ScALING
The theory contains no prescriptions for scaling or analysis, but again
operational practices have developed to guide research. Not all cultivation
researchers conform to the practices used by the theoreticians, however. In
some studies, the single measure of viewing per day is used as the exposure
measure, but in other studies several measures are aggregated in some way to
construct a viewing scale. There are three forms of aggregation.
Summative scales: Weaver and Wakshlag ( 1986) computed two indexes:
non-crime related viewing and crime related viewing. For every hour-long
crime show a respondent checked in his/her seven-day diary, 60 minutes
were added to the crime viewing index. A similar procedure was used to
translate check marks on non-crime shows into index minutes. The authors
report that reliability was marginal for crime related viewing (KR-20 = .64)
and non-crime related viewing (KR-20 = .69). But the Kuder-Richardson test
is a split half method of testing internal consistency of a scale. It splits the
items on the scale into two sets and examines the degree to which measures
on one half of the items are related to measures on the other half of items. It
does not seem that the results of Kuder-Richardson would address the issue
of reliability here, because there is no reason to expect someone who watches
w. / AMES P OTTER
one crime show to watch all other crime shows. So a low KR-20 coefficient
is less an indication of the quality of the scale than it is an indication that
viewers have differential viewing preferences.
Hawkins and Pingree (1980, 1981) summed the minutes recorded in
respondents' diaries to construct six program type exposure scales: news,
situation comedies, crime adventures, dramas, game shows, and cartoons.
They also had respondents use a 4-point scale ("never" to "often") to indicate
how often they viewed programs in these seven genres. They found that
"except for cartoons (r = .46), correlations between the two forms of
measurement were quite low and inconsistent" (p. 205). Given the choice of
the two measures, they concluded "while the two methods of measurement
do not relate well to each other, viewing of the content types measured by the
TV diary is better related to the cultivation dependent variables, suggesting
that the diary is also a better measure of long-term viewing habits than the
averaging measures" (p. 205).
Several researchers have computed summative scales where the elements
summed were not hours of exposure. For example, Carlson (1983) summed
the responses (1 ="never watched"; 4 ="almost always watch") to 16 crime
shows to construct a Crime Show Viewing scale that had a range of 16 to 64.
Volgy and Schwarz (1980) used responses to a list of25 prime-time programs
(how regularly viewed?) and computed exposure scales for medical and
minority show viewing.
Multiplicative scales: Some researchers have computed a weekly viewing
figure from daily viewing data by multiplying the daily viewing response by
7. For example, Carveth and Alexander (1 985) used their respondents' hours
of viewing per day and multiplied by 7 to get the weekly hour figure. This
scale is based on the assumption that viewing is the same all seven days of the
week. Ogles and Sparks (1989) multiplied the weekday viewing measure by
5 and the weekend day viewing measure by 2 and then summed the two
products to get a weekly figure.
Average exposure: Another type of scale is the average. For example, in
two studies, exposure was operationalized as the average between two
measures: amount of television watched yesterday (a weekday) and the
amount of television that usually watched on an average weekday (Perse,
Ferguson, & McLeod, 1994; Rubin, Perse & Taylor, 1988).
In another study, exposure was operationalized as the sum of hours of
TV viewed on a typical weekday and a typical weekend day (Hawkins,
Pingree, & Adler, 198 7). This type of scale is merely an operational device that
appears to be built on a feeling that a single item is not a good indicator of
viewing, nor is any item by itself. This scale has no conceptual meaning, that
is, it does not reflect daily or weekly viewing. It can also be misleading. For
example, if person A watches 4 hours on a typical weekday and 6 hours on
a typical weekend day, her score on this scale is 5. IfpersonB watches 2 hours
on a typical weekday and 8 hours on a typical weekend day, his score on the
scale is 6. The scale indicates that B is a heavier TV viewer despite the fact that