0

I get All values of t are equal to 0.999999999999995 \n Cannot calculate confidence intervals

My data is this:

$data
    y x2 x3
1   2  1  1
2   0  2  1
3  -2  3  1
4  -4  4  1
5  -6  5  1
6   5  1  2
7   3  2  2
8   1  3  2
9  -1  4  2
10 -3  5  2
11  8  1  3
12  6  2  3
13  4  3  3
14  2  4  3
15  0  5  3
16 11  1  4
17  9  2  4
18  7  3  4
19  5  4  4
20  3  5  4
21 14  1  5
22 12  2  5
23 10  3  5
24  8  4  5
25  6  5  5

I did this:

b<-boot(data=d,statistic=lmcoefs,R=5)
> b

ORDINARY NONPARAMETRIC BOOTSTRAP


Call:
boot(data = d, statistic = lmcoefs, R = 5)


Bootstrap Statistics :
    original        bias     std. error
t1*        1 -9.436896e-15 4.831384e-15
t2*       -2  1.110223e-15 1.169693e-15
t3*        3  1.776357e-15 8.881784e-16

Then I calculated the confidence intervals using boot.ci: which gives

c<-boot.ci(b,conf=0.95,type="all")
[1] "All values of t are equal to  0.999999999999995 \n Cannot calculate confidence intervals"

And I get this error I cant get rid of. I dont understand why either.

Siguza
  • 21,155
  • 6
  • 52
  • 89
LoveMeow
  • 3,858
  • 9
  • 44
  • 66
  • Maybe because your standard errors are effectively ~0? Your boot strapped estimates are also your CI. – Vlo Aug 11 '14 at 19:02
  • 2
    You only did 5 bootstrap replicates. How much variation in the bootstrap samples did you expect in only 5 replications, particularly if there isn't much variation in your data to begin with? – joran Aug 11 '14 at 19:17

1 Answers1

1

The problem is that I did not do enough bootstrap replications. Hence I did not have enough variation in my data, and that's why I encountered this error.

user213544
  • 2,046
  • 3
  • 22
  • 52
LoveMeow
  • 3,858
  • 9
  • 44
  • 66