rt, .. RTE-3"

PREDICTORS OF TELEVISION VIEWING
AMONG JUNIOR HIGH SCHOOL STUDENTO

TIIE$I8 FOR THE OEOBEE OF M. A. _
MICHIGAN STATE UNIVERSITY

BRENDA OEBVIN
”I 9 5 O

.' . '-
" .-,~‘.
. 4-?"
. ..‘..‘,.‘I.“= 1

 

AUG 0; 200“

 

ABSTRACT
PREDICTORS OF TELEVISION VIEWING
AMONG JUNIOR HIGH SCHOOL STUDENTS

by Brenda Dervin

The four purposes of this study were: 1) to replicate some
previous work on correlates of frequency of child television viewing;
2) to include certain variables which have been partly or wholly over-
looked in past work; 3) to extend the analysis to a multivariate method;
and QI‘to specifically look at the types of relationships which exist
between the criterion variable and its predictors.

Respondents for this study were 252 seventh and eighth grade boys

‘I

K

'nd girls -- the entire junior high school class in a suburban school
system outside Milwaukee, Wisconsin. The data used here were drawn
from a larger survey tapping mass media behaviors and family life style
patterns. Survey questionnaires were self-administered in regular
classes during the 196u-65 school year.

From the available data, 63 variables were selected -- 62 pre-
dictors and the criterion. Among the major results and conclusions of
the study were the following.

1) Of the 62 predictors, 21 were significantly related (by X2

and/or r analyses) to the criterion variable.

 

 

Brenda Dervin

2) No single variable alone eXplained a great deal of variance
in the criterion. Maximum variance explained by any one variable was
18% with a curvilinear model, 9% with a linear model. This result
suggests the need for multivariate approaches.

3) Variables were subdivided into seven categories and multiple
Rs were run within categories to determine which variables contributed
significantly to variance explained, assuming a linear model. The

"best" within category predictors were: a) parent media use -- amount

 

of parent TV viewing and variety of parent radio use; b) respondent

 

media use -- amount of respondent radio use, variety of respondent
radio use,‘variety of respondent book preferences, frequency of
reapcrdent movie attendance, and medium respondent would miss most;

c) family cohesiveness -- none; d) community integration -- none;

 

 

e) self-orientation -- re5pondent outside home employment and respondent

 

math knowledge level; f) consumer orientation -- re3pondent attitude

 

toward credit and variety of respondent Spending; and g) demography_--

 

occupational prestige.

n) Assuming a linear model, the best category of predictors
(in terms of variance accounted for in the criterion ) was respondent
media use (18%). Second best was parent media use (1u%), followed by
self-orientation and consumer orientation (8% each). Demography accounted
for 6%. The multiple Rs for family cohesiveness and community integration
were not significant. One of the better predictor categories -- consumer
orientation -- included almost all new variables -- i.e. variables not

looked at as predictors of TV viewing in prior research.

Brenda Dervin

5) When the "best" within category predictors were pooled in one
multiple R.equation, the resulting R.was .50, accounting fOr 25% of the
variance in the criterion. When all 62 vairables, regardless of category,
were included in one multiple R equation, the resulting R was .65,
accounting for 42% of the variance. '

6) A comparison of the linear r with the curvilinear Eta
correlations for the relationship of each predictor to the criterion
indicated that the curvilinear model fit the data better.

7) A comparison of the current results with prior research
suggested that certain assumptions derived from early child-television-
research need re-examination, particularly in light of today's high
media saturation environment. As an example, the present results offered
little support fer two often-used generalizations about television

iewing -- the frustration hypothesis and the functional displacement

C +‘ .
hypotneSLS.

PREDICTORS or TELEVISION VIEWING
AMONG JUNIOR HIGH SCHOOL STUDENTS

By

Brenda Dervin

A THES IS

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

MASTER OF ARTS
Department of Communication

1968

JI

Accepted by the faculty of the Department
of Commmication, College of Communication Arts, Michigan
State University, in partial fulfillment of the

requirements for the Master of Arts degree.

.‘ /Z f
(rs
' __ ‘ flaflél'l -
Director of. lThesis

ACKN OW LB DGIEN TS

Special thanks go, first and foremost, to Dr. Bradley Greenberg,
who has been a most patient advisor and perceptive critic. Thanks go
also to Dr. Vincent Farace and Dr. Thomas Baldwin who both served on
the author's thesis committee; Albert Talbott, Jeffrey Katzer, and
Anita Immele who all gave considerable computer assistance; and

Mrs. Shirley Sherman who typed the final manuscript.

Po
[-0.

TABLE OF CONTENTS

CHAPTER

I INTRODUCTION . . . . . .

Rationale . . . . . .

Generalizations from Prior Research . . . . . . . . .
Classification of Variables

II HYPOTHBSES . . . . . .

Respondent Perceptions
ReSpondent Media Use .
Family Cohesiveness .
Self-Orientation . . .
Consumer Orientation .
Demography . . . . . .

III METHODOLOGY . . . . . .

of Parent Media Use . . . . . .

Questionnaire Administration . . . . . . . . . . . . .
The Respondents and Their Community . . . . . ... . .

nItem Selection . . . .
tem Measurement . . .
Item COding O O O 0 O

I v RES ULTS O O O O O O O 0

Reduction of Items by Deletion and Indexing . . . . .
Description of the Criterion Variable . . . . . . . .
Statistical Analyses . . . . . . . . . . . . . . . . .

Re5pondent Perceptions of Parent Media USe . . . . . .
Re3pondent Media Use . . . . . . . . . . . . . . . . .

Family Cohesiveness .
Community Integration
Self-Orientation . . .
Consumer Orientation .
Demography . . . . . .

Comparison of Multiple Linear Correlations Withi

Variable Categories

Results of Multiple Linear

Variable Categories
v SUMMARY AND CONCLUSIONS

BIDLIOGRAPHY . . . . . . . . . .

O O O O O O O O O O O O O C
O O O O O O O O O O O O O O
O O O O O O O O O O O O O O
O O O O O O O O C O O O O C
O O O O 0

Correlation Analyses Across

Page

.17
.21
.26
.36
.39
.42

.U8
.H9
.99
.50
.51
.52
.53
.59

.67
.77
.89
.95
102
111
118

126
127
132

151

LIST OF TABLES
Table Page

 

1. Means for Respondent Scores on the Top 16 Television
how Checklist, Reported by Questionnaire Administration

Date 0 I O I O O O C O O O C O O O C O 0 O C O O O O O O O O 56

2. One-Way Analysis of Variance Table Testing the Difference
Between Scores on the Top 16 Television Show Checklist by
Questionnaire Administration Date . . . . . . . . . . . . . 56

3. Means for ReSpondent Scores on the 10 Miscellaneous Adult
Drama Television Show Checklist, Reported by Questionnaire
Administration Date 0 O O O O O O O O O O O O O O O C O O O 5 7

4. One-Way Analysis of Variance Table Testing the Difference
Between Scores on the 10 Miscellaneous Adult Drama Television
Show Checklist by Questionnaire Administration Date . . . . 57

