CODEWITHSUNDEEP
r
user969113
user969113

Reputation: 2429

Quantitative analysis of treatment effects

I need to analyse a biological study. The study design is rather simple. It includes 3 groups. 1 control group and 2 test groups. The two test groups are treated with the same drug but different doses. Each group has approximately 14 observations and I look at around 600 variables. I already used an one way ANOVA to determine the significant hits between the groups. However, I now want to know if there is a quantitative effect between the treatment doses. In other words, is the treatment effect statistically greater when you compare test group 1 (lower dose) with the control group or is the treatment effect bigger when you compare test group 2 (higher dose) with the control group. I also need to include 1 or 2 co-variates in the analysis.

Unfortunately, I do not know if there is a statistical test/technique in R which I can use to answer this question. Any advise is very much appreciated.

Syrvn

Upvotes: 0

Views: 1374

Answers (1)

EDi
EDi

Reputation: 13280

This sounds like a job for the multcomp-package, there are a lot of resource out there. Also have a look at this book.

Here is an example comparing the response of two groups to a control group:

require(multcomp)
mice <- data.frame(group=as.factor(rep(c("C","1","2"),rep(6,3))), 
                   score=c(58, 32, 59, 64, 55, 49, 73, 70, 68, 71, 60, 62, 53, 74, 72, 62, 58, 61))
# reoder factor, so that Control is the 1st level
levels(mice$group) <- c("C", "1", "2")

plot(score ~ group, data = mice)

enter image description here

# Anova
mod <- aov(score ~ group, data = mice)
# Multiple Comparisons with Dunnett contrasts (=Compare to control)
summary(glht(mod, linfct=mcp(group = "Dunnett")))

     Simultaneous Tests for General Linear Hypotheses

Multiple Comparisons of Means: Dunnett Contrasts


Fit: aov(formula = score ~ group, data = mice)

Linear Hypotheses:
           Estimate Std. Error t value Pr(>|t|)  
1 - C == 0   -4.000      4.965  -0.806   0.6427  
2 - C == 0  -14.500      4.965  -2.920   0.0195 *
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 
(Adjusted p values reported -- single-step method)

HTH,

Upvotes: 1

Related Questions