[R] Stats help for dissertation project

[Richard O'Keefe]

In fairness to Raija Hallam, I've met masters and doctoral students whose
supervisors hadn't a clue.  (Heck, I once worked at a University where the
staff evaluation process involved taking the means of 5-point ordinal
variables...)  Two of these cases stick in my mind: one where the student
had more observed variables than cases, and one where the student was
controlling for the effect of the order in which the treatments were
applied by applying the treatments in the same order to every subject.

To Raija Hallam, do web searches for "linear regression in R" and "logistic
regression in R".  Look at the R web site r-project.org, specifically at
https://www.r-project.org/doc/bib/R-books.html
Print off a copy of that list of books, go to your University library, and
borrow a couple, maybe an introductory book and "An R Companion to Applied
Regression".

It is incredibly important that you have a very clear idea of your research
question, your experiment structure, and the nature of your data.
Here's one point which even your supervisor is very likely to miss:
the micronutrient analysis *MAY* best be regarded as COMPOSITIONAL DATA.
R has a package called "compositions" that may be helpful.  The description
says "Regression, classification, contour plots, hypothesis testing and
fitting of distributions for compositional data are some of the functions
included. The standard text-book for such data is John Aitchison's (1986)
``The statistical analysis of compositional data''. Chapman & Hall."

If I am right about this, then analysing your data using "standard"
techniques may be *seriously* misleading.  Ask your supervisor to
find a statistician who can talk to you about your problem; at my former
university we'd probably suggest inviting co-supervision to make sure
your statistical analyses are credible.


On Sat, 19 Sep 2020 at 22:30, Michael Dewey <lists using dewey.myzen.co.uk> wrote:

> Dear Raija
>
> This list is primarily for R programming questions so most of your post
> is off-topic here. If you are registered for a degree presumably someone
> is paying a fee to your institution and someone there is being paid to
> supervise your project so I would have thought they would be the first
> port of call. If they fail to meet their obligations then there is a
> site Cross Validated where you may have better luck.
>
> Michael
>
> On 18/09/2020 18:19, Raija Hallam wrote:
> > Hello,
> >
> > I am a conservation Masters student who is new to R and in need of some
> > confirmation of my methods/ stats help!
> >
> > My dissertation project is looking at the micronutrient intake (iron,
> zinc,
> > calcium, magnesium and folic acid/folate) of 18 female monkeys, 14 of
> which
> > are reproductively 'successful' (their infant survived past 1 year) and 4
> > of which are reproductively 'unsuccessful' (their infant did not survive
> to
> > 1 year). The females go through a number of stages throughout their
> > pregnancy, and I would like to focus on 2 of these stages, early
> gestation
> > and early lactation, as these appear in the literature to be important
> > stages in terms of nutrition. Each stage is broken down into 4 week
> periods
> > so I have G1, G2 and G3 as early gestation and L1, L2, L3 and L4 as early
> > lactation. These could also be combined into just 2 reproductive stages;
> > early gestation (EG) and early lactation (EL), to make the model a bit
> > simpler.
> >
> >
> > *I first would like to investigate how micronutrient intake is affected
> by
> > the reproductive stage of females.*
> >
> > To investigate this I am thinking of doing a multivariate multiple
> > regression general linear model, controlling for Female ID:
> > ironintake ~ repphase + (1/FemaleID)
> > zincintake ~ repphase + (1/FemaleID)
> > etc.
> >
> >
> > *I would also like to investigate how the micronutrients they intake
> affect
> > reproductive 'successfulness'.*
> >
> > To investigate this I am thinking of doing a binomial logistic regression
> > generalised linear model, again controlling for Female ID:
> > Repsuccess ~ ironintake + zincintake + calciumintake etc... +(1/FemaleID)
> >
> >
> > As I'm new to R and a bit rusty with my stats knowledge I would be very
> > grateful for any comments on my current stats tests and methods outlined
> > above. If there are other tests that would fit my data better then I'm
> all
> > ears!
> >
> > Thank you!
> >
> >       [[alternative HTML version deleted]]
> >
> > ______________________________________________
> > R-help using r-project.org mailing list -- To UNSUBSCRIBE and more, see
> > https://stat.ethz.ch/mailman/listinfo/r-help
> > PLEASE do read the posting guide
> http://www.R-project.org/posting-guide.html
> > and provide commented, minimal, self-contained, reproducible code.
> >
>
> --
> Michael
> http://www.dewey.myzen.co.uk/home.html
>
> ______________________________________________
> R-help using r-project.org mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
> http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>

	[[alternative HTML version deleted]]