5. Summary of Results for Individual Hypotheses on the
Relationship of Parent Media Use Variables to Frequency

of Child Television Viewing . . . . . . . . . . . . . . . . 68
I

6. Results of Multiple Linear Correlation Analysis for the
Parent Media Use Variable Category . . . . . . . . . . . . . 71

7. Contingency Tables and Chi-Square Values for the Relationship
of the Parent Media Use Variables to the Criterion Variable 73

8. Summary of Results for Individual Hypotheses on the
Relationship of Re3pondent Media Use Variables to Frequency
of Child Television Viewing . . . . . . . . . . . . . . . . 77

9. Results of Multiple Linear Correlation Analysis for the .
Respondent Media Use Variable Category . . . . . . . . . . . 82

10. Contingency Tables and Chi-Square Values for the Relation-
ship of Respondent Media USe Variables to the Criterion
variable 0 O C O O O O O O O O O O O C O O C O O O O O O C 8“

ll. Summary of Results for Individual Hypotheses on the
Relationship of Family Cohesiveness Variables to Frequency
of Child Television Viewing ~. . . . . . . . . . . . . . . . 89

12. Contingency Tables and Chi-Square Values for the Relation-

ship of Family Cohesiveness Variables to the Criterion
vari able 0 O O O O C O C O O O I O O O O O O O O O O O O O 9 1

iv

List of Tables--continued

Table Page

 

13. Summary of Results for Individual Hypotheses on the
Relationship of Community Integration Variables to
Frequency of Child Television Viewing . . . . . . . . . . 95

14. Contingency Tables and Chi-Square Values fer the
Relationship of Community Integration Variables to the
Crite rion Vari ab le 0 O O O O O O O O O O O O O O O O O O 9 9

15. Summary of Results for Individual Hypotheses on the
Relationship of Self-orientation Variables to Frequency
of Child Television Viewing . . . . . . . . . . . . . . . 102

16. Results of Multiple Linear Correlation Analysis for the
Self-orientation Variable Category . . . . . . . . . . . 105

17. Contingency Tables and Chi—Square Values fer the
Relationship of Self-orientation Variables to the
crite rion variab le 0 O O I. O O O O O O O O O O O O O O 0 lo 7

, I
18. Summary of Results for Individual Hypotheses on the
Relationship of Consumer Orientation Variables to
Frequency of Child Television Viewing . . . . . . . . . 111

19. Results of Multiple Linear Correlation Analysis for the
Consumer Orientation Variable Category . . . . . . . . . llu

20. Contingency Tables and Chi-Square Values for the
Relationship of Consumer Orientation Variables to the
Criterion Variable . . . . . . . . . . . . . . . . . ._ 115

21. Summary of Results fer Individual Hypotheses on the
Relationship of Demographic Variables to Frequency
of Child Television Viewing . . . . . . . . . . . . . . . 118

22. Results of Multiple Linear Correlation Analysis for the
Demographic Variable Category . . . . . . . . . . . . . . 121

23. Contingency Tables and Chi-Square Values for the
Relationship of Demographic Variables to the
Criterion variable 0 O O O O O O O C C O O O O O O O O 0 l2 3

2n. Summary of Within Variable Category Multiple Linear
Correlation Analyses . . . . . . . . . . . . . . . . . . 126

25. Result of Multiple Linear Correlation Analysis with
"Best" Predictors from Within Variable Categories . . . . 128

List of Tables--continued

Table

26.

27.

28.

29.

30.

31.

33.

3“.

37.

Result of Multiple Linear Correlation Analysis of All
Predictor Variables Regardless of Variable Category . . . .

Items, Codes, and Interjudge Coding Reliabilities Listed
by Variable Category and Variable Name . . . . . . . . . .

Code Ranges, Means, Standard Deviations, and Non-ReSponse
Counts for the Five Items Used to Create Two Indexes of
Permissiveness’ . . . . . . . . . . . . . . . . . . . . . .

Matrix of Pearson Product Moment Correlations Between
the Five Items Used to Create Two Indexes of
PemiSSj-veness O O O O O O O O O. O O O O O O O O O O O O 0

Code Ranges, Means, Standard Deviations, and Non-Response
Counts for the Three Items Used to Create an Index of
Child-Parent commication o o o o o o o o o o o o o o o 0

Matrix of Pearson Product Moment Correlations Between the
Three Items Used to Create an Index of Child-Parent

Communication . . . . . . . . . . . . . . . . . . . . . .
1 .

Code Ranges, Means, Standard Deviations, and Non-Re3ponse
Counts for the Two Items USed to Create an Index of
Fm]: 1y Togethemess O O O I O O O O O O O O O O r. O O O O 0

Code Ranges, Means, Standard Deviations, and Non-
Response Counts for the in Items Used to Create an Index
of Re3pondent Home Responsibilities . . . . . . . . . . . .

Matrix of Pearson Product Moment Correlations Between the
1M Items Used to Create an Index of ReSpondent Home
Respons ibilities O O ' O O O O O O O O O O O O O O O O O O 0

Code Ranges, Means, Standard Deviations, and Non-Re3ponse
Counts for the Feur Items Used to Create an Index of
Respondent Knowledge of Family Operation . . . . . . . . .

Matrix of Pearson Product Moment Correlations Between
the Four Items Used to Create an Index of Respondent
Knowledge of Family Operation . . . . . . . . . . . . . . .

Code Ranges, Means, Standard Deviations, and Non-ReSponse
Counts fer the Two Items Used to Create an Index of
ParentGregariousness ..................

Code Ranges, Means, Standard Deviations, and Non-Re5ponse
Counts for the Four Items Tapping Re5pondent Gregariousness

VJ.

 

130

157

178

178

179

180

180

182

182

18%

184

185

186

#1.

42.

1‘3.

'44.

.1.
(.VI

”6.

47.

#8.

Page

Matrix of Pearson Product Moment Correlations Between
the Four Items Tapping Respondent Gregariousness . . . . . 186

Code Ranges, Mean , Standard Deviations, and Non-Response
Counts for a “s ‘ ed to Create the Outside
Home Orientation Index . . . . . . . . . . . . . . . . . 187

Matrix of Pearson Product Moment Correlations for the
Seven Items Used to Create the Outside Home
orientation Index 0 0 O O O O O O O O O O O O O O O O O 188

Code Ranges, Means, Standard Deviations, and Non-Response
Counts for the Two Items Used to Create the Frequency
of Church Attendance Index . . . . . . . . . . . . . . . . 188

Code Ranges, Means, Standard Deviations, and Non-Response
Counts for the Five Items Used to Create the Math
KnOVIIlEdge Level Index. 0 O O O O O O O O O O O O O O O O O 189

«ix of Pearson Product Moment Correlations Between the
Items Used to Create the Math Knowledge Level INdex 190

Code Ranges, Means, Standard Deviations, and Non-ReSponse
Counts for the Two Items Used to Create the Family

Shopping Orientation Index . . . . . . . . . . . . . . . . 191
Code Ranges, Means, Standard Deviations, and Non-Response
Counts for the Four Items Used to Create the Socio-

Economic Status Index of Possessions . . . . . . . . . . . 192 -

Matrix of Pearson Product Moment Correlations Between the
Four Items USed in Creating the Socio-Bconomic Status
Index of Possessions . . . . . . . . . . . . . . . . . . 192

Code Ranges, Means, Standard Deviations, and Non-Response
Counts, Listed by Variable Category and Variable Name . . 191+

APPENDIX

A.

