digest, Vol 37, Issue 20
> **************************************************
>
More information about the R-sig-mixed-models
mailing list