137
479
THS ~

 

TWIN? 3'

 

Seam LIBRARY
Michigan State
University

 

 

 

This is to certify that the
thesis entitled

ABUSED CHILDREN AND TELEVISION: EXAMINING
PARASOCIAL INTERACTION IN SPECIAL POPULATIONS

presented by
SARAH FRANCES CROOK

has been accepted towards fulfillment
of the requirements for the

MA. degree in COMMUNICATION

WM LRMmQr

V “Q05 Profess‘od Signature
82 j '7 l 05
Date

MSU is an Affinnative ActiorVEqual Opportunity Institution

v.—---.u—y----n-.-----o-u---~-------o-»-<-.-—.. ,.

 

PLACE IN RETURN Box to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 cJClRC/DateDuepas-pts

ABUSED CHILDREN AND TELEVISION: EXAMINING PARASOCIAL
INTERACTION IN SPECIAL POPULATIONS
By

Sarah Frances Crook

A Thesis

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF ARTS
Department of Communication

2003

ABSTRACT
ABUSED CHILDREN AND TELEVISION: EXAMINING PARASOCIAL
INTERACTION IN SPECIAL POPULATIONS
By

Sarah Frances Crook

The purpose of this exploratory study was to examine the difference between abused and
non-abused children’s relationships with television characters. Abused and non-abused
children between the ages of 7 and 16 years old were interviewed about their favorite
television character. Analyses were conducted to investigate parasocial interactions and
how they differ by abuse status (abused vs. non-abused) and gender (boys vs. girls).
Findings suggested that abused children parasocially interact with their favorite TV
characters less Often than non-abused children. Girls were more likely to identify social
behavior as an important character attribute in their favorite character, while boys were
likely to report strength. Conversation topics during these parasocial interactions
typically were about episode information for non-abused girls. Abused girls reported
seeking and giving help as the reason why they engaged in these relationships, while non-

abused girls described affective reasons.

Cepyright by
SARAH FRANCES CROOK
2003

TABLE OF CONTENTS

LIST OF TABLES ..................................................................................... v
INTRODUCTION .................................................................................... 1
METHOD ............................................................................................ 10
RESULTS ............................................................................................ 16
DISCUSSION ....................................................................................... 23
REFERENCES ...................................................................................... 32
FOOTNOTES ....................................................................................... 36
APPENDICES ....................................................................................... 39

iv

LIST OF TABLES

TABLE 1 ........................................................... . .................................. 42
TABLE 2 ............................................................................................. 43
TABLE 3 ............................................................................................. 44
TABLE 4 ............................................................................................. 45

INTRODUCTION

Children are our future. When children are growing up, they look to other people
to help them develop their personality, style, and other defining features. Since they are
developing a sense of self, not only do children imitate their parents, teachers, and friends
(Weiner & Elkind, 1972, p. 83), but they also have the television and other media as
sources of possible imitation. The television offers a plethora of interesting people to be
like. Even though real-life models are important to children’s development, “many of the
adults witnessed by children come from prime-time television programs they come to
prefer and frequently view” (F emie, 1981, p. 47). It is not that unexpected that children
would prefer these models to real-life people. Such characters are often physically
attractive, engaging, and the ideal prototypes of success. Thus, it becomes important to
ask what impact television characters have on children’s socialization?

Several studies have attempted to answer this very question. Most of the research
has focused on those characters that foster young viewers’ identification. Typically,
identification has been defined as the process of “the viewer, in fantasy, [putting] himself
in the place of a character and momentarily [feeling] that what is happening to that
character is happening to himself ’ (Maccoby & Wilson, 1957, p. 76). In fact,
identification has been found to mediate the impact of television violence on aggression
as well as thin body ideals on eating disorders in adolescents (Harrison, 1997; Huesmann,

Lagerspectz, & Eron, 1984). Thus, identification with media personae may be an

important intervening variable between television exposure and other behaviors (Reeves
& Miller, 1978).

Scholars interested in the process of identification have examined primarily
gender differences. Studies have generally shown that boys and girls are more likely to
identify with male characters on television (Eisenstock, 1984; Hoffner, 1996; Lonial &
Auken, 1985; Miller & Reeves, 1976; Reeves & Miller, 1978). These findings have been
explained by the fact that TV continues to feature a disproportionate amount of men
when compared to women (Gerbner, Gross, Morgan & Signorielli, 1994, p. 31). Thus,
male characters seem to be potent role models among both young males and females.

Although children are identifying frequently with males on television, research
reveals that young boys and girls are drawn to different attributes of media characters.
Studies show boys typically identify with characters that are active and physically strong
whereas girls are drawn to physically attractive characters (Miller & Reeves, 1976;
Reeves & Greenberg, 1977; Reeves & Lometti, 1979; Reeves & Miller, 1978). In
addition, some studies have cited that children identify with characters based on
intelligence and humor (Reeves & Greenberg, 1977; Reeves & Lometti, 1979; Reeves &
Miller, 1978). As such, character traits seem to be important predictors of the likelihood
of children’s identification.

Another body of research indicates that people not only identify with characters
on television, but also have cognitive or affective relationships like parasocial
interactions. Parasocial interactions are imaginary relationships that viewers form with
media characters. Originally, the idea of parasocial interactions was developed by

Horton and Wohl (1956) who were interested in viewers’ relationships with television

news anchors. The researchers argued that newscasters can create an illusion of intimacy
because the audience members perceive the message as directed towards them (Horton &
Wohl, 1956). When audience members start to anticipate these illusionary conversations,
they have been noted to talk back to their television screens (Noble, 1975) or feel as if
they know these characters as well as their friends and neighbors (Hoffner, 1996).

Most of the research on parasocial interaction has used adults as subjects. One
series of studies evaluated adult’s interactions with television newscasters (Choi, 1993;
Cortez, 1991; Perse, 1990; Rubin, Perse, & Powell, 1985). Houlberg (1984) interviewed
respondents over the telephone and found that over half of the participants reported
having parasocial relationships with news anchors. Levy (1979) also assessed whether
adults engage in parasocial relationships with people in the news. He found that a
majority of adults in his sample reported having such interactions. Further, parasocial
interaction was positively correlated with exposure to TV news. That is, as individuals’
exposure to television news increased so did the frequency of their parasocial
interactions. Or, the reverse process is equally plausible; individuals who parasocially
interact with news anchors may be more likely to seek out and watch television news
programming.

In addition to newscasters, research reveals that other characters on television can
also create a sense of intimacy with the audience. One such media personality is the
television- shopping host. Grant, Guthrie, and Ball-Rokeach (1991) examined the
relationship between exposure to television shopping networks and parasocial
interactions with emcees. Their results revealed that heavy viewers of television

shopping programming were more likely to develop parasocial relationships with the

hosts. Such findings are presumably due to the fact that shopping hosts, much like TV
news anchors, talk directly and explicitly to audience viewers thereby heightening an
illusion of intimacy.

Soap opera stars have also been examined in relation to parasocial interaction
(Perse & Rubin, 1988; Yanof, 1991). It has been argued that female college students
often schedule their classes around soap operas and this persistence in viewing may be an
important precursor for parasocial relationships (Lemish, 1985, p. 281). Indeed, results
from several studies (Perse & Rubin, 1989; Rubin & Perse, 1987; Turner, 1993) suggest
that adults do parasocially interact with soap opera characters, especially when the
viewers perceive soap operas as real.