LIST OF APPENDICES

ITEMS, CODES, AND INTERJUDGE CODING
RELIABILITIES LISTED BY VARIABLE CATEGORY.
FUND VAR-IABLE NAIE O O O O O O O C O O O O O O 157

INDEXING PROCEDURES . . . . . . . . . . . . . 176
CODE RANGES, MEANS, STANDARD DEVIATIONS,

AND NON-RESPONSE COUNTS LISTED BY VARIABLE
CATEGORY AND VARIABLE NAME . . . . . . . . . 193

viii

I ...._, . as -_.+_ A . 44

CHAPTER I

INTRODUCTION

Rationale

Ever since television burst upon the U.S. scene in the 1950's,
the attention of researchers, critics, and the public has been focused
on possible effects of the new, imminently attractive, and pervasive
medium on children. Generally, attention to the medium has been rooted
in a concern for possible detrimental effects. Does viewing television
ruin eyesight? Does it cause children to withdraw from real life
concerns? Does it lead to increased aggression, cause juvenile
delinquepcy? Does it debase tastes?

While the critics and researchers have been busy looking for
effects, the public has been busy accepting the new medium. In a true
success story of the 20th century, the proportion of 0.8. homes with
television sets rose from 7% in 1950 to 82% in 1957 (Witty 1963,

Bogart 1958). By 1960, saturation had reached 90% (Bogart 1962) and,.
in 1965, reports of 9A to 100% saturation were common. -(Witty 1966,
Rainwater, Coleman, and Handel 1959, Huber 1965.) Television continues
to be the major leisure time activity with the typical family
television set operating 5 hours a day, the average man viewing for

2 1/2 hours, average woman for 3 1/2 hours, and average child for u

hours (Bogart 1962).

While the popularity of television rose with relative ease,
the search for effects did not. Indeed, the search might well be
summarized by saying that the more researchers looked for effects,
the less they found. In 1957, Meyersohn noted that few of the
Inmdreds of studies on television and children had "succeeded in
providing much of consequence." (Meyersohn 1957) As late as 196u,
Ekbramm asserted that one should not expect "too much, too soon, and
too specifically" from effects research (Schramm 1964).

In delineating the reasons for the lack of clarity offered by
effects research, there seems to be general agreement in pointing to
the failure of the "hypodermic model" of communications research.
Schraum, White, Klapper and others have all pinpointed the evolution
of research on the mass media from a "direct effects" model to the
currently accepted phenomenalistic, functional, situational, or uses
and gratifications approaches (Schramm 1962, White 196u, Klapper 1960,
Katz and Foulkes 1962, Bauer 1964).

While once the mass media researCher asked "what do the mass
media do to people", today he more often asks "what do people do with
the mass media?" While once the researcher viewed the individual
child in isolation from his family, social background, and environment,
today the researcher looks to these "situational and contextual"
elements fer explanation of the television-child interaction. While

once the researcher was concerned primarily with measuring effects

almost totally in terms of source expectations, today he looks to

umsequences which may be entirely independent of source intent.
lerper's generalization that mass communication ordinarily does "not
serve as a necessary and sufficient cause of audience effects but
rather functions among and through a nexus of mediating factors and
:hufluences" best sums up the result of the evolution (Klapper 1960,
p. 8).

A major consequence of this newer approach to media research is
that the task is now immensely more complicated. The hypodermic model
filits simplicity failed to yield clarity while the functional model
in its complexity must be concerned with an enormous number of variables
axltheir interactions in order to begin to yield clarity.

In ahcepting a more functional approach to their subject, re-
searchdrs have Spread their concern from analysis of aggregates or
demographic characteristics (White 196%) to concern for the ability,
maturity, personality, peer group relations, and family relations of
the hild (Himmelweit, Oppenheim, and Vince 1958, Riley and Riley 1955,

Bailyn 1959, Maccoby 196M, Schramm, Lyle, Parker 1961).1

Along with the emergence of the functional approach, there has
been an increasing awareness of the conditions necessary for developing
causality statements in social science research. While once the

researcher was quite ready to infer effect statements from field

 

1Two of these references -- Himmelweit, Oppenheim, and Vince (1958)
and Sohramm, Lyle, and Parker (1961) will be referred to extensively
in Chapters I and II. All further citations of these two references
will be made simply to Himmelweit (1958) and Schramm (1961).

Iemxxth, today researchers are careful in emphasizing that their

fkfid studies on the relationship of television and children are
cmnelational in nature. After careful analysis of the methods of
ldfldng for the effects of television, Maccoby concluded that only
theleboratory experiment can come close to causality requirements.
Quanoted that even precise before-after field surveys using matching
tedudques do not offer true randomization, so causal connections cannot
begfinned down from such research (Maccoby 196u).

As a result of these two trends 1. the acceptance of the functional
axuoadh to research and the awareness of the limitations of field
zesearCh -- the current state of affairs in television and children
research seems to include:

1) an acceptance of'non-experimental field research as a

search for "correlates" of media use rather than "effects"
of media use;

2) an acceptance of the need to look at a number of areas

of the child's life in order to more fully understand
and describe the child-television interaction; and
3) a call for increased eXperimental work from which "effect

statements" may be derived.

We find Schramm summarizing his intensive review of the past
television-children research by saying that the classic field surveys
'have gone about as far as it is possible to go with survey methods. . . ."

(Sdntmm1196u, p. 7). Yet, rejection of the "correlate" approach to

researdh on television and children may be a bit too hasty. While the

 

-u

MIA

 
 

correlational-type approach will not yield effect statements, the
possibilities of the approach have not been tapped to their fullest
extent.

Most of the past work, while acknowledging a correlational
approach, has been very much rooted in a search for effects. Findings
on television and children have come mainly from studies concerned
with the differences between matched reSpondents in television and
non-television households. The two classic studies -- Himmelweit (1958)
and Schramm (1961) —- were based on a concern for the effects of the
introduction of television although both did look somewhat at correlates
of amount of viewing.

' Whileithere is a great deal of research on television and
childrea only a small subset is directly relevant to the television
environment of the 1960's with its almost 100% saturation of TV
households. The absence-presence of viewing is almost no longer
relevant. White, Bdelstein and others have made references to the
need to take a new look at audience composition and utilization of
media in terms of the current media environment. (Edelstein 1966,
White 1964) They suggest that once well-supported principles of mass
media behavior like the "all or none principle" (first posited by
Lazarsfeld and Kendall 1948) may no longer hold in today's context of
media overload. One recent factor analytic study of media exposure
patterns of teenagers, for example, suggests that once rather clear-
cut relationships between various aspects of media usage may no longer

be quite so clear. (Troldahl and Costello 1966).

 

u--.
r ..
p. ..

 

 

4... ‘

w..-

 

~-. -
- .
uh...

‘--¢

 

a.
h.

 

 

If:

‘\
‘v

 

Another reason the correlational approach still has merit is
that many crucial variables have been completely or partly ignored in
the past work. For example, we find that family consumer behaviors
have been rarely included as possible correlates of child television
behaviors. Yet, with one of the main family functions being its
consumption role in society, this seems a lucrative area in which to
extend the television-children focus. Chapter II more fully details
variables which have been partly or wholly overlooked in the past work.

