R help - Mar 2015 - t-test: many changing groups.

I'd like to perform a t-test between groups 'A' and 'B'. The
difficulty is that although there is only one response variable, there are many
observations, and the grouping (A or B) differs with each observation. My code
for generating the input data is shown below.
I'd like to know how to approach doing the test, ideally so that the t-test
results for each observation are presented in a table. I'm not sure where to
start as other searches have been futile... something along the lines of
t.test(ind_vars ~ resp_vars) , maybe using rapply, and separating groups by A
and B each time...
# Matrix for response variableN_samples <- 20resp_vars <-
matrix(runif(n=N_samples, min=0, max=1))sample_names <-
paste0("sample_", 1:N_samples )rownames(x=resp_vars) <-
sample_namescolnames(x=resp_vars) <- "resp"resp_vars [1:5,]
# Matrix for independent variablesN_observations <- 100ind_vars <-
matrix(NA, N_observations, N_samples)ind_vars <- apply(ind_vars, c(1,2),
function(x) sample(c("A", "B"),1))ind_var_names <-
paste0("obs_", 1:N_observations)rownames(x=ind_vars) <-
ind_var_namescolnames(x=ind_vars) <- sample_namesind_vars[1:3,1:5]
	[[alternative HTML version deleted]]

Hard to disentangle your code in the non-HTML world of r-help.  Please use
plain text in the future.  Below is one approach you could try.

Jean

# response variables
resp_vars <- runif(20)
names(resp_vars) <- paste0("sample_", seq(20))

# independent variables
ind_vars <- matrix(sample(c("A", "B"), 2000, TRUE),
ncol=20,
  dimnames=list(paste0("obs_", seq(100)), names(resp_vars)))

# save the t-test results to a list
t_fit <- apply(ind_vars, 1, function(obs) t.test(resp_vars ~ obs))

# grab the estimates from each t-test and put it in a data frame
library(broom)
t_est <- do.call(rbind, lapply(t_fit, tidy))
names(t_est) <- c("diff", "meanA", "meanB",
"t", "p.value", "df",
  "conf.low", "conf.high")

head(t_est)



On Sat, Mar 7, 2015 at 10:05 AM, White Sky <bluebarbarossa at hotmail.com>
wrote:
> I'd like to perform a t-test between groups 'A' and
'B'. The difficulty is
> that although there is only one response variable, there are many
> observations, and the grouping (A or B) differs with each observation. My
> code for generating the input data is shown below.
> I'd like to know how to approach doing the test, ideally so that the
> t-test results for each observation are presented in a table. I'm not
sure
> where to start as other searches have been futile... something along the
> lines of t.test(ind_vars ~ resp_vars) , maybe using rapply, and separating
> groups by A and B each time...
> # Matrix for response variableN_samples <- 20resp_vars <-
> matrix(runif(n=N_samples, min=0, max=1))sample_names <-
paste0("sample_",
> 1:N_samples )rownames(x=resp_vars) <- sample_namescolnames(x=resp_vars)
<-
> "resp"resp_vars [1:5,]
> # Matrix for independent variablesN_observations <- 100ind_vars <-
> matrix(NA, N_observations, N_samples)ind_vars <- apply(ind_vars, c(1,2),
> function(x) sample(c("A", "B"),1))ind_var_names <-
paste0("obs_",
> 1:N_observations)rownames(x=ind_vars) <-
ind_var_namescolnames(x=ind_vars)
> <- sample_namesind_vars[1:3,1:5]
>         [[alternative HTML version deleted]]
>
> ______________________________________________
> R-help at 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]]