same biological conclusions were reached regarding traits
associated with fighting ability. As the importance of
fighting experience increased and the initial variation in
individual RHP decreased, the lower the correlation between
initial RHPs and position in the hierarchy (Beaugrand
1997). Thus, the presence of an experience effect
is unlikely to result in spurious associations between fighting
ability and individual traits (Type I error), although
under some circumstances (very strong experience effect)
it may result in failure to detect effects of subtle individual
differences in RHP (Type II error). The effects of casque
height and flank patch area emerge as being statistically
significant in the present study, even under this potential
loss of sensitivity caused by experience effects.
The B–T model is a useful tool for analysing contest data
because it maximizes the information that can be obtained
from a limited number of individuals. For instance,
with 36 individuals, using each individual only once
would result in 18 contests (18 dyads of unfamiliar males).
This is similar to the sample sizes used in many studies of
animal contests but results in poor statistical power and
a greatly inflated chance of Type II error in a multiple
regression context (Cohen & Cohen 1983; Cohen 2000;
MacNally 2000). In contrast, the B–T model allows the
simultaneous estimation of the relative influence of several
individual traits, while controlling for ‘contest
history’, based on a robust sample size (in this case, 107
contests); yet it does not require that every individual contests
every other, thereby minimizing stress to the test subjects.
In addition to identifying the role of male contest
competition in the evolution of multiple male signals
and sexual dimorphism in dwarf chameleons, our study
highlights the utility of tournament designs and the structured
B–T model for analysing contest data.
same biological conclusions were reached regarding traits
associated with fighting ability. As the importance of
fighting experience increased and the initial variation in
individual RHP decreased, the lower the correlation between
initial RHPs and position in the hierarchy (Beaugrand
1997). Thus, the presence of an experience effect
is unlikely to result in spurious associations between fighting
ability and individual traits (Type I error), although
under some circumstances (very strong experience effect)
it may result in failure to detect effects of subtle individual
differences in RHP (Type II error). The effects of casque
height and flank patch area emerge as being statistically
significant in the present study, even under this potential
loss of sensitivity caused by experience effects.
The B–T model is a useful tool for analysing contest data
because it maximizes the information that can be obtained
from a limited number of individuals. For instance,
with 36 individuals, using each individual only once
would result in 18 contests (18 dyads of unfamiliar males).
This is similar to the sample sizes used in many studies of
animal contests but results in poor statistical power and
a greatly inflated chance of Type II error in a multiple
regression context (Cohen & Cohen 1983; Cohen 2000;
MacNally 2000). In contrast, the B–T model allows the
simultaneous estimation of the relative influence of several
individual traits, while controlling for ‘contest
history’, based on a robust sample size (in this case, 107
contests); yet it does not require that every individual contests
every other, thereby minimizing stress to the test subjects.
In addition to identifying the role of male contest
competition in the evolution of multiple male signals
and sexual dimorphism in dwarf chameleons, our study
highlights the utility of tournament designs and the structured
B–T model for analysing contest data.
การแปล กรุณารอสักครู่..
same biological conclusions were reached regarding traits
associated with fighting ability. As the importance of
fighting experience increased and the initial variation in
individual RHP decreased, the lower the correlation between
initial RHPs and position in the hierarchy (Beaugrand
1997). Thus, the presence of an experience effect
is unlikely to result in spurious associations between fighting
ability and individual traits (Type I error), although
under some circumstances (very strong experience effect)
it may result in failure to detect effects of subtle individual
differences in RHP (Type II error). The effects of casque
height and flank patch area emerge as being statistically
significant in the present study, even under this potential
loss of sensitivity caused by experience effects.
The B–T model is a useful tool for analysing contest data
because it maximizes the information that can be obtained
from a limited number of individuals. For instance,
with 36 individuals, using each individual only once
would result in 18 contests (18 dyads of unfamiliar males).
This is similar to the sample sizes used in many studies of
animal contests but results in poor statistical power and
a greatly inflated chance of Type II error in a multiple
regression context (Cohen & Cohen 1983; Cohen 2000;
MacNally 2000). In contrast, the B–T model allows the
simultaneous estimation of the relative influence of several
individual traits, while controlling for ‘contest
history’, based on a robust sample size (in this case, 107
contests); yet it does not require that every individual contests
every other, thereby minimizing stress to the test subjects.
In addition to identifying the role of male contest
competition in the evolution of multiple male signals
and sexual dimorphism in dwarf chameleons, our study
highlights the utility of tournament designs and the structured
B–T model for analysing contest data.
การแปล กรุณารอสักครู่..