A third reason for not yet eulogizing the correlational approach
to television and children is the almost total absence in the past
work of multivariate analytic approaches. Himmelweit (1958) and
SchramN (1961) illustrate the common use of contingency analysis of
frequencies and percentages. While researchers have developed lists
of what they consider to be the most important predictor variables
of television behavior these have usually been derived from analyses
of one predictor variable at a time and developed into a kind of
jigsaw puzzle portrait. An exception is Bailyn (1959) who used a
multivariate technique to arrive at a statement of the four variables
which accounted for approximately 47% of the variance in amount of
exposure to the pictorial media (comics, films, television).

Multivariate analysis techniques offer several interesting
possibilities for the area of television and children. First is the
obvious advantage of using techniques which detect spurious corre-
lations and isolate interactions. Second is the ability of these

techniques to handle the larger numbers of variables demanded by the

more functional analysis of media behaviors. Third is the parsimony
offered by techniques which reduce a large set of variables down to
the set most important in terms of variance accounted for.

Related to the lack of use of multivariate approaches, the past
work on television and children also shows an almost total absence of
concern with the exact nature of the relationship of predictor variables
to the criterion media behavior. Most predictions of relationships
' have been stated in a linear fashion. Further, even contingency
analysis which can tap curvilinear relationships has been essentially
used with dichotomous classification of the criterion television
viewing behavior. This makes detection of curvilinearity impossible.
The pichotomization of the criterion variable was a necessary con-
straint for those researchers who, in the past, looked at the absence
and presence of television. However, it is rare to find any more than
two categories even when researchers were looking at amount of
television viewing. Schramm (1961), for example, used simply high and
low television viewing categories. An exception is the Himmelweit
(1958) study in which three categories of television viewing were used.

In summary, then, the correlational approach to studying the
relationship of television and children seems to still have merit
mainly because it has not been utilized to its fullest. Much of the
past research does not apply to today's 100 per cent television
saturation. Many crucial variables have been partly or wholly over-
looked. Multivariate analysis techniques have not been used, and

specific attempts to look at the nature of the relationships between

predictor variables and the criterion media behavior have not been
made.

With this background, the present study aims to look at correlates
of the amount of television viewing by junior high school students.
The purposes of this study are:

1) To replicate much of the past work on correlates of

frequency of child television viewing;

2) To go beyond a replication by including some

variables which have been partly or wholly overlooked
in past works;

3) To extend the analysis to a multivariate method; and

4) To Specifically look at the types of relationships

which exist between the criterion variable and its

predictors.

Generalizations from_prior research

 

It seems most efficient at this point to draw together the
most parsimonious generalizations from across the literature and
reserve Specific citations for supporting the hypotheses formulated in
Chapter II. The generalizations stated in this section come primarily
from the conceptual discussions of Himmelweit (1958), Schramm (1961),
Campbell (1962), Maccoby (1964), and Klapper (1960).

A thread seems to tie the various recent approaches to predicting
amount of television viewing together. This thread might best be
called the functionalist's agreement to look at the child as an
integrated human being acting in a reasonable fashion within the

context of his environment. We find such opening generalizations as

"the mass media exist because they are useful in meeting human needs
and that TV has come into use ... because it meets some of these needs
better than any other alternative." (Schramm 1961, p. 74). This unifying
vieWpoint ties together the more firm generalizations derivable from
past work.

Most writers have phrased their generalizations in terms of
effects rather than in terms of relationships. Thus, one finds
reference to the direct and indirect effects possible from television
in Maccoby (1964; or a list of possible effects in terms of physical,
emotional, cognitive, and behavioral effects in Schramm (1961), or to
"displacement of time effects". and effects of content in Himmelweit

O

(1959 . The generalizations stated below have rephrased "effect"
nations into "correlational" notions.

A good deal of the prior research from which these generalizations
are derived have dealt with both the quantity and quality of a child's
television viewing behavior. Since the present study deals only with
amount of television viewing, findings and generalizations are derived
from past work only when relevant to the quantity of viewing. We have

derived the following five generalizations:

1. The parental imitation generalization: While not explicitly

 

Stated as such in past literature, an underlying theme of the results
might be framed: "children tend to do what their parents do, all other
things equal." Thus, consistently researchers have feund that one of
the best predictors of any child's media behavior is the media behavior

of his parents. Three major researchers all concur that if parents

10

view a lot, children tend to do likewise; if parents view moderately,
children tend to also. (Schramm 1961, Maccoby 1964, Himmelweit 1958).

A logical consequence of the above generalization is that a number
of demographic characteristics which predict parent television behavior
predict the child's behavior as well. Rather consistently the findings
agree that the occupational level and education of parents predicts
the amount their children view (Bailyn 1959, Schramm 1961). Specific
findings on social class will be offered later but the variable is
offered now as an example of the class of readily tapped variables
which might be termed attributes of parents which predict the child's
behavior.

.

2. The demographic attributes of the child generalization:
Again this generalization has not been stated explicitly in the prior
research. It suggests that there are a number of good predictors of
television viewing which are not accounted for by parental behavior or
attributes nor strictly by generalizations 3, 4, and 5 below. For
example, the child's age is often found to be strong predictors of
various media behaviors. In terms of television viewing, findings
rather consistently concur that the peak television viewing time for

children is roughly from the ages of 11 to 13 (SChramm 1961).

3. The functional displacement generalization: This

 

_generalization was first framed by Himmelweit (1958) and has
structured numerous approaches since. In essence, the generalization
is that a child must choose between activities and that he will
sacrifice in lieu of television those activities which satisfy the

same needs as television but do so less effectively. He will not

11

sacrifice, however, those activities which serve needs different than
those served by television. The term "functional" is essential in the
_generalization as what is "functional" to one group of children may
not necessarily be for another. As an example, it is a rather consistent
finding that, movie attendance is sacrificed for television viewing
(Himmelweit 1958, Schramm 1958 and others). However, when a child
reaches adolescence the movie theater becomes a meeting place fer
friends and the displacement phenomenon no longer holds.

One problem with the di5placement generalization is that it has
most often been tested with a criterion variable of absence vs presence
of television in the home. It is difficult to make the leap from a

O

finding in that context to a prediction that would hold fer amount of
teievision viewing. It seemed quite clear that when families purchased
television sets movie viewing dropped off. However, after years of
television ownership, it is not so clear that the person with a high
desire for the kind of need fulfillment offered.by both television

and movies will fulfill that need primarily through television.,

4. The frustration generalization: First stated by Maccoby

 

(1951) this generalization States that the more frustrated a Child

is, the more time he will Spend in front of his television set. The
usual indices of frustration have been the degree of the child's
integration within both his family and his peer group social systems.
In a theoretical discussion of the use of mass media as escape, Katz
and Foulkes (1962) point to the impressive evidence supporting the
notion that alienation, deprivation, and frustration leads to increased

exposure.