Although most of the research on parasocial relationships has focused on adults,
children may be another subgroup prone to such illusionary interactions. To date,
however, only one published study in the US. has actually looked at children’s parasocial
interactions with television characters. Hoffner (1996) interviewed a sample of 7- to 12-
year olds about their parasocial relationships with their favorite television characters.
She was not only interested in the frequency of such illusionary relationships but also the
attributes of characters that predict parasocial activity. Similar to the results on
identification with television characters above, the study revealed that parasocial
interaction with a male television character was predicted by physical attractiveness and
intelligence for both boys and girls and by strength for boys only. Females’ parasocial
interaction with female characters was predicted solely by physical attractiveness. In
addition, parasocial interaction was more common with characters that were regarded as

kinder, more helpful, more caring, less mean, and less selfish.

Even though children have parasocial interactions, they may not affect all children
in the same way. As Wilson and Smith (1998) noted, children’s reactions to the
television “are mediated to some degree by certain characteristics they bring to the
viewing environment” (p. 551). An example of an individual difference variable that
may affect television’s impact is intelligence. Sprafkin, Gadow, and Abelman (1992)
discussed how gifted children are less apt to perceive television content as real when
compared to children of normal intelligence. Conversely, Van Evra (1998) stated that
children who have learning, behavioral, and emotional disorders usually view more
television, interpret the content as realistic, and are in turn more vulnerable to television’s
influences (p. 130). Perhaps, disadvantage is another variable that can affect the way
children interpret television. For example, Sprafldn, Gadow, and Dussault (1986) found
that emotionally disturbed children perceived television as real even when the study
controlled for intelligence (cited in Van Evra, 1998, p. 125).

One group of young viewers that may fit into this category of disadvantage is
abused children. Maltreated children are not only emotionally disturbed, but are more
likely to be disadvantaged in other ways. Niolon (1999) reported several ways in which
abused children are underprivileged. They are more likely to come from a low socio-
economic status (SES) family, a single-parent home, have witnessed a parent being
abused, had fewer fiiends during their childhood, had parents who were dissatisfied with
their marriage, and have received little affection from their father.

Since abused children may differ in their interpretation of television reality, the
prevalence and intensity of parasocial interactions may also manifest differently for this

special population. Theory and research supports this idea in three different ways. The

first reason is based on an attachment explanation. Cole and Leets (1999) have theorized
that peeple who have insecure attachments may be more likely to satisfy their relational
needs in different outlets, like parasocial activity (p. 497). Consistent with their thinking,
Cole and Leets (1999) found, when compared to secures and avoidants, anxious-
ambivalent adults were more likely to form parasocial relationships with their favorite
television characters. Since abused children tend to be classified with an attachment
pattern known as insecure-disorganized attachment, otherwise known as anxious-
ambivalent attachment (Carlson, Cicchetti, Barnett, & Braunwald, 1989), they may be
more likely to have parasocial interactions with their favorite television characters than
non-abused children.

The second reason is based on a depression explanation. Research reveals that
depression and parasocial activity are related. At least one study with adults showed that
when depression is not temporary in nature it is positively correlated with parasocial
interaction (Vogel, 1993, p. 65). Several scholars have discussed the rate of depression
among maltreated children. For example, Toth, Manly, and Cicchetti (1992) compared
physically abused children to neglected and non-maltreated children on depression rates.
The results indicated that children in the abuse sample had significantly higher depression
scores than children who were physically neglected and the control group (Toth, Manly,
& Cicchetti, 1992, p. 105). Similarly, an interview survey by F lisher et al. (1997) found
that 24% of a sample of children with a history of physical abuse were candidates for a
mood disorder, which included major depression (p. 128). Because they are more likely
to be depressed, abused children should be more likely to engage in parasocial

relationships than non-abused children.

The third reason is based on a functional alternative explanation. Studies with
adults have found that parasocial relationships may substitute or replace something else
to satisfy a specific need. For example, Levy (1979) found that individuals who were
lacking in social interaction had an increased chance of developing a parasocial
interaction with a television newscaster. Similar results were obtained by Yanof (1991).
Not surprisingly, several scholars have noted the difficulty abused children have with
social relationships. Feldman et a1. (1995) found that abused children received lower
adaptive functioning ratings by their teachers than a control group (p. 324). Additionally,
Salzinger, Feldman, Hammer, and Rosario (1993) established that abused children were
preferred less as social companions by their fellow peers than non-abused subjects (p.
175). This difficulty abused children have in forming relationships seems like a
reasonable explanation for why they may be more likely to have parasocial interactions
than non-abused children (Heineman, 1998).

In summary, the research undoubtedly has established that adults and children do
have parasocial interactions with characters on television (Choi, 1993; Cortez, 1991;
Grant, Guthrie, & Ball-Rokeach, 1991; Hoffner, 1996; Houlberg, 1984; Levy, 1979;
Perse, 1990; Perse & Rubin, 1988; Perse & Rubin, 1989; Rubin & Perse, 1987; Rubin,
Perse, & Powell, 1985; Turner, 1993; Yanof, 1991). Research indicates that these
parasocial relationships are more likely to occur in populations that are clinically
depressed, use television as a functional alternative, and have an anxious-ambivalent
attachment style (Cole & Leets, 1999; Levy, 1979; Vogel, 1993; Yanof, 1991).
Literature on abuse suggests that maltreated children are more likely than non-maltreated

children to exhibit all of these characteristics (Heineman, 1988; Wolfe, 1999). Therefore,

it is likely that abused children will engage in parasocial interactions more often than
their non—abused counterparts.
Present Study, Hypotheses, and Research Questions

To test this prediction, the present study replicated and extended Hoffner’s (1996)
investigation. Abused and non-abused children between the ages of 7 and 16 years old
were interviewed about their relationships with media characters, as well as their
motivations for such relationships. Based on the literature review above, it is anticipated
that differences in the prevalence of parasocial interactions would be observed in

maltreated and non-maltreated subjects. In specific, Hypothesis 1 states:

H1: Abused children will report parasocially interacting with their favorite characters on

television more often than non-abused children.

I In addition to prevalence, the present study is interested in attributes of characters
that influence parasocial interactions among abused and non-abused children. Hoffner
(1996) developed a scale that measured what character attributes attract the child to
interact with a particular character. To date, research has only investigated non-abused

children; therefore, the following question is asked:

RQl: What character attributes attract abused and non—abused children to their favorite

television character?

Lastly, this study is not only interested in what characters children are drawn to
when they participate in a parasocial interaction, but what needs these relationships are
filling for children. Past research has only presented results on the prevalence of
children’s parasocial relationships and the attributes that attract them to media personas.
In other words, what are the reasons children state for engaging in parasocial interactions.
To date, no published studies have investigated the reasons why children have these
imaginary relationships with television characters. Therefore, this examination calls for

the following question:

RQ2: What motivates abused and non-abused children to have parasocial relationships?

