Free Index Values

Index values are most often some type of count or an average of such counts over several sampling units. Some biologists use these index values as if they represented 'relative abundance', even though a very large literature confirms that it is difficult to interpret such index values without making ludicrous assumptions. Such assumptions nearly always include the notion that the underlying detection probability is constant across a wide variety of variables associated with (1) the observer, (2) the environment, and (3) the species of interest. In addition, these variables must be assumed fixed and not time-varying. Without these assumptions, index values cannot be interpreted. Detection probability often varies by an order of magnitude, thus cannot be simply ignored.

            Index values are really just numbers, they do not represent data. The word 'data' has the connotation that information is contained in an interpretable manner. Index values do not represent data as they lack interpretability. If the heroic assumptions could be met (even approximately), then the numbers (index values) could be interpreted. However, it is clear from decades of research that the assumptions concerning a constant detection probability are blatantly false.

             Several sound approaches have been available for decades to collect meaningful data and avoid the problems with index values. Double sampling with ratio or regression estimators is an obvious approach. Others have used distance sampling or capture-recapture sampling to allow for the proper estimation of detection probabilities. I certainly advocate such approaches as the collection of index values seems increasingly pointless (embarrassing, in fact).

A huge amount of money is spent each year on the collection of index values. Thus, I am providing free index values (below). These can be used in about any way that index values 'from the field' are currently used. That is, one can take the raw index values, or they can be transformed (e.g., logarithms and square roots are popular), or can be multiplied by some constant (for example, if one wants larger index values than I have available, they can be multiplied by 3.14159 or 10 or 30 or whatever), or one can select (by visual inspection) strings of index values that tend to increase or decrease. This last alternative allows many forms of trend analysis to be performed on the numbers. Those wanting integer values can merely 'truncate off' the real parts of the index values. These values are provided without cost; thus saving substantial sums of money on wasted field research.


HTML view

Excel File

TXT file (comma and space delimited text file)


Other Numbers Coming Soon:

  • Wild Outliers
  • Various Positive Numbers

There may be a small charge from some categories (e.g., bogus integers).



Recent papers in Auk regarding index values

PDF Format

Rosenstock et al.2002

Thompson 2002