5. The information void generalization: While more directly

 

related to the quality of a child's viewing rather than the quantity,
this generalization is relevant to some aspects of the present study.
Posited first by Himmelweit (1958), the generalization states that the
conditions under which television content is likely to have an effect
on a child's values and outlooks are: a) if the values and views
recur over and over again in TV content; b) if the ferm of presentation
is dramatic; c) if the contentis linked with the child's needs and
interests; d) if the child is uncritical and attached to the medium;
and e) if the child is not presented through peers or relatives with
a standard against which to assess the views offered on television.
Thus? Himmelweit found that female adolescent television viewers were
more concerned than matched non-viewers with the problems of growing up
and marrying. From this generalization, one might reason that since
television content in the United States tends to be somewhat of the
same ilk, that there should be some relationship between the amount
a child views television and the child's attitudes on topics which
television emphasizes but parents and peers rarely talk about.

The five'generalizations above form a pattern which agrees
with the functionalist's basic view of predicting the child's behavior
within the context of his environment. In essence, the generalizations
say that when predicting the amount a child views television, the
research can account for a certain amount of variance by looking at
the demographic characteristics of the child's parents and at the

parent's media behavior.’ Another portion of the variance can be

 

 

 

,-
.
, u.

 

-.
l
....
u
I...
.

n.-
in:

 

 

 

‘V

 

 

 

 

13

accounted for by looking at the child's own demographic characteristics
and his other media behaviors (which, not surprisingly, are often

a close replica of the media behaviors of his parents). After using these
rather efficient approaches (efficient in the sense that the variables
involved are relatively easy to tap), the further differences between
children must be accounted for by looking at the values and attitudes

of the child and the child's integration within his family unit and

within his peer group.

Classification of variables

 

Researchers in this area increasingly agree that the greatest
explanatory results are yielded by surveys which seek information on
a‘variety of dimensions of a child's life. Thus, Himmelweit, in
describing television effects, had to time and again "consider the ways
these effects differed according to the ability, maturity, background,
and personality of the children concerned." (1958, p. xiv). Essentially,
there is agreement that a sizeable number of variables are needed
to predict the child's television behavior. With this in mind; the
present study deals with a large number of predictor variables, 62
in all.

With such a large number of predictors, some efficient method
of classifying the variables is necessary. The 62 predictor variables
tapped in this study were grouped into seven categories on the

following criteria:

14

1) Agreement with the methods used to group variables in past

 

research. It has been rather standard procedure in past work to group
attributes of the child's media use as a separate category of
variables. Other consistently used categories have included: parent
media use, family integration, community integration, personal ex-
pression (e.g. hobbies), and demography. These are essentially the
categories used for the present study.

2) Facilitation of the formation of hypotheses. A second

 

criterion in the classification of variables used here-is to group
together variables for which predictions form a unified whole. For
example, past literature suggests that predictions about the re-
lationship of1elevision viewing to family integration may all

essentially be derived from the frustration hypothesis. When a

00

romp of variables seemed to easily fall into such a unified category,
this advantage was utilized.

3) Concern for the number of variables within each category.

 

Since one of the purposes of the present study is to utilize a multi-
variate analysis as well as a variable-by-variable prediction method,
an effort was made to keep the number of variables within each
category large enough so that multivariate analysis could be utilized
both within variable categories and then across the entire set of 62
predictors.

Based on the above criteria, the 62 predictors in this study
were organized in seven categories: 1) parent media use; 2) respondent
media use; 3) family cohesiveness; 1+) community integration; 5) self

orientation; 6) consumer orientation; and 7) demography. The specific

15

rationale for each category is in Chapter II. Throughout this report,
these seven categories are used as a means of organizing the statement

of hypotheses, findings, and implications.

CHAPTER II

H‘IPOTHESES

In review, the purposes of this study are: l) to replicate much

of the past work on correlates of frequency of child television viewing;
2) to go beyond a replication by including some variables which have
been partly or wholly overlooked in past work; 3) to extend the analysis
to a multivariate method; and it) to Specifically look at the types of

relationships which exist between the criterion variable and its

predictors. Since very little of the available work in this area has

attacked‘purposes 3 and 1+, formal hypotheses will be stated only for

Purp‘ose l and 2.
The stage for hypotheses might best be set by a brief review

01‘

the general television behavior of children who are around the

3838 of 13.-11+. Past work suggests that these respondents are at the

peak of their television viewing time -- viewing some 20 or more
hours a week (Schramm 1961, Witty 1963, Himmelweit 1958, Maccoby .1963).
Piu<izlngs also agree that television is the dominant medium in their
lives (Baxter 1960, Maccoby 1964, Schramm 1961). However, this is

also the age range within which children begin to develop adult media
use patterns. As Schramm put it: this is "the time near the beginning
of the teens when adult patterns begin to replace childish ones and

Spectacular changes in media behavior occur." (1961, p. 117)

16

17

Within this context, hypotheses for this study are stated in

terms of the seven variable categories outlined in Chapter I.

Respondent perceptions of parent media use
The first category includes variables tapping the media

behaviors of the reSpondent's parents. An impressive array of past

research shows that one of the strong predictors of the child's own

behavior is the behavior of his' parents (Schramm 1961, Himmelweit 1958,

Maccoby 1961+, Bailyn 1959).
Derivation of hypotheses for this category of variables involves

a combination of both the parental imitation and functional dis-
A prediction may

placement generalizations outlined in Chapter I.

be cieduced directly from the parental imitation generalization that
the amount of a child's viewing will reflect the amount of his
Par-e nt's viewing. When dealing with other media behaviors (e.g. radio,

mOVi es) a two step prediction is involved. First, the displacement

generalization is used to predict how the parents' amount of

tale Vision would relate to other media behaviors of the parents. Then,

if the parental imitation hypothesis holds, one would expect the

child's behavior to again mirror that of his parents.

Amount of parental television viewing. Schramm (l96l),
Hil"Xlz‘nelweit (1958) and Maccoby (1961+) all concur that the amount of
parental television viewing is one of the strong predictors of the

E”‘T‘C-‘nlnt of a child's viewing. Hypothesis 1 is:

If”

18

H1: Frequency of parental television viewing
will be positively related to frequency of
child television viewing.
Variety of parental television viewing. Few prior studies
have looked at variety of parental viewing as a predictor of amount
of child viewing. However, logic would suggest that greater variety

is, to some degree at least, related to greater quantity. Thus,

hypothesis 2 is:

H2: Variety of parental television viewing will be
positively related to frequency of child
television viewing.

Number of television sets in household. Again, none of the
imiterature cited here has included this as a predictor variable but

!
.l<qgic would suggest that the family owning more television sets

I
.p;leaces more emphasis on television as an activity. Thus, hypothesis

3 :Lsn

H3: Number of television sets will be positively
related to frequency of child television
viewing.

Amount of parental radio listening, None of the past work
reviewed here looked Specifically at the relationship between parental
rec1:10 use and child television behavior. However, the bulk of past
"<>JFJ< agrees that with the introduction of television, radio listening
:3‘1iFiered for both adult and child television viewers (Coffin 1955,
Abrams 1956, Maccoby'lQSl, Campbell 1962, Schramm 1961, Himmelweit
1958). The relationship of amount of viewing to amount of radio use
iii not quite so clear. Parker (1960) reported a non-significant

'relationship as did Troldahl and Costello (1966). Bailyn (1959)

however, found a small but significant negative correlation between

the two variables. In view of this conflict in evidence, the pre-

diction for this variable is based on a logical inference from both the
functional displacement and parental imitation generalizations outlined

in Chapter I. These generalizations suggest that radio use should be

displaced, to some degree at least, by increasing television usage

and that children should imitate their parents in these media behaviors.

