Boot strap values, confidence level, weights and split values - (Nov/02/2006 )

Hullo friends,


I am pretty confused with some of this terminology.

Boot strap values as i understand is an estimate of good a branching is in a phylogeny tree...or more clearly how true a branch is as a seperate entity....

On the face of this... out of 1000 replicates, a boot strap value of 900 for a branch is supposed to be good...aint it??

So, then what would the confidence value give the measure of ...

What would a branch with a boot strap value of 200 out of 1000 replicates with a confidence value of 100% mean???. ....is it a reliable branching that can be trusted.

are weights the same as boot strap values.???.. what would split values mean??

I hope i did not overwhelm with questions...any help in this regard will be greatly appreciated.

In statistics bootstrapping is a method for estimating the sampling distribution of an estimator by resampling with replacement from the original sample. It is distinguished from the jackknife procedure, used to detect outliers, and cross-validation, whose purpose is to make sure that results are repeatable. There are more complicated bootstraps for sampling without replacement, two-sample problems, regression, time series, hierarchical sampling, mediation analyses, and other statistical problems.

Bootstrapping is becoming the most popular method of testing mediation because it does not require the normality assumption to be met, and because it can be effectively utilized with smaller sample sizes (N < 20). However, mediation continues to be (perhaps inappropriately) most frequently determined using the logic of Baron and Kenny or the Sobel test

weights are not the same as bootstrapping.

really to get answers to this kind of stuff one should look first to wikipedia for the geeks are far wiser there.

In "normal" statistics, the confidence values give you the percentage of values fitting in their range, normally it is calculated as 95% limits, e.g you calculated a mean of 200 with 95% confidence values 191 - 215, then 95% of the values covered by mean are within 191-215. The more narrow the limits the more precise is your measurement/mean.
If you use confidence intervals as introduced from Felsenstein, as estimation of phylogenetic trees, this method provides assessments of "support" for each clade of a tree. It is based on the proportion of bootstrap trees showing that same clade.
So your example I did not understand completely, perhaps you look in the manual of the program for the meaning. I would interpret it as 100% of the bootstrap replicates supporting this branching, but this is in conflict with the 200 from 1000 replicates. So a 20% support of branching (with a 100% chance that the values are within it) is also possible as meaning. (I would prefer the first explanation.)