recent periods and as standard deviations values that are first 25 percent (best guess) and then 50 percent (high value) of the respective mean. The differences between the members of each pair (the mid 1970s average minus the late 1990s average) were computed in each funding case, and the cumulative histograms of the differences were plotted on normal probability paper. The cumulative probability value for which the difference was less than zero (1970s funding minus 1990s funding < 0) was also computed and designated in the cumulative distribution plots.
The cumulative probability plots and information on them may be used to answer the posed question. If the zero-level cumulative probability value mentioned above is near 50 percent (say within 10 percent up or down), then it is doubtful that differences in funding levels are significant. If it is significantly lower than 50 percent then there is high likelihood that the late 1990s funding level is indeed lower than the mid 1970s level. If it is significantly greater than 50 percent then there is high likelihood that the late 1990s funding level is higher than the mid 1970s level. Results and significant findings for total water resources funding and for funding in each major FCCSET category are presented below.