Data were analyzed using a generalized linear model
approach [27,28]. First, we tested for differences between
years, between seasons and between seasons within each
year (year X season interaction) in the occurrence of
viruses or the abundance of indicator bacteria. For viruses,
the response variable (presence/absence) was assumed to
follow a Bernoulli distribution; the abundance of bacteria
was assumed to follow a quasiPoisson distribution (to
compensate for overdispersion when a Poisson distribution
was used). A full factorial model with factors Year
(2001 vs. 2002) and Season (cold-dry vs. warm-rainy) was
fitted to each response variable (occurrence of rotavirus,
enterovirus and astrovirus; abundance of TC, FC, and FE)
separately. The terms' significance was judged with basis
on the change in deviance that its deletion from the
model produced, which approximately follows a Chisquare
distribution [27].
Data were analyzed using a generalized linear modelapproach [27,28]. First, we tested for differences betweenyears, between seasons and between seasons within eachyear (year X season interaction) in the occurrence ofviruses or the abundance of indicator bacteria. For viruses,the response variable (presence/absence) was assumed tofollow a Bernoulli distribution; the abundance of bacteriawas assumed to follow a quasiPoisson distribution (tocompensate for overdispersion when a Poisson distributionwas used). A full factorial model with factors Year(2001 vs. 2002) and Season (cold-dry vs. warm-rainy) wasfitted to each response variable (occurrence of rotavirus,enterovirus and astrovirus; abundance of TC, FC, and FE)separately. The terms' significance was judged with basison the change in deviance that its deletion from themodel produced, which approximately follows a Chisquaredistribution [27].
การแปล กรุณารอสักครู่..