the coefficient of variation for the l-cm samples is the same as that
for the 5-cm samples at Site II. The alternate hypothesis is that the
coefficient of variation for the l-cm samples is not the same as that
for the 5-cm samples. The t~ statistic for comparing the two coefficients of variation is 6.21 with four degrees of freedom.
The critical

value at the 95 percent confidence level is 3,18.
Therefore, the ruling
hypothesis is rejected and the alternate hypothesis is accepted.
Site III
The sampling coefficient of variation for the l-cm sampling was 0.390 as
compared to 0.29 for the 5~cm samples.
The ruling hypothesis is that
the coefficient of variation for the l-cm samples is the same as that

for the 5-cm samples at Site III.

The alternate hypothesis is that the

coefficient of variation for the 1-cm samples is not the same as that
for the 5-cm samples.
The t* statistic for comparing the two coefficients of variation is 2.33 with four degrees of freedom.
The critical
value at the 95 percent confidence level is 3.18.
Therefore, the ruling
hypothesis is not rejected.
Site IV
The sampling coefficient of variation for the l-cm sampling was quite

large, at 0.428 compared to the small value of 0.191 for 5-cm sampling.
Again, the ruling hypothesis is that the coefficient of variation for
the 1-cm sampling is the same as that for the 5-cm samples. The t”

Statistic for comparing the two coefficients of variation is 6.71 which

is larger than the critical value at the 95 percent confidence level of

3.18.
Therefore, the ruling hypothesis is rejected and the alternate
hypothesis that the coefficients are not equal is accepted.

SUMMARY

Assessment of the coefficients of variation of sampling depths, treating
each plot independently, and using t~ seems to indicate significance in
some sites, not in others.
Considering the clouded validity of using
t*, an alternate method of assessment of the coefficients of variation
was considered.
For paired samples (in this case, the paired coefficients of variation

for each site), the Wilcoxon signed rank test (Gibbons, 1976; pp. 131-137)

can be helpful in defining a possible level of significance to the count
of the direction of differences in paired data.
In this case, the
probability that only one data pair out of four pairs indicates a differ-

ence in a direction opposite of the other three (i.e., Cs > C, (Site
II)

as opposite C, > Cs

(Sites I, III, and IV)) is 0.25.

Even if all

pairs had been in the same direction (i.e., C; > C5 in all cases), the

678

Select target paragraph3