Discrete Events
Another measure that is often used as an index of central tendency is the rate of
occurrence of some event. As a prototypical example, consider the rate of
contracting polio for children inoculated with the Salk vaccine. Even though
individual children either get the disease or do not, the rate tells us something about
the ability of inoculated children, as a group, to fight the disease.
How can we view a rate (or probability) as a measure of central tendency? First,
a probability can be formally viewed as a mean through what some would regard as
a bit of trickery. If we code the event “polio” as a 1, and the event “no polio” as a 0,
then the probability of getting polio is merely the mean of these Boolean values.
Producing a formal average, however, does not automatically give us a measure of
central tendency. We need to be able to interpret this average as a signal related to
the causes of polio. Compare the distribution of values in the dichotomous case to
the ideal case of the weighing example. In the dichotomous case, the mean is not a
value that can actually occur in a single trial. Rather than being located at either of
the peaks in the distribution, the mean is located in the valley between, typically
quite far from the observed values. Thus, it is nearly impossible to think about the
rate or probability as the true-value component of any single observation and the
occurrence or nonoccurrence of an individual case of polio as the sum of a true
value and a random error component. We suspect this is largely why the idea of a
central tendency in dichotomous situations is the least tangible of all.
It might help in reasoning about this situation to conceive of some process about
which the rate or probability informs us. In the disease example, the conception is
fairly similar to the earlier height example: A multitude of factors influence the
propensity of individuals to get polio—level of public health, prior development of
antibodies, incident rate of polio, age—all leading to a rate of getting the disease in
some population. So even though individuals either get polio or do not, the
propensity of a certain group of people to get polio is a probability between 0 and 1.
That value is a general indicator of the confluence of polio-related factors present in
that group.
As with our height example, although an absolute rate may have some meaning,
we think it is much easier to conceptualize the meaning of a signal when we are
Discrete Events
Another measure that is often used as an index of central tendency is the rate of
occurrence of some event. As a prototypical example, consider the rate of
contracting polio for children inoculated with the Salk vaccine. Even though
individual children either get the disease or do not, the rate tells us something about
the ability of inoculated children, as a group, to fight the disease.
How can we view a rate (or probability) as a measure of central tendency? First,
a probability can be formally viewed as a mean through what some would regard as
a bit of trickery. If we code the event “polio” as a 1, and the event “no polio” as a 0,
then the probability of getting polio is merely the mean of these Boolean values.
Producing a formal average, however, does not automatically give us a measure of
central tendency. We need to be able to interpret this average as a signal related to
the causes of polio. Compare the distribution of values in the dichotomous case to
the ideal case of the weighing example. In the dichotomous case, the mean is not a
value that can actually occur in a single trial. Rather than being located at either of
the peaks in the distribution, the mean is located in the valley between, typically
quite far from the observed values. Thus, it is nearly impossible to think about the
rate or probability as the true-value component of any single observation and the
occurrence or nonoccurrence of an individual case of polio as the sum of a true
value and a random error component. We suspect this is largely why the idea of a
central tendency in dichotomous situations is the least tangible of all.
It might help in reasoning about this situation to conceive of some process about
which the rate or probability informs us. In the disease example, the conception is
fairly similar to the earlier height example: A multitude of factors influence the
propensity of individuals to get polio—level of public health, prior development of
antibodies, incident rate of polio, age—all leading to a rate of getting the disease in
some population. So even though individuals either get polio or do not, the
propensity of a certain group of people to get polio is a probability between 0 and 1.
That value is a general indicator of the confluence of polio-related factors present in
that group.
As with our height example, although an absolute rate may have some meaning,
we think it is much easier to conceptualize the meaning of a signal when we are
การแปล กรุณารอสักครู่..
