Message-ID: <74cfd9161001251231x254bdb8ame60f028b9b9320c4@mail.gmail.com>
Date: 2010-01-25T20:31:16Z
From: Doug Adams
Subject: Science Fair data
In-Reply-To: <74cfd9161001251229t73902338g1dbb2c02d9b2dcbf@mail.gmail.com>
Whoops, I should say also that we're trying to test and see if
'Division' (representing student age) has a significant effect, (or
can be used to roughly predict) student score.
On Mon, Jan 25, 2010 at 1:29 PM, Doug Adams <dougadams53 at gmail.com> wrote:
> Hi everyone,
>
> I had posted a question on the R Help forum previously about my syntax
> of what I was trying to do with this dataset... ?But now I'd like to
> get an opinion if I'm thinking about it correctly after all. ?I've got
> data for a science fair, and here's what I have for each student:
>
> School District
> School
> Teacher
> Division (Elementary, Junior, High)
> Student, and several scores for each student project
>
> My committee & I decided there aren't enough teachers within each
> school to warrant using Teacher as a level in a hierarchical model, so
> we decided to go with just 'School' and 'Average Student Project' as
> the two levels in the model. ?The problem is that sometimes there is
> only 1 student from a given school, although it doesn't happen too
> often. ?Is HLM theory robust enough that that will still give pretty
> reliable results?
>
> Does this all sound correct so far? ? :-) ? And another question: if I
> were to include District as a third level, would this look right?
>
> lmer(score ~ division + (1|school|district), data=Age6m)
>
> Thanks for reading my novel ?:)
>
> Doug Adams
>