The TcCB data are much more strongly skewed than the Chlorophyll A data. The latter are
quite close to being normally distributed, as revealed by the QQ plot. Thus I expect the CIs
based on the t distribution to perform relatively well.
The 95% CI of the mean based on the t distribution is [29.66, 70.95] for Chlorophyll A and [-
0.628, 8.458] for TcCB (I obtained these by running a two-tailed, one-sample t-test; the
confidence intervals are included as part of the output). For Chlorophyll A, the bootstrap means
are very nearly normally distributed (based on the QQ plot), so we can use the Normal CI that
uses the standard error of the bootstrap means, giving [31.43, 69.00]. This is a little bit
narrower than the t-distribution CI, but the difference is modest. For TcCB, the bootstrap means
are strongly skewed, so the BCa interval is best; it is [1.363, 14.151]. This is quite a bit wider
than the t-distribution CI, as well as being skewed towards higher values. (Note that your values
won’t be exactly the same as mine; see # 4.)
The bias is -1.466 for chlorophyll and -2.813 for TcCB. Thus in both cases the sample
standard deviation is likely to be an underestimate of the population standard deviation.
For chlorophyll the values were -1.551, -0.896, -1.329, -1.403, and -1.377. For TcCB the values
were -2.881, -3.872, -2.127, -2.723, and -2.634. Thus the bias estimates aren’t very precise
with 1000 replicates. The CI limits vary by a similar amount.
Bootstrapping TcCB with 100 replicates, I get the following BCa CIs:
( 1.52, 38.31 )
( 2.30, 32.71 )
( 2.31, 38.08 )
( 6.26, 41.86 )