METHOD

Participants

A total of 60 children (M = 9.75, SD = 2.07, range = 7 - 16) were sampled for this
investigation, 30 abused and 30 non-abused. There were 25 males and 35 females, with
the majority of the sample between the ages of 7 and 12. Of the total sample, 38% of the
children were White, 33% were Black, 8% were Hispanic, 5% were Asian, 2% were
American Indian, and 13% labeled themselves as other. All children were from low SES
families. Each child was required to receive permission from their parent or guardian and
their own assent prior to participation.

The goal of this study was to obtain a sample of abused children identified by
trained professionals, rather than trying to identify them through scales that correlate with
abuse. Given the Obstacles involved in finding an abused sample of children to
participate in research, it seemed appropriate to allow a non-random sample. Therefore, a
convienence sample of abused children participated through a Human Service Agency in
a Midwestern city. Non-abused children were obtained through an elementary school in
the same city‘.

Procedure

The interview covered six groups of questions: parasocial activity, prevalence of
parasocial interactions, character attributes, motivations for viewing, affect, and
demographics. Non-abused children and abused children received the same

questionnaire. Following Hoffner’s (1996) protocol, most of the non-abused children

10

were interviewed individually in a quiet room at the children’s school. A researcher was
trained on how to administer the questionnairez. Each question was read aloud and the
child was instructed to respond orally or by pointing to their chosen response from a
larger printed version of the scales. Older children, in the non-abused sample, completed
the questionnaire through a group interview; this was necessary to keep interview time to
a minimum.

Abused children were all interviewed individually by a volunteer or employee of
the Human Service Agency. These individuals were also trained by the university
researchers on how to administer the questionnaire. Again, as done with the non-abused
sample, each question was read aloud and the child was instructed to respond orally or by
pointing to their chosen response from a larger printed version of the scales.

After answering all of the survey questions, each child was debriefed about the
contents of the questionnaire. The survey administrators explained that the questions
tried to determine what TV characters children like, why they like them, and how
different types of children respond to different characters on television. Two forms of the
survey were constructed by randomly ordering the parasocial interaction and character
attributes scales; this helped to minimize order effects.

Measures

In order to answer the questions on the survey each child went through a carefully
constructed series of questions to determine their favorite television character. Children
started by indicating how often they view television, their three favorite television shows,
their three favorite television characters and the show in which they appear, and then

finally answered the question, “Who is your favorite TV character? ” The children were

11

then instructed to refer to that favorite television character when they answered survey
questions.

Parasocial interaction. Children’s illusionary inter-actions with their favorite
television characters was measured with the lO-item Parasocial Interaction Scale (PSI)
adapted by Hoffner (1996), originally developed by Rubin and Perse (1987). Reliability
was reported at a Cronbach’s a of .70 (Hoffner, 1996). Because of this marginally
acceptable alpha level, additional questions were added to the PSI scale. Examples of
new items include, “If something happens to [ insert name of favorite character] I feel
bad ” and “If [insert name of favorite character] lived in my neighborhood we would be
friends. ” In addition, one item was modified because the content was confusing, “I see
[insert name of favorite character] as a natural, down-to-earth person ", was changed to

“[ insert name of favorite character] seems like he/she would be easy to talk to. ” Lastly,
“I find [insert name of favorite character] to be attractive” was removed from the scale
because an overlapping item occurred in the character attribute scale.

Research also confirms that the lO-item scale and longer versions of the scale are
internally valid. Discriminability, a form of construct validation, was determined because
first, the scale is correlated to related constructs. For example, Rubin and Perse (1987)
found that the PSI scale was correlated to voyeurism and escapist relaxation. Perse and
Rubin (1987) also found that the scale was uncorrelated to unrelated constructs, like
passing the time and co-viewing distraction. In addition, the scale has shown evidence of
convergence because similar results have been obtained from a number of studies that
have utilized the scales (Auter, 1992). Responses to the items were answered on an

ordinal 5-point Likert scale, from “disagree a lot ” to “agree a lot ” Higher scores on the

12

scale indicate a greater parasocial interaction with the character. These items were
summed and averaged to form a single parasocial interaction scale (M = 3.91, SD = .72,
range = 1.71 - 5). Confirmatory factor analysis of these items suggested a
unidimensional scale with a standardized item a = .793 (Hunter & Gerbing, 1982) (see
Appendix A for complete scale).

Prevalence. In addition to the PSI scale, other questions were used to determine
the prevalence of these interactions between children and their favorite television
characters. First, a direct yes or no question was asked to determine whether the subjects
ever participated in these interactions, “Do you have times you think about [insert name
of favorite character] and talk back and forth with him/her in your head? " In addition, a
four item scale was constructed with several items like, “How often do you talk back and
forth with [insert name of favorite character] in your head? ” Response options were on
a 4—point Likert scale ranging from “never” to “all of the time " (M = 2.02, SD = .85,
range = .5 - 3.75, a: .81).

Character traits. Each subject also completed a character attribute scale
developed by Hoffner (1996). Sixteen ordinal items were used to measure children’s
perceptions of their favorite character’s personal traits. This study looked at the same
categories: attractiveness, strength, humor, intelligence, and social behavior. The
children were asked to agree or disagree with each item on a 5-point Likert scale,
pertaining to their favorite character on television.

Hoffner (1996) conducted a principal axis factor analysis on her scale. Each of
the items received moderate to high levels of reliability in the study. Cronbach’s alpha

for each trait were as follows, social behavior (kind, helpful, caring, mean, violent, and

13

selfish) or = .90, attractiveness (handsome/pretty, good-looking, and somewhat ugly) 01 =
.89, humor (funny, and says and does funny things) or = .81, strength (physically strong,
and has more strength than most people) or = .79, intelligence (smart, good at figuring
things out, and dumb) (I = .87. The scale also showed face validity because of the
questions asked. For example, to measure if the child perceives their favorite character to
be strong the question read, “[ insert name of favorite character] is physically strong. ”
For this investigation, reliability estimates were adequate for the subscales; social
behavior (M = 4.05, SD = .88, range = 2- 5, a = .75), attractiveness (M = 3.40, SD = 1.42,
range = l — 5,01 = .82)4, humor (M= 4.5, SD = .98, range = 1 — 5, a: .76), strength (M=
3.13, SD = 1.41, range = 1 - 5, a: .79), and intelligence (M= 3.94, SD = 1.20, range = 1
— 5, a = .68) (see Appendix B for complete scale).

Motivation. Two open-ended questions were asked in the interview to explore
what motivates children to engage in these relationships. Open-ended, items were, “What
do you usually talk about? ”, and “My do you talk back and forth with [insert name of
favorite character] in your head? ”

Affect. These questions assessed how happy children felt when engaging in
parasocial interactions. Closed-ended items were, “How sad/happy do you feel when you
talk back and forth with [insert name of favorite character] in your head? ” Response
options were on a 5-point Likert scale ranging from “Not at all Sad/Happy " to “Very
Very Sad/Happy”. For the sad items (M = 1.5, SD = 1, range = 1 — 5). For the happy
items (M= 3.62, SD = 1.67, range = l -5).

Demographics. Lastly, demographic information for each subject was collected,

including gender, age, race, and the parent’s occupation. Since abused children may not

14

live with their parents, the abused sample questionnaire did not include questions
regarding the children’s parental occupations. Information from the Human Service

Agency confirmed that the children were primarily from low SES households.

