CODEWITHSUNDEEP
rstatisticsaggregate
user5577851
user5577851

Reputation: 11

R: ttest across columns

I have the following dataset (abbreviated below). On occasion, I want to run a t-test (or other test) on a subset of data, for instance, comparing dcxd in data with d=1 & c=1 vs d=0 & c=0. The closest I've come is using aggregate() to provide the means for these, but have been unable to perform any tests on the data. Any ideas on how to achieve this?

(df <- read.table(header = TRUE, text = "   exp   n s d t   dcxd  brdud    cod
1    1 966 0 1 1  44444  63248  20513
2    1 967 0 0 1  69124 165899 101382
3    1 968 0 0 1 126627 338462 195266
4    1 969 0 1 0  25517  10207   7655
5    1 970 0 0 0  62374  46278  28169
6    1 971 1 1 1  48366  73203  41830
7    1 972 1 0 1  78292 138790  65243
8    1 973 1 1 0  99379  49689  37267
9    1 974 1 0 0  52724   8787   1757
10   2 978 0 0 0  11686   6678   1669"))

#    exp   n s d t   dcxd  brdud    cod
# 1    1 966 0 1 1  44444  63248  20513
# 2    1 967 0 0 1  69124 165899 101382
# 3    1 968 0 0 1 126627 338462 195266
# 4    1 969 0 1 0  25517  10207   7655
# 5    1 970 0 0 0  62374  46278  28169
# 6    1 971 1 1 1  48366  73203  41830
# 7    1 972 1 0 1  78292 138790  65243
# 8    1 973 1 1 0  99379  49689  37267
# 9    1 974 1 0 0  52724   8787   1757
# 10   2 978 0 0 0  11686   6678   1669

Upvotes: 1

Views: 55

Answers (1)

etienne
etienne

Reputation: 3678

Here are two solutions :

create subsets of df :

d1<-df[df$d==1 & df$s==1,]
d2<-df[df$d==0 & df$s==0,]
t.test(d1$dcxd,d2$dcxd)

or without subsets :

t.test(df[df$d==1 & df$s==1,'dcxd'],df[df$d==0 & df$s==0 ,'dcxd'])

Same results for both of them

Welch Two Sample t-test

data:  df[df$d == 1 & df$s == 1, "dcxd"] and df[df$d == 0 & df$s == 0, "dcxd"]
t = 0.185, df = 2.759, p-value = 0.866
alternative hypothesis: true difference in means is not equal to 0
95 percent confidence interval:
 -109662.0  122501.5
sample estimates:
mean of x mean of y 
 73872.50  67452.75

Upvotes: 1

Related Questions