'Thus hypothesis 4 is:

Amount of parental radio use will be negatively

an:
related to frequency of child television viewing.

Variety ofgparental radio use. Pew past studies have looked

especifically at this variable in relation to child television usage.
.Iir their comparisons of television vs non-television children, both

ESCiaramm (1961) and Himmelweit (1958) found that radio became a more

SPecialized medium in television households. While the leap from the

CijLcehotomous television vs no television situation to actual amount

C>i?' viewing is tenuous, these findings suggest hypothesis 5:

H5: Variety of parental radio use will be negatively
related to frequency of child television viewing.

Number of newSpapers subscribed to. Again, none of the work

clfiLted here has looked at the relationship of parental use of the print

rneedia to child's use of television. However, considerable prior work

flee looked at the relationship of adult print media use to adult
‘television use and child print media use to child television use.
The findings fairly consistently agree that, with the introduction of

television, some aSpects of viewer reading suffered (Belson 1959,

20

Coffin l955, Campbell 1962, Himmelweit 1958, Bailyn 1959, Schramm 196l,

and Parker 1963). The reduction appeared heaviest for books and

magazines but affected some aSpects of newSpaper reading. However,
several trend studies indicated that the difference between television

vs non-television users tended to disappear over time. Fewer studies

have looked Specifically at the relationship between amount of

television viewing and use of the print media. For use of newspapers,

the findings conflict. Westley and Severin (1961;) found a non-

! significant relationship between adult time spent on television and
time Spent on newspapers while Himmelweit (1958) found that television

viewing related negatively to newSpaper reading for their children

respondents. Despite some contradictions, hypothesis 6 is:

I
; H6: The number of newspapers subscribed to by

f parents will be negatively related to
' frequency of child television viewing.

" Number of magazines parents read. Here again evidence is

uncle ax. Himmelweit (1958) found a tendency for magazine reading

t° suffer with increasing amounts of viewing. The television V5 “'3
talevision comparisons (as indicated in support for hypotheSiS 5)
generally agreed that magazine reading decreased with the introduction
0f television. Thus hypothesis 7 is:

‘ H7: Amount of parental magazine reading will be
negatively related to frequency of child
television viewing.

 

 

-_ A. -
-.-

W s .
J ‘._...-—— ..WWHP‘ ....

..q

21

Parental sources of news. None of the literature cited here

 

3155 looked at the relationship between parental sources of news and
axxnmt of child television viewing. However, other work in the
germral area of media credibility and media perceptions suggests that
perceptions of major news sources is related positively to amount of
media use (Carter and Greenberg 1965, Westley and Severin 1961+).
Thus, hypothesis 8 is:

H8: Children who perceive their parents' major

source of news as television will be more
frequent television viewers than children
who. perceive their parents' major news source
as a medium other than television.

Number of phonographs owned. None of the television use
studies cited here have looked at phonograph usage as a predictor
Variab 1e. The only support available comes from consumer studies
(Such as Huber 1965, Caplovitz 1963) that indicate that families that
have gone in debt for television sets have very often also done so for
phonOgmaphs. On the basis of this slim evidence, hypothesis 9 is:

H9: Number of phonographs owned will be

positively related to frequency of child
television viewing.

Re\Sp°\rldent media use
The second category of variables treated in this study taps the
re . .
SpQT‘Adent'S own media behaViors. Most category schemes developed by

re . .
Searchers have a class of this sort (e.g. Schramm 1961, Himmelweit

 

 

22

Predictions within this category have been derived mainly from
me functional displacement generalization, stating that other activities
will be sacrificed in lieu of television if they serve the same needs
as television but do so less well.

Amount of respondent radio use. Support for a prediction here
is essentially the same as that offered for hypothesis 1 stated on
page 18. Hypothesis 10 is:

H10: Amount of respondent radio use will be

negatively related to frequency of
television viewing.

Variety of respondent radio use. Again, support is essentially
the same as that offered for hypothesis 2 stated on page 18.
Hypothesis ll is:

Hll: Variety of respondent radio use will be

negatively related to frequency of television
Viewing.

Variety of respondent newspaper use. Most of the prior work
has looked at another dimension of newspaper usage, namely time SPént
on neWSpaper reading. The evidence is fairly well summed up in
Support of hypothesis 6 stated on page 20 on parental newspaper use --
newsp aI>er reading suffered with the introduction of television but the
mlat ionship of actual newspaper use to time Of viewing is “0': clear.
_Schralnm (1961) further suggests that with junior high school age
childPen, at least, newspaper usage is jUSt beginning to expand 5°

V““'\‘
culanee on the variable is restricted. This evidence suggests

my? Q‘ihes is 12:

23

H12: Variety of reSpondent newspaper use will be

negatively related to frequency of television
viewing.

Number of magazines respondent reads. This variable is also
bunkered by lack of support and conflicting evidence as noted in the
discussion preceding hypothesis 7 stated on page 20 on parental
xnagazine use. DeSpite the conflict, the prediction on respondent

magazine use agrees with that on parent magazine use. Hypothesis 13 is:

H13: Number of magazines the respondent reads will
be negatively related to frequency of
television viewing.
Variety of reSpondent book preferences. As with the other print

media use‘variables, evidence here is unclear. However, in predicting

he rgelationship of book reading to amount of television viewing,

rt

SEVTBIFal researchers have found.negative relationships between the
Quéuaftzitative dimensions of both variables (Himmelweit 1958, Bailyn 1959)
”hiJLfie others have found non-significant relationships. Parker(l963)
e”(Pl—Eiined the lack of significance in terms of the displacement
gene I‘alization which suggests that television viewing should displace
only certain kinds of book reading, namely fiction which fulfills the
mane: :fantasy gratification function. While the bulk of evidence
Sug‘gests a negative relationship bet-teen amount of book reading and
amo‘nl‘t of television viewing, interestingly the only finding available
for the variable -- variety of respondent book preferences -- suggests
a Positive relationship. Himmelweit(1958) found (in' her ‘ absence vs
pbe‘S-ence of television viewing situation) that television widened the

heading tastes of viewers. While making the leap from this finding

24

to amount of television viewing is, perhaps, tenuous, hypothesis 1% is:
H14: Variety of respondent book preferences will be
positively related to frequency of television

viewing.

Respondent preference for comics. Evidence here is clearer

 

than for the other print media. Parker(l961l Himmelweit(l959l and
Bailyn<1959)all found a positive relationship between amount of comic
book reading and amount of television viewing -- a relationship that
is particularly strong for addict television viewers. Thus,
hypothesis 15 is:
H15: ReSpondent preference for comics will be
positively related to frequency of television

viewing.

' ReSpondent library use. This variable has received attention

 

mainly from Parker (1963) and Bogart (1958). In his absence vs
presence of television analysis, Parker reported that television
displaced library fiction circulation mainly but that library use

did decrease generally. Again, the leap to a prediction for amount
of television viewing is tenuous. However, on the basis of the above
evidence, hypothesis 16 is:

H16: Respondent library use will be negatively

related to frequency of television viewing.