15

RESULTS

Data Analyses

For all interval-level data, unweighted-means analyses Of variance were
conducted with abuse status (no, yes) and gender (boys, girls) as factors. For all
categorical level data, multidimensional contingency tables were formed using the same
two factors. Log-linear analyses were executed on the frequencies of responses for such
data. This statistical procedure provides a parallel to the traditional AVOVA for nominal
level data (Marascuilo & Levin, 1983). Below, only means or percentages with no letter
in common in their subscripts are significantly different at the p < .05 level. Results are
presented below by each hypothesis and/or research question and not in the order the
measures were asked on the questionnaire.
Parasocial Interaction

Hypothesis I predicted that abused children would parasocially interact with their
favorite television characters more often than non-abused children. This hypothesis was
tested in 3 different ways. First, children answered questions from the parasocial
interaction scale. Participants indicated their agreement with statements such as, “[ insert
name of favorite character] makes me feel comfortable, like I ’m with a friend. ”
Responses ranged from, (1) disagree a lot to (5) agree a lot. An analysis of variance on
the children’s summed parasocial interaction scale scores was calculated. No significant
differences emerged by abuse status or gender. Exactly half of the children (50%) in the

sample reported parasocially interacting with their favorite television characters.

16

Second, children were asked the following yes or no question: “Do you have
times that you think about [insert name of favorite character] and talk back and forth
with him/her in your head? ” A log-linear analysis of yes/no responses revealed a
significant effect for abuse status, G2 (N = 58) = 9.80, p < .01, V* = .12. Contrary to
expectation, a higher proportion of non-abused children reported parasocially interacting
with their favorite television character (72%,) than did abused children (31%.)

Third, a series of scaled questions were summed to estimate the prevalence or
frequency of these intrapersonal relationships. To illustrate one question, the children
were asked, "In a typical day, like today, how frequently do you imagine talking with
your favorite character? ” Responses Options ranged from (1) never to (4) all of the time.
The three remaining questions were similarly worded and featured the same response
options. An analysis of variance on the summed prevalence items revealed a significant
main effect for abuse status, F (1, 47) = 3.91, p < .05, n2 = .07. Again, non-abused
children were significantly more likely to report parasocially interacting with their
favorite television character (M = 2.241,) than were abused children (M = 1.723). No other
significant effects emerged here. Overall, the results are directly opposite to the
prediction offered in Hypothesis 1.

Character Attributes

Research question 1 inquired about what character attributes attract abused and
non-abused children to their favorite television characters. As described above, several
character attributes were assessed. For each attribute, the response options ranged from
(1) disagree 0 lot to (5) agree a lot. The first was social behavior. As you may recall,

the children were asked questions about how mean, violent, selfish, kind, helpful, and

17

caring they saw their favorite character. An analysis of variance of these ratings revealed
a significant main effect for gender, F (1, 55) = 5.70, p < .05, n2 = .09. Girls were
significantly more likely to perceive their favorite character as having socially acceptable
behavior (M = 4.281,) than were boys (M = 3.74,.) No other main effects or interactions
were significant.

The second character attribute measured was attractiveness. The children were
asked how handsome/pretty and good—looking they found their favorite character. An
analysis of variance was performed on these ratings; no significant effects emerged in
this analysis. Over a third (3 8%) of the sample saw their favorite character as attractive.

The third attribute analyzed was humor. Children indicated how funny their
character is and whether their favorite character says and does funny things. An analysis
of variance revealed no significant differences. More than three quarters of the sample
(86%) found their character to be humorous.

The fourth character attribute assessed was strength. The participants answered
questions that asked if they agreed that their favorite character was physically strong and
has more strength than most people. An analysis of variance revealed a significant main
effect for gender, F (1, 55) = 5.35, p < .05, 172 = .08. For these results, boys were more
likely to see their favorite character as strong (M = 3.601,) than were girls (M = 2.78,).
There were no other significant main effects or interactions for strength detected.

The fifth attribute analyzed was intelligence. Children indicated if they thought
their favorite television character was smart, and good at figuring things out. No
significant effects emerged for this factor. Over half of the sample (66%) saw their

favorite character as intelligent.

18

Motivation

Research question 2 inquired about abused and non-abused children’s motivations
for engaging in parasocial relationships. Two open-ended questions were asked to
address this research question. First, the interviewer asked “Do you have times that you
think about [insert name of favorite character] and talk back and forth with him/her in
your head? " If the child answered yes, the interviewer then inquired, “Mat do you
usually talk about? ” This question aimed to elicit conversational topics that the children
recalled engaging in with their favorite television character. Responses to this question
were coded by indicating the presence or absence in the child’s response of liking the
character (i.e., “like watching her show”, “funny stuff”), episode information (i.e., “we

I) H

talk about the show , what else will he show on T V ”), interaction with the character
(i.e., “stupid stuff—how the day went ", “strength and sports ”), and other television
shows (i.e., “cartoons ”, “TV shows "). Coding categories for these and all other open-
ended responses were developed from the responses given by the children. Several
coding categories were collapsed to represent the children’s responses on a more global
level. The actual coding procedure was then performed by two independent judges and
disagreements were resolved through discussion with the primary researcher. Using
Scott’s 1r (1955), intercoder reliability was 97% for character liking, 96% for episode
information, 90% for interaction with the character, and 100% for other television shows.
Log-linear analyses were executed on each of the conversational topics. First,
liking the character was examined and no significant differences emerged for this

variable. Overall, about a third (30%) of the sample indicated liking the character as a

conversation topic.

19

Next, a log-linear analysis of episode information revealed a significant effect for
gender, 02 (N = 30) = 3.90, p < .05, V* = .09. Girls were significantly more likely to talk
about episode information (42%,) than were boys (9%,)6. No other effects for episode
information were significant.

Interaction with the character was also examined as a possible topic of
conversation. No significant effects emerged from these analyses. Just under half (47%)
of the sample indicated interaction as a topic of conversation.

Lastly, the conversation topic of other TV shows was analyzed. A log-linear
analysis did reveal a significant effect for gender, (P (N = 30) = 4.08, p < .05, V* = .10.
Boys were significantly more likely to talk about other television shows (18%,) than
were girls (0%,). This analysis should be interpreted with caution because a total of two
boys reported this answer. No other significant effects emerged in this analysis.

The second open-ended question asked, “Why do you talk back and forth with
[insert name of favorite character] in your head? " Responses to this question were
coded by indicating the presence or absence of positive affect (i.e., “cause I like talking
to her”, “because sometimes he is really really funny") and help (i.e., “to help him ",
“get over anger”). Intercoder reliability was perfect for both categories at 100%.

A log-linear analysis of these frequencies revealed a significant interaction effect
for positive affect, (P (N = 35) = 5.82, p = .01, V* = .12. Non-abused girls expressed
positive affect as a reason for their parasocial interactions more often (7 5%,) than did
abused girls (17%,). A main effect was detected also for gender, (P (N = 35) = 4.02, p <

.05, V* = .08. Qualified by the interaction effect, girls were more likely to talk with their

favorite character for affective reasons (59%b) than were boys (23%,).

