To determine if job class versus LHD was driving the variation
in agreement, three two-factor ANOVAs without replication,
using job class and LHD as factors, were conducted: one each
for tasks, knowledge, and resources. Three additional ANOVAs
with replication were conducted using job class and two groups as
factors: smaller LHDs with less than 50 FTEs and larger LHDs with
50 or more FTEs. These categories were based on the size of the
LHDs in our data. Seven of eight job classes that were common
across all LHDs were used for these ANOVAs. Agency Director
was not used for the job class versus LHD analysis because with
only one director in each LHD, agreement could not be calculated.