7422 - Laserfiche WebLink

Home Browse Search

. DRAfT 3.5 STATISTICAL TESTS . A series of statistical tests were conducted to examine for each species: 1. The distribution of depths, velocities and substrates obtained with the 2-hour monitoring design versus samples taken over a full 24 hours. . 2. The relationship between habitat parameters measured at the fish (point) with those taken one meter in or out. 3. The difference in habitat measurements for fish under ice versus fish in ice-free conditions. . 4. The difference in habitat measurements taken during year 1 vs year 2. . Tests conducted between the 2-hour and the 24-hour monitoring programs were designed to investigate whether the 2-hour program represented the habitats utilized by a fish throughout the day and night. Distributions created for the 2-hour program included all 2-hour samples except those collected in conjunction with 24-hour monitoring. Distributions for the 24-hour program included the 2-hour samples collected in conjunction with the 24-hour moni toring. However, tests conducted to compare distributions for point measurements with those taken one meter in or out used all 2-hour monitoring data, including those collected in conjunction with the 24-hour sampling. The distributions of raw depth or velocity data between two data sets were tested using the KolmorgorovjSmirnov (KjS) cumulative frequency test (Press et al. 1986). A '0' statistic was calculated that is defined as the maximum absolute difference between two cumulative distribution functions. If the probability level associated with '0' was less than 0.05, the two distributions were considered to be significantly different. Means of each depth and velocity partition were also calculated. . . . Raw frequency histograms of depth, velocity and substrate were plotted on a conunon maximum independent scale to facilitate a visual comparison of the respective sample sets. In addition, statistical tests for the variance (F- test), the mean (student t-test) and Chi-Square two sample tests were conducted on the data partitions. Sample variance and mean tests were conducted for depth and velocity, while the Chi-Square tests were utilized for comparisons of the substrate data. Sample sizes were normalized to the lowest sample size prior to test of significant differences. In the case of sample tests for the mean, the appropriate t-test with either equal or unequal variance was selected . based on the results of the F-test statistics. All statistical tests were conducted with alpha = 0.05. . 12 .