Frequency of respondent movie attendance. Most of the evidence
here applies to the absence vs presence of television situation and
generally agrees that movie attendance was hard hit by the introduction
of television (Coffin 1955, Campbell 1962, Belson 1959, Abrams 1956,

Maccoby 1951). However, Himmelweit (1958) emphasized that, for

25

teenagers, television should not functionally replace movies as the
movie theater becomes an arena for social interaction at this age.
While that finding might suggest a non-significant relationship
between movie attendance and televisionrviewing, Bailyn (1959) found
a significant positive relationship between the two variables.
Hypothesis 17 is:

H17: Frequency of reSpondent movie attendance

will be positively related to frequency of
television viewing.

‘ Variety of respondent record preferences. Support here is the
same as that offered for hypothesis 8 stated on page 21 on parental
ownership of phonographs. Hypothesis 18 is:

H18: Variety of reSpondent record preferences

will be positively related to frequency
of television viewing.

I

Media respondent would miss most. Hypotheses for respondent
media preference and media credibility ratings must be derived mainly
fumistudies other than those done on children's television use.

Wenfley and Severin (196R) and Carter and Greenberg (1965) suggest that
gwxeeptions of the media are directly related to media use.
Hypothesis 19 is:
H19: ReSpondents who indicate they would miss
television most will be heavier television
Viewers.

3 Respondent media credibility ratings. Based on the support

oflrred for hypothesis 19 above, hypothesis 20 is:

26

D

n:
r.)
0

Respondents who indicate that television
is their most believed medium will be
heavier television viewers while those
who indicate television is their least
believed medium will be lighter
television viewers.

Family cohesiveness

 

The third category of variables tapped in this study includes
respondent reports on the character of interaction within his family.
An impressive number of researchers have emphasized the need to look
at the child's media behavior within the context of his family life.
(Freidson 1955, Freedman 1961, Meyersohn 1957, Campbell 1962,

Clark 1965, Riley and Riley 1955).

s

DeSpite agreement on the need to look at the child's family life,
regatively little work has been done in the area of family relation-
ships and mass media. One recent review of child development re-
search, for example, lists not a single reference in a one-hundred-
plus item bibliography on the relationship of the mass media within
the context of the child's development (Douvan and Gold 1966). .Most
of the support cited for hypotheses below, therefbre, is quite recent
and indicative of an attempt (pointed out by Clarke 1965). to study
the development of media use patterns by looking at parental
socialization techniques.

A growing body of evidence suggests that this is a fruitful
approach but one also fraught with problems. Himmelweit (1958),
Schramm (1961), and other of the major children-television researchers

have emphasized their collective findings that the quality of a

child's home life is a predictor of his preoccupation with television.

27

1

Host of their findings may be derived from the frustration
generalization which simply states that the more frustrated a child
is the more he will turn to television as an escape. Freidson (1953)
and Katz and Rmdkes(l962) have dampened the clarity of this gen-
eralization however, with their assertion that high exposure to
television may be as much a sign of lack of frustration in a closely
knit family as it is a sign of frustration and resulting efferts to
escape in a disruptive family. Other researchers (e.g. Clausen 1966)
have also begun to Stress the need for looking at the structure of a
family as a complex class of variables needing analysis in at least
several dimensions in order to yield clarity. DeSpite these

:
admonitions which point up glaring weaknesses in existing attempts
to look at the mass media within the family life context, the
available work does provide a baseline fer making predictions for the

present study.

Parental‘permissiveness. This variable seems most closely

 

aligned with the variable past researchers have used to test the
frustration hypothesis in the context of family life. Maccoby (1954),
for example, found that highly frustrated middle class children Spent

more time viewing television than children who were not frustrated.

A major portion of Maccoby's index of frustration was the severity of

punishment and the degree of percental permissiveness. Becker (1964)

 

1Most of the literature has applied this prediction to middle
class reSpondents only. The respondents for the present study are for
the most part middle class, as will be pointed out in Chapter III.

28

also notes that restrictive socialization techniques tend to lead
to fearful, dependent, submissive children. The Maccoby finding has
been consistently replicated by Schramm (1961), Bailyn (1959), and
Lyle (1962). Thus, hypothesis 21 is:

H21: Parental permissiveness will be negatively

related to frequency of child television
viewing.

Child-parent communication. None of the literature cited here

 

deals Specifically with the degree to which the child feels able to
talk to his parents. However, several of the major television and
children studies make references to child-parent communication and
child-parent conflicts. Thus, Maccoby (1954) used as one of her
indices of frustration the degree of warmth in the child-parent in-
teraction. Schramm (1961) found that when children saw a conflict
between themselves and their parents they Spent more time with
television. Fine and Maccoby (1962) and Campbell (1962) indicated
in their absence vs presence of television studies that while
television families spent more time together there was less inter;
personal interaction during that time. The intersection of this
evidence suggests hypothesis 22.

H22: The degree to which the child sees himself

as being able to talk to his parents will be

negatively related to frequency of child
television viewing.

Family togetherness. A variable related to the one above is

 

the number of activities family members share together. Findings from

the absence vs presence of television studies agree that while

29

families with television tended to Spend more time in the home, the
time spent on non-television activities was reduced. (Himmelweit
1958, Coffin 1955, Hamilton and Lawless 1956, and Maccoby 1951).
Thus, hypothesis 23 for this study is:
H23: The number of activities in which family
members mutually share will be negatively

related to frequency of child television
viewing.

Parent orientation. None of the literature cited here has

 

——

specifically dealt with the degree to which the child sees one or
both of his parents as having done the most fer them. However, this
variable lOgically seems like another indice of the nature of the
child-parent relationship and is cited (Douvan and.Adelson 1966) in
child develoPment literature (usually labelled as emotional attachment
to parents or its opposite, emotional autonomy). One study
(Campbell 1962) found that television children cited their parents
less frequently as "ego ideals." Since most evidence suggests that
junior high school students are just beginning to broaden their
social contacts to outside family members (Remmers 1957), it would
seem that the child who expresses open separateness from his parents.
is indicating a degree of parent-child conflict. Thus, hypothesis 2#
for this study is:
H2“: Those reSpondents who name neither parent

as "having done the most fer them" will

be most likely to be heavy television

viewers; those respondents who name only

one parent will be next most likely; and

those who name both parents will be least
likely to be heavy viewers.

30

Parent decision making. Some studies have tapped various behaviors

 

indicating whether the mother or father dominates various household
activities or whether both parents mutually Share in authority. One
absence vs presence of TV study indicated that in television families,
both parents tended to mutually share in the settling of program
disagreements and selection of family activities (Hamilton and Lawless 1956).
While sparse, this evidence provides hypothesis 25:

H25: Where parents mutually share authority,

their children will be heavier television

viewers than where one parent holds most
of the authority.

