I would assume woof means that you need to compare the % of starters from a barrier to the % winners from a barrier. ie if 10% of runners start from barrier 5 you would expect to get 10% winners, if barriers were insignificant.

There is some problems when you get to wider draws or starts that are used irregularly with regards to to using stats as the small numbers produce unreliable statistics. The method I use is based on places rather than wins as you achieve a more valid set of stats of fewer races. Before doing this I checked some starts which have had thousands of races to compare the win/place results and they were remarkably similar. From this data I have a set of base barrier penalties which I use for guidance when looking at each individual horse in an upcoming race. They are not just mechanically adjusted as this practice ignores the uniqueness of each event.

There are some issues I personally still have with how the barrier data is executed (such as the influence of prices, wet tracks) but feel it gives a reasonable guide to work from.