20

A log-linear analysis of the help category also revealed a significant interaction
effect, 02 (N = 35) = 7.48, p < .01, V“ = .15. In this analysis, abused girls saw help as a
reason for interacting with their favorite character more often (50%,) than did non-abused
girls (0%,). No other effects emerged in this analysis.
Affect

The interviewer asked two closed-ended items to assess the enjoyment children
received from engaging in these interactions. The interviewer first asked, “How sad do
you feel when you talk back and forth with [insert name of favorite character] in your
head? ” Responses ranged from (1) not at all sad to (5) very very sad. An analysis of
variance of these ratings revealed no significant effects. About a quarter (27%) of the
sample reported being at least a little bit sad when interacting with their favorite
character.

The interviewer then asked, “How happy do you feel when you talk back and
forth with [insert name of favorite character] in your head? " Responses ranged from (1)
not at all happy to (5) very very happy. An analysis of variance of these ratings also
revealed no significant findings. Approximately three quarters (79%) of the sample
indicated at least feeling a little bit happy when interacting with their favorite character.
Correlations Among Measures

Final analyses were conducted by computing partial correlations for all variables
used in the study controlling for gender and abuse status (see Table 4). First, the PSI
scale was positively correlated with social behavior (r = .39, p < .01), attractiveness (r =
.42, p < .01), and intelligence (r = .56, p < .01). Prevalence, on the other hand, was

correlated with the other two attributes, humor (r = -.3 1 , p < .05) and strength (r = .28, p

21

< .05). These partial correlations suggest that children are in fact interested in characters
with these specific attributes, except humor which resulted in a negative correlation.
Second, affect was negatively correlated with help (r = -.38, p < .05). This finding
proposes that children either find these relationships fun or taxing. Third, prevalence was
positively correlated with happiness (r = .65, p < .01). This result suggests that most

individuals who engage in these relationships find them to be fun.

22

DISCUSSION

Generally, the findings from this study reveal that abused children parasocially
interact with their favorite television characters differently than non-abused children.
Unexpectedly, abused children were found to engage in these relationships less than their
non-abused counterparts. Although the first hypothesis was unsupported, convergent
findings with Hoffner (1996) and other identification literature were obtained in the
analyses of character attributes. The most notable consistency was that boys are
interested in strong characters. This investigation also attempted to take a first look at the
motivations behind children’s parasocial interactions. At least for girls, these analyses
revealed positive affect and help as the most important reasons for engaging in parasocial
relationships.

In terms of specific predictions, it was expected in Hypothesis 1 that abused
children would parasocially interact with their favorite television character more often
than non-abused children. There was no support for this hypothesis. In fact, the opposite
pattern emerged where non-abused children were engaging in these relationships more
often; however, these findings are hard to ignore given the convergent results on two
analyses. Approximately three quarters of the non-abused sample reported participating
in these illusionary interactions with their favorite character, compared to only a third of
the abused sample.

Why do abused children report participating in parasocial interactions less often

than non-abused children? There are at least four possible explanations for this finding.

23

First, Salzinger et a1. (1993) explains that abused children tend to be more aggressive,
show fewer leadership skills, and less cooperation when playing with other children.
Typically, these offensive social acts are responded to negatively by other children and
the abused child is rejected by their peers (Salzinger et al., 1993). This rejection may
caution abused children from engaging in any type of friendship, be it imaginary as in a
parasocial context or a real relationship.

Second, because abused children are isolated from social situations by their
parents, they are less likely to develop appropriate social schemas for interacting with
other children (Coleman, Butcher, & Carson, 1980; DiMaggio, 1997; White & Watt,
1973; Wolfe, 1999). Therefore, when they are presented with a social situation, they may
alienate themselves from creating fiiendships because they don’t know how to develop
them. This alienation may include any type of friendship, including parasocial
companions.

A third possibility is that abused children may fear punishment from their parents
if they engage in friendships. Since abused children are typically isolated from social
relationships, they may think that if they engage in friendships their parents will punish
them (Wolfe, 1999). A parasocial relationship may seem like a safe alternative to the
observer, but the child may not be cognitively aware of that distinction. Their
interpretation of any friendship may be associated with fear; therefore, the child may
avoid any type of social relationship including parasocial interactions.

Finally, the abused sample may have included extreme cases of abuse because the
sample was obtained from programs that are designed to care for maltreated children.

Considering the three reasons stated above, this population may have been more likely to

24

avoid the possibility of rejection in an illusionary friendship than less severely abused
children. To this end, a Tukey a post hoc analysis was conducted to see what trends
emerged when looking at high SES, low SES, and abused children. The PS1 scale
revealed no significant differences between the means of these three groups, but the trend
did suggest that high SES children (M = 3.67) interact in these relationships less often
than low SES (M = 3.89) and abused children (M = 3.93). The alternative PSI
measurement used in this study which summed four prevalence questions again revealed
different findings compared to the PSI scale. This analysis suggested that high SES (M =
1.71) and abused children (M = 1.72) interact with television characters at about the same
level and low SES children (M = 2.24) are significantly more likely to engage in these
relationships than the other two groups.

Since the abused sample was in the care of organizations designed to help the
children, this group may consume similar levels of television to the high SES group and
therefore have similar prevalence ratings because of the amount of exposure. Although
this is a viable hypothesis, mean comparisons showed a trend of abused children (M =
3.85) as the highest consumers of television compared to low income (M = 3.15) and high
income (M = 2.42) children in the study7. Although the explanation offered here is a
possibility, it seems that the results are again inconsistent with each other and call for
future research to confirm a specific pattern.

Overall, it seems that previous research and the current findings indicate that
abused children are unlike non-abused children when it comes to parasocial interactions.
Other investigations like Hoffiter (1996) have found that children do engage in these

relationships. This examination confirms the same findings; 75% of the non-abused

25

sample reported engaging in these interactions. Considering this information, it seems
that abused children are again showing that they have incurred negative effects from the
maltreatment of their parents because they avoid these parasocial relationships. Future
research should investigate this same question with a larger sample to see if similar
findings are revealed.

The first research question was interested in what types of character attributes
attract abused and non-abused children to their favorite TV characters. This research
question found two significant gender effects. First, boys were more likely to see their
favorite character as strong. This finding is consistent with previous research in
identification and parasocial interaction. Typically, children are attracted to male
characters and specifically boys like characters that are physically strong (Eisenstock,
1984; Hoffner, 1996; Lonial & Auken, 1985; Miller & Reeves, 1976; Reeves &
Greenberg, 1977; Reeves & Lometti, 1979; Reeves & Miller, 1978). Boys most likely
select these types of characters because men in American society are expected to be
strong individuals emotionally and physically (Gray, 1992). These television characters
display a potent version of that strength and understandably become someone to look up
to.

Second, girls were more likely to view their favorite character as having socially
acceptable behavior; for example, they chose characters that were more kind and helpful.
Hoffner (1996) found a similar finding where boys and girls were more likely to identify
female characters as engaging in more positive social behaviors than male characters.
These parasocial interactions may be an opportunity for girls to learn scripts or schemas

related to social behavior.

