> Date: Thu, 23 Jun 2011 15:41:25 -0700
> From: jmo...@student.canterbury.ac.nz
> To: r-help@r-project.org
> Subject: Re: [R] Time-series analysis with treatment effects - statistical
> approach
>
>
> Mike Marchywka wrote:
> >
> >> I discovered a way to do repetitive tasks that can be concisely specified
> >> using
> >> something called a computer.
>
> Now that's funny :)
well, there is a point to that and that is that with cheap computations you can
do different analyses than you did in the past.
>
> There were not controlled tests. It was a field experiment testing the
> effects that various pavement designs have on underlying soil moisture. Two
> designs incorporated a porous pavement surface course, while two others were
> based on standard impervious concrete pavement...the control was just bare,
> exposed soil.
>
> As you can see from the graph, the control responds quickly to rainfall
> events, but dries out quickly as well due to evaporation. The porous
> pavement allows for quick infiltration of precipitation, while the
> impervious pavement eventually allows infiltration of rainfall, but it's
> delayed.
>
> My objective is to be able to differentiate between the pavement treatments,
> such that I can state with statistical confidence that porous pavements
> affects underlying soil moisture differently than impervious pavements.
>
> I think this is obvious just looking at it, but I wanted to be able to back
> it up with stats. What I'd done previously is to average by week. But as I
> mentioned, I thought that an anova table with 104 rows relating to each week
> was a poor way of analyzing the data. But that being said, it effectively
> allows me to check for treatment-related differences.
I don't think we've mentioned R in the past few posts but I guess pointing
people to useful things that R can do is not too big a problem and
if you have ever dealt with "analysis for the sake of rationalization"
you can appreciation that is a huge problem :) Generally you'd
like to have reproducible results and if you don't have IID ( stationary
parameters of the population you wish to characterize) you
are not even asking a good question about the system. It may
be helpful as quick check of something but otherwise difficult to
interpret- do your results mean these things are "different" in the
desert? You appear to have data points from a bunch of different situations.
After the fact selection is often helpful but generally stats people frown
on that as "backing up" anything ( unless it supports sponsor's opinion LOL).
http://www.itl.nist.gov/div898/handbook/prc/section4/prc432.htm
I guess I'd either go with dynamic model or convert into dollars and then
see if you have clinically and statistically significant differences in things
of relevance.
>
> Thanks for the suggestions to date. Maybe the more I explain what I'm trying
> to achieve, the more focussed the suggestions will be. The vaguer the
> question, the broader the response, right?
>
> Thanks again,
> Justin
>
> --
> View this message in context:
> http://r.789695.n4.nabble.com/Time-series-analysis-with-treatment-effects-statistical-approach-tp3615856p3621179.html
> Sent from the R help mailing list archive at Nabble.com.
>
> ______________________________________________
> R-help@r-project.org mailing list
> 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]]
______________________________________________
R-help@r-project.org mailing list
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.
