Does it make a difference when gathering a sample size for STRATUM to use 3-6% sample of total street segments or 3-6% sample of total street miles? The information on sampling for STRATUM indicates both; see below. I'm looking at a community in New Jersey where I have taken 6% of all street segments to sample and I noticed that this happens to equal 6% of all street miles for the community as well. Coincidence? Thanks.
"For STRATUM, sample inventories are generally a 3-6% sample of total street segments, depending on community size and variation from segment to segment, and will produce about a 10% standard error for the total number of trees citywide. The following, general guidelines can be used as a starting point for determing sample sizes in communities based on human population sizes:
• For communities with less than 50,000 persons, sample size is 6% of total street miles"
Which is it? Street Miles or Street Segments for Stratum.
Moderators: i-Tree Support, i-Tree Team
-
Brian NJFS
- Posts: 6
- Joined: Tue Nov 27, 2007 11:03 am
Hmmm
In statistical terms, though, doesn't the standard error of a mean depend on the number of times the population is sampled (i.e., n) rather than the percentage of the population being sampled? Put another way, won't 30 samples one-tenth of a mile long usually produce a better estimate than three samples that are one mile long? If that is right--and I am certainly not a statistician so this might be total hooey--then shouldn't it be preferable to take the percentage of the samples rather than of the miles?
It may be that in the real world it won't make much difference in many communities, unless the distribution of sample lengths is skewed right (i.e., the average sample is unusually long). If the sample set is being created using the i-Tree utility, by the way, the user can look at the distribution of segment lengths for a community's street segments using ArcMap: highlight all the segments, click (for ArcMap 9.1) Selection --> Statistics, then select Length under Field.
It may be that in the real world it won't make much difference in many communities, unless the distribution of sample lengths is skewed right (i.e., the average sample is unusually long). If the sample set is being created using the i-Tree utility, by the way, the user can look at the distribution of segment lengths for a community's street segments using ArcMap: highlight all the segments, click (for ArcMap 9.1) Selection --> Statistics, then select Length under Field.