26

Other character attributes, although not significant by gender or abuse status, were
also important attributes in the favorite television character for the child. Nearly a third
of the children saw their favorite character as attractive, three quarters as humorous, and
over half as intelligent. Previous research has confirmed that girls typically identify or
parasocially interact with attractive characters (Hoffner, 1996; Miller & Reeves, 1976;
Reeves & Greenberg, 1977; Reeves & Lometti, 1979; Reeves & Miller, 197 8). In
addition, other studies have suggested intelligence and humor as attributes that encourage
identification with television characters (Reeves & Greenberg, 1977; Reeves & Lometti,
1979; Reeves & Miller, 1978).

On the whole, it is clear that children do interact with characters that possess these
attributes. Since abused children are less likely to engage in these interactions, they don’t
benefit from some of the lessons that these characters may teach them. Future research
should investigate other attributes that may be more interesting to abused youngsters.
Perhaps, this scale was not inclusive enough to identify attributes that abused children are
more likely to seek out.

The second research question was interested in the motivations that abused and
non-abused children have for interacting with their favorite television characters. These
analyses contributed new information in the investigation of parasocial relationships.
Open-ended questions were coded in order to evaluate what types of topics these
interactions typically involve and why children have these relationships. By and large,
these analyses revealed some interesting findings. In terms of conversational topics, girls
were more likely to discuss episode information with their favorite TV character than

were boys. One popular stereotype of the sexes is that women like storylines and men

27

fixate on action sequences (Gray, 1992). Conceivably, girls may be using illusionary
companions as a way to learn more about these stereotypical behaviors.

As for why children engage in these relationships, two significant effects were
revealed. First, non-abused girls were more likely to express positive affect as a reason
for their illusionary conversations than were abused girls. Generally speaking, these
relationships seem to be fun for non-abused girls because a similar trend was suggested in
analyses of the affect questions. Children were asked how happy or sad they felt as a
result of interacting with their favorite character. These results indicated that non-abused
girls enjoyed these friendships; they tended to be happier than other children when
involved in these parasocial relationships.

The results also suggested that abused girls were the only group who engaged in
these relationships because they needed or wanted to give help. Again, a trend from the
affect analyses suggested that abused children were sadder than non-abused children
when interacting with their favorite character. Therefore, engaging in these interactions
for help reasons may be taxing on the individual rather than fun.

Additionally, some research has reported what types of abused children tend to
seek help or receive more help from governmental agencies. Typically, these children
have been abused by their parents or foster parents, have had multiple incidents of
maltreatment, have had suicidal feelings, and have either had a medical check-up or have
been in medical care for their abuse (Saunders, Resnick, Hoberrnan, & Blum, 1994;
United States Department of Health and Human Services, 2000). This information
suggests that children who have been severely abused may be more likely to turn to

characters for help or to give their help. Therefore, this parasocial relationship is not an

28

alternative to friendships like it may operate for non-abused children, but a functional
relationship where tough problems are addressed. Perhaps, if a larger sample of abused
children was collected, an analysis could be computed for the severity of abuse that
children incur to understand why children engage in these relationships across different
levels of abuse.

As well, the literature suggests, abused children not only have problems in their
household, but they also have a hard time making friends (Salzinger et al., 1993; Wolfe,
1999). It is understandable that when they are presented with an opportunity to express
themselves or to converse with someone that seems like an interested friend, they may
use that time to take care of troubling problems or to see if that person needs someone
like they do. Perhaps, they offer help because they know children at school do not help
them when they are in need of companionship.

From a methodological point of view, there were several limitations of the study.
First, it was exploratory. This examination was not only a first attempt looking at abused
children and parasocial relationships, but also the motivations behind these interactions.
Given the results obtained from this study, future research should make considerations
for the types of predictions associated with parasocial interactions and abused children.
In addition, modifications to the motivation questions would help capture different
reasons why abused children, in particular, may be turning to these illusionary
companions.

Second, there were several obstacles to obtaining a sufficient abused sample.
Data collection was extended for one full year in order to collect 30 abused sample

interviews. Several organizations that agreed to participate did not anticipate a lack of

29

interest by their volunteers or interns and had fewer children within the required age
range than they originally estimated. Consequently, many participating areas were not
able to produce any data. This small sample size inevitably lowered the power of
statistical analyses for this investigation. Given this, a larger sample size would offer
much more confidence behind the interpretations offered.

Third, considering the barriers related to data collection, the primary researchers
were not able to maintain the level of control typically exhibited in experimental
situations for the abused sample. This data was collected by on site social workers,
volunteers, or interns and inevitably some data was not collected. For the most part,
uncollected data consisted of questions from the prevalence scale, affect section, and the
open-ended motivation questions. Future research should attempt to have more control
over the collection of this data.

Fourth and finally, all participants used in the data analyses were from low SES
households. This decision was made because abused children typically come from low
SES homes. Although this variable was constant in the analyses, it is important to note
that Hoffiier (1996) used a sample that was from a more affluent area. Perhaps, some of
the differences between this study and Hoffner (1996) are due to the fact that this study
involved low SES children.

Lastly, a few important contributions to parasocial research should be highlighted
from this investigation. Future research should consider using the new PS1 scale
questions added in this examination. Although these questions may be more appropriate
for children, they did increase the reliability of the scale (Hoffner (1996) or = .70; this

investigation a = .79). Also, a new measure of prevalence was established in these

30

 

 

analyses. This scale was very reliable and produced consistent findings with a similar
yes/no question. The use of this new scale could help increase the validity of parasocial
interaction if consistent findings were established across similar investigations and within
examinations.

Overall, this study reveals some compelling information concerning abused
children and parasocial relationships. Although the results are tenuous considering the
sample composition, this study can guide future research investigating this fragile
population. From this examination, the results propose that abused children may use
these parasocial friendships as functional activities where they can attempt to solve some
of their problems. This study suggests that even a fun activity like television viewing
may be a burdened action for this particular group of children. On the whole, these
results are only a starting point to understanding more about television effects on
disadvantaged populations. It is essential that this trend continues because understanding
how individual differences can change the relationship between content and viewer can

not only be applicable to society, but guide theory in the future.

31

REFERENCES

Auter, P. J. (1992). TV that talks back: An experimental validation of a parasocial
interaction scale. Journal of Broadcasting and Electronic Media, 36, 173-181.

Carlson, V., Cicchetti, D., Barnett, D., & Braunwald, K. (1989). Disorganized/disoriented
attachment relationships in maltreated infants. Developmental Psychology, 25,
382-393.

Choi, Y. (1993). Process of parasocial interaction in local television news watching
(Doctoral dissertation, The University of Florida, 1993). Dissertation Abstracts
International, 55, 9-A.

Cole, T., & Leets, L. (1999). Attachment styles and intimate television viewing:
Insecurely forming relationships in a parasocial way. Journal of Social and
Personal Relationships, 16, 495-51 1.

Coleman, J. C., Butcher, J. N., & Carson, R. C. (1980). Abnormal psychology and
modern life. London, England: Scott, Foresman and Company.

Cortez, C. A. (1991). Mediated interpersonal communication: The role of attraction and
perceived homophily in the development of parasocial relationships (Doctoral
dissertation, The University of Iowa, 1991). Dissertation Abstracts International,
53, 8-A.

DiMaggio, P. (1997). Culture and cognition. Annual Review of Sociology, 23, 263-287.

Eisenstock, B. (1984). Sex-role differences in children’s identification with
counterstereotypical televised portrayals. Sex Roles, 10, 417-430.

Feldman, R. S., Salzinger, 8., Rosario, M., Alvarado, L., Caraballo, L., & Hammer, M.
(1995). Parent, teacher, and peer ratings of physically abused and nonmaltreated
children’s behavior. Journal of Abnormal Child Psychology, 23, 317-334.

Femie, D. E. (1981). Ordinary and extraordinary people: Children’s understanding of
television and real life models. In H. Kelly & H. Gardner (Eds.), New directions
for child development (pp. 47-5 8). San Francisco, CA: J ossey-Bass.

Flisher, A. J ., Kramer, R. A., Hoven, C. W., Greenwald, S., Alegria, M., Bird, H. R.,
Canino, G., Connell, R., & Moore, R. E. (1997). Psychosocial characteristics of

physically abused children and adolescents. Journal of the American Academy of
Child and Adolescent Psychiatry, 36, 123-131.

32

Gerbner, G., Gross, L., Morgan, M., & Signorielli, N. (1994). Growing up with
television: The cultivation perspective. In J. Bryant & D. Zilhnan (Eds), Media
effects: Advances in theory and research (pp. 17-42). Hillsdale, NJ: Erlbaum.

Grant, A. E., Guthrie, K. K., & Ball-Rokeach, S. J. (1991). Television shopping: A
media system dependency perspective. Communication Research, 18, 773-798.

Gray, J. (1992). Men are from mars, women are from venus: A practical guide for
improving communication and getting what you want in your relationships. New
York: Harper Collins.

Harrison, K. (1997). Does interpersonal attraction to thin media personalities promote
eating disorders? Journal of Broadcasting & Electronic Media, 41 , 478-500.

Heineman, T. V. (1998). The abused child. New York: The Guilford Press.

Hoffner, C. (1996). Children’s wishful identification and parasocial interaction with
favorite television characters. Journal of Broadcasting & Electronic Media, 40,
389-402.

Horton, D., & Wohl, R. R. (1956). Mass communication and para-social interaction.
Psychiatry, 19, 215-229.

Houlberg, R. (1984). Local television news audience and the para-social interaction.
Journal of Broadcasting, 28, 423-429.

Huesmann, L. R., Lagerspectz, K., & Eron, L. D. (1984). Intervening variables in the TV
violence-aggression relation: Evidence from two countries. Developmental
Psychology, 20, 746-75.

Hunter, J. E., & Gerbing, D. W. (1982). Unidimensional measurement, second-order
factor analysis, and causal models. Research in Organizational Behavior, 4, 267-
320.

Lemish, D. (1985). Soap opera viewing in college: A naturalistic inquiry. Journal of
Broadcasting & Electronic Media, 29, 275-293.

Levy, M. R. (1979). Watching TV news as para-social interaction. Journal of
Broadcasting, 23, 69-80.

Lonial, S. C., & Auken, S. V. (1985). Wishful identification with fictional characters: An
assessment of the implications of gender in message dissemination to children.

Journal of Advertising, 15, 4-42.

Maccoby, E. E., & Wilson, W. C. (1957). Identification and observational leaming from
films. Journal of Abnormal and Social Psychology, 55, 76-87.

33

Marascuilo, L. A., & Levin, J. R. (1983). Multivariate statistics in the social sciences: A
researcher's guide. Monterey, CA: Brooks/Cole.

Miller, M., & Reeves, B. (1976). Dramatic TV content and children’s sex-role
stereotypes. Journal of Broadcasting, 20, 35-50.

Niolon, R. (1999). Child abuse and domestic violence. Retrieved January 29, 2002, from
http://www.psychpage.com/learning/library/counseling/childvio.htrnl

Noble, G. (1975). Children in front of the small screen. Beverly Hills, CA: Sage.

Perse, E. M. (1990). Media involvement and local news effects. Journal of
Broadcasting & Electronic Media, 34, 17-36.

Perse, E. M., & Rubin, R. B. (1988). Audience activity and satisfaction with favorite
television soap opera. Journalism Quarterly, 65, 368-375.

Perse, E. M., & Rubin, R. B. (1989). Attribution in social and parasocial relationships.
Communication Research, 16, 59-77.

Reeves, B., & Greenberg, B. S. (1977). Children’s perceptions of television characters.
Human Communication Research, 3, 113-127.

Reeves, B., & Lometti, G. E. (1979). The dimensional structure of children’s perceptions
of television characters: A replication. Human Communication Research, 5, 247-
256. ‘

Reeves, 8., & Miller, M. M. (1978). A multidimensional measure of children’s
identification with television characters. Journal of Broadcasting, 22, 71-86.

Rubin, A. M., & Perse, E. M. (1987). Audience activity and soap opera involvement.
Human Communication Research, 14, 246-268.

Rubin, A. M., Perse, E. M., & Powell, R. A. (1985). Loneliness, parasocial interaction,
and local television news viewing. Human Communication Research, 12, 155-
180.

Salzinger, S., Feldman, R. S., Hammer, M., & Rosario, M. (1993). The effects of
physical abuse on children’s social relationships. Child Development, 64, 169-
187.

Saunders, S. M., Resnick, M. D., Hobennan, H., M., & Blum, R. W. (1994). Formal help
seeking behavior of adolescents identifying themselves as having mental health
problems. Journal of the American Academy of Child & Adolescent Psychiatry,
33, 718-728.

34

Scott, W. (1955). Reliability of content analysis: The case of nominal scale coding.
Public Opinion Quarterly, 17, 321-325.

Sherman, A. (1997). Poverty matters. Retrieved March 20, 2002, from
http://www.childrensdefense.org

Sprafldn, J ., Gadow, K. D., & Abelman, R. (1992). Television and the exceptional child:
A forgotten audience. Hillsdale, NJ: Erlbaum.

Toth, S. L., Manly, J. T., & Cicchetti, D. (1992). Child maltreatment and vulnerability to
depression. Development and Psychopathology, 4, 97-112.

Turner, J. R. (1993). Interpersonal and psychological predictors of parasocial interaction
with different television performers. Communication Quarterly, 41, 443-453.

United States Department of Health and Human Services. (2000). Child Maltreatment.
Washington, DC: US. Government Printing Office.

Van Evra, J. (1998). Television and child development. Mahwah, NJ: Erlbaum.

Vogel, R. D. (1993). Loneliness and depression and their relation to parasocial
interaction (Doctoral dissertation, The University of Tennessee, 1993).
Dissertation Abstracts International, 54, 325.

Weiner, I. B., & Elkind, D. (1972). Child development: A core approach. New York:
John Wiley & Sons.

White, R. W., & Watt, N. F. (1973). The abnormal personality. New York: The Ronald
Press Company.

Wilson, B. J ., & Smith, S. L. (1998). Children’s responses to emotional portrayals on
television. In P. Anderson & L. Guerrero (Eds.), Handbook of communication and
emotion: Theory, application, and contexts (pp. 533-569). New York: Academic
Press.

Wolfe, D. A. (1999). Child abuse: Implications for child development and
psychopathology. Thousand Oaks: Sage.

Yanof, D. S. (1991). The para-social and interpersonal relationships of heavy, light, and
nonviewers of daytime television serials (Doctoral dissertation, California School
of Professional Psychology at Berkley/Alameda, 1991). Dissertation Abstracts
International, 52, l3-A.

35

FOOTNOTES

lNon-abused children were obtained through a large scale education-based
longitudinal study which assesses how children learn. The education research team
originally obtained the participating schools by randomly sampling low SES school
districts in a metropolitan area. The schools that were available to this investigation were
selected based on their enthusiasm during the education project. It was important to
obtain children from low SES families because, as Niolon (1999) stated, abused children
are more likely to be from low SES families. Therefore, in order to control for income
level, the participants all come from the same level of wealth. However, this factor may
seem troubling because it is likely that abused children would be in the non-abused
sample. From a report by The Children’s Defense Fund, children from low SES families
are 2.8 times more likely to have inadequate parental care than wealthier children
(Sherman, 1997). Even though there is a risk of abused children being in our non-abused
sample, that risk is minimal because it should lie within sampling error of the obtained
results.

2University researchers trained all survey interviewers. For both the abused and
non-abused sample, a researcher explained all the facets of the questionnaire to the
specified interviewers (Parasocial Interaction, Prevalence, Character Attributes,
Motivations, Affect, and Demographics). The instructions were explained in detail,
making sure each interviewer understood how to ask the scale questions to solicit the

child’s spontaneous answers. They were also sensitized to stop immediately if the child

36

said ‘STOP’ at any time during the interview process. Lastly, the interviewers had full
training for the debriefing of subjects and were informed of all necessary project
information if the subjects had any questions regarding the research study. After these in
depth training procedures, the interviewer practiced giving the questionnaire using the
researcher as a subject or a co-worker from their office.

3 Please note that all alpha levels reported for this investigation are standardized
item alpha, except for the few that are specifically labeled as Cronbach’s alpha.

4The attractiveness and intelligence subscales were slightly modified for this
investigation because they both had problem items that considerably lowered their
reliability levels. This investigation dropped the item, “[ insert name of favorite
character] is somewhat ugly” for the attractiveness subscale. The statement “[ insert
name of favorite character] is dumb " was dropped for the intelligence subscale. Even
though these items were unreliable the results did not change significantly when the items
were eliminated from analyses.

5 Participants were classified as high on the parasocial interaction scale with an
average score of 4 or higher, meaning they at least “agreed a little” to each scale item.
Proportions for the character attributes and the affect scaled items were determined in the
same fashion. Character attributes were high when above 4 or higher on the scale,
meaning at least “agreed a little Affect items were considered high when they were at
2 or higher, they had to be at least “a little bit happy " or “a little bit sad

6Please note that all log-linear analyses for the motivation questions had, at least,
one cell with less than 5 participants. These results should be interpreted with caution

because of the small cell sizes.

37

7Television use means were computed by averaging the scores for the amount of
television use on a typical weekday and weekend. Response options ranged from 0 to 6

or more hours.

38

APPENDICES

39

APPENDIX A

Parasocial Interaction Scale

Answer Selections:

1 = Disagree A lot, 2 = Disagree a Little, 3 = Neutral, 4 = Agree a Little, 5 = Agree A lot

1.

2.

8.

9.

_ makes me feel comfortable, like I’m with a friend.

I look forward to watching _ when his/her show is on.

If _ appeared on another Show, I would want to watch it.

__ seems like he/she would be easy to talk to.

__ seems to understand the kinds of things I want to know.

If I saw a story about _ in a newspaper or magazine, I would want to read it.
I miss seeing __ when his/her show isn’t on for some reason.

If I could, I would like to meet __ in person.

I feel sorry for _ when he/she makes a mistake.

10. __ would fit in well with my group of fiiends.

11. If something happens to _ I feel bad.

12. I would invite _ to my birthday party.

13. __ is the kind-of person I would like to play or hang out with.

14. If _ lived in my neighborhood we would be friends.

40

APPENDIX B
Character Attributes
Answer Selections:

l = Disagree A lot, 2 = Disagree a Little, 3 = Neutral, 4 = Agree a Little, 5 = Agree A lot

1. __ is kind.

2. __ is helpful.

3. _ is caring.

4. __ is mean.

5. __ is violent.

6. __ is selfish.

7. _ is handsome/pretty.

8. __ is good-looking.

9. __ is somewhat ugly.

10. _ is funny.

11. __ says and does funny things.
12. __ is physically strong.

13. __ has more strength than most people.
14. _ is smart.

15. _ is good at figuring things out.
16. __ is dumb.

41

TABLE 1

Parasocial Scale-Item by Item Comparisons

 

 

 

Item # Abused M Non-abused M t p
1 4.03 4.00 .11 .92
2 4.67 4.37 1.03 .31
3 4.20 4.03 .48 .64
4 3.73 3.50 .62 .54
5 3.07 3.47 -1.06 .29
6 4.17 4.23 -.19 .85
7 3.63 3.59 .14 .89
8 4.24 4.27 -.07 .94
9 3.63 3.69 -.15 .88
10 3.66 3.89 -.58 .57
11 3.80 3.70 .27 .79
12 4.03 4.24 -.64 .53
13 4.17 3.80 1.07 .29
14 4.34 4.40 -.19 .85
*p < .05.

42

TABLE 2

Mean Differences in Interval Level Variables by Abused Status and Gender

 

 

 

 

 

Abuse status ' Gender

Variable Abused Non-abused Boy Girl
Parasocial scale 3.93 3.89 3.80 3.98
Prevalence 1.72* 224* 1.79 2.21
Social behavior 4.05 4.06 3.74* 4.28*
Attractiveness 3.67 3.12 3.32 3.46
Humor 4.64 4.37 4.46 4.53
Strength 3.05 3.20 3.60”“ 2.78*
Intelligence 3.90 3.99 4.00 3.90
Affect

Sad 1.60 1.45 1.41 1.56

Happy 3.31 3.76 3.06 3.96
*p < .05.

43

 

TABLE 3

Percentage Differences in Nominal Level Variables by Abused Status and Gender

 

 

 

 

Abuse status ' Gender

Variable Abused Non-abused Boy Girl
Prevalence 31%** 72%** 46% 56%
Conversation Topics

Liking character 11% 38% 27% 32%

Episode information 22% 33% 9%* 42%*

Interaction w/ character 33% 52% 55% 42%

Other TV 1 1% 5% 18%* 0%*
Why Parasocial Interaction

Positive affect 27% 54% ‘ 23%* 59%*

Help 36% 13% 31% 14%

 

*p < .05. **p < .01.

44

35-22 .8. v a3.... Eases .3. v a...

,8qu

3m

Bo:

Combat

8:03:35

fiweobm

mmr 585$
ON. N. M . D>Eomut<
5.- sea. * fl m. Boom
*wm. ... _ m.. S. m L .- 85365
o _. em. 1?. 3am. 2. 3088mm

 

new Bo: 00:03:85

Emmobm 8:53 o>tofit< 300m coco—$65 3688mm 03359

 

was” 345V 38 REESE LO\M=.S9:=OD 836.23; \o 9.2333200 3.:er

v mqufi.

45

NI

1| IIIII illl‘llilillli
3 1293 O

IIIIIIIIIIIIIIIIII'I
470 0548

 

2