Respondent's home responsibilities. A number of prior studies

 

have tapped as another index of the character of the child's home
life, the degree to which the respondent himself Shares in household
responsibilities. Evidence from such child development researchers as
(Hansen (1966) indicates that American children in general Share
minimally in household tasks and that children who have a high involve-
ment in household tasks tend to be more compliant and submissive.
These latter attributes are often applied to children who are heavier
tears of television (Himmelweit 1958)., Specific findings from the
television and children studies indicate that fer the absence vs
presence of television situation, members of television families
1rnded to spend more time on household chores and share the burden

thousehold tasks (Hamilton and Lawless 1956, Belson 1960). The above

31

emdence suggests hypothesis 26:

H26: The number of home reSponsibilities a

child has will be positively related to
frequency of child television viewing.

ReSpondent knowledge of family operation. None of the prior
work reported here has specifically looked at the respondent's
knowledge of such everyday household operation questions as whether
his family has banking accounts, what type of heat is used in the
household, and so on. However, Himmelweit (1958) and others have
hypothesized that television accelerates the impact of adult life,
suggesting that the heavier television user may be more attuned to
such questions. Thus, hypothesis 27 is:

H27: Respondent knowledge of family operation
. will be positively related to frequency
of television viewing.

Parent-child agreement on television program choices. Little
of the prior work has Specifically looked at this. variable. However,
the intersections of evidence provided for several prior hypotheses
suggests that parent-child agreement on television choices Should
relate positively to amount of child television viewing. If amount
0f child viewing is posifirely related to amount of parent viewing, then
the child has, in part, at least acquired his viewing habit from
Parental imitation. Further Himmelweit (1958) and others have pointed
out that heavy television users tend to come from families where

television is used as a child distractor and "babysitter" and where
Childreri's choices, implicitly at least, tend to dominate program

" 1 c . o . ~ - o o o
°e-eCtionS. This then prov1des anor: “313 on which parent-child

television choices should seem in agreement as amount of TV viewing
Idses. Thus, hypothesis 28 is:
H28: Parent-child agreement on television

program choices will relate positively to
frequency of child television viewing.

Respondent's Perceptions of His Own and His Parents Community Integration

 

The fourth category of variables tapped in this study include
reSpondent perceptions of the degree of his own and his parents
integration within the community. The same researchers who have
stressed the need to study media use patterns within the context of
infily life have also stressed the need to look at these behaviors
ydthin the context of the individual's integration into peer and
ammunigy groups. Additional support for the emphasis is offered by
research on teenagers which suggests that friendship relations are
a.major problem for adolescents as they begin for the first time to

mkmch outside their families for their contacts (Remmers 1957).
ufleman (1961) further suggests that there is increasing evidence
that today actual family relationships have less influence on
behavior as adolescents look more to each other and less to adults
fin~their social rewards.

As with the family cohesiveness class of variables, most pre-
(fictions of the relationship between community integration and amount
cfi television viewing are derived from the frustration generalization.
The prediction fer variables which seem to be tapping any type of

alienation would be for a positive relationship between degree of

alienation and amount of exposure. Again, this class of variables
presents a problem in terms of predictions. Much of the literature
suggests that such variables relate to quality of use differences
between reSpondents rather than quantity of use differences (Riley

and Riley 1955). Television may be escape for a low gregarious reSpon-
dent and a method of entertaining friends fer a high gregarious
respondent.

Parent gregariousness. None of the literature cited here has

 

specifically looked at the relationship between parent gregariousness
and frequency of child television viewing. However, the literature
generally shows that the introduction of television led to a reduction
in gregariousness (in terms of interaction with peers informally)

I
for both adults and children (Campbell 1962, Belson 1960, Hamilton
and Lawless 1956). Himmelweit (1958). emphasized that one of the
characteristics of ‘.her "addict" viewers was their less frequent
visiting with friends and their feeling of rejection by peers.
Conflicting evidence, suggesting a non-significant relationship, is
offered by Maccoby (1951) in her television vs no television com-
parison and Troldahl and Costello (1966). Both found no relationship
between time Spent with teenage friends and time Spent viewing.
When gregariousness is Operationalized in terms of participation in
community activities, the bulk of the evidence applies only to the

television vs no television comparison and shows no relationship

(Abrams 1956, Belson 1959, and Himmelweit 1958). Despite the conflict

34

in findings, a combination of the parental imitation and frustration
generalizations yields hypothesis 29:
H29: Parental gregariousness will be negatively

related to frequency of child television
viewing.

Respondent gregariousness. On the basis of the evidence cited

 

for hypothesis 29, hypothesis 30 is:
H30: Respondent gregariousness will be

negatively related to frequency of television
viewing.

Outside home orientation. A good deal of the literature has been

 

concerned with the relationship between television behaviors and the
i

family's use of community resources. Most of the evidence comes

F“

I
rem television vs no television comparisons and shows that television
use did relate to a reduction in attendance at outside home events.
Hypothesis 31 is:
H31: The degree of a family's outside home
orientation will be negatively related

to the frequency of child television
viewing.

Length of time in community. Little evidence is available on

 

this variable. However, Himmelweit (1958) noted that one of the

characteristics of -her "addict" viewers was that they were newer to

their neighborhoods. Based on this sparse evidence, hypothesis 32 is:
H32: Length of time in community will be negatively

related to frequency of child television
viewing.

35

Frequency of attendance at church. Evidence here conflicts

 

with several researchers finding that the introduction of television
lead to a reduction in church attendance (Hamilton and Lawless 1956,
Campbell 1962); others finding no such reduction (Himmelweit 1958);
and others finding no relationship between time Spent on church
activities and time Spent on television (Westley and Severin 1964).
Bailyn (1959) found, interestingly, that Catholics were more exposed
to her "pictorial media", including television, than non-Catholics
and this variable was one of her four strong predictors of exposure.
Despite the conflict in findings, if church attendance is conceptualized
as a measdre of community integration, hypothesis 33 results:

I H33: Frequency of attendance at church will

be negatively related to frequency of child
television viewing.

Respondent knowledge of local and state_public figures. The one

 

tut of evidence available for this variable comes from Schramm (1961)
who found that light television viewer children were more able to,
identify statemen than heavy television viewers. Ordinarily, this
variable is placed in the category of variables dealing with
reSpondent knowledge levels (called the "self-orientation" category
:hxthe present study). However, the variable has been placed in the
community integration category for this study because the respondents
were not actively involved in the study of local and state affairs.
This suggests that their knowledge in the area would more logically
he an index of the degree to which their parents are concerned with

<xmmunity and state issues. Based on the evidence above, hypothesis 3“ is:

HBH: Respondent knowledge of local and state
public figures will be negatively related
to frequency of television viewing.

Self orientation

 

The fifth category of variables in this study is a composite of
'dmachild's reports of his own knowledge levels and interests with his
perceptions of his parents' interests. Most of the major television and
duldren studies have generated a category of variables of this type.
Hmdlton and Lawless (1956), for example, termed this general area the
fiersonality" action area of their respondents' orientations. Others,
mmh as Belson (1959), more simply stated their concern as fecused on
fie interests: activities, and initiatives of their respondents.

Hypotheses for this class of variables are derived mainly from
he functional displacement and information void generalizations out-
lhmd.in Chapter I. When the focus is more on actual activities, such
ashobbies, the functional displacement generalization applies.
Mmever, when the focus is more on knowledge levels, the infermation

vmkigeneralization applies.

figmer of respondent hobbies.

Belson (1959), Campbell (1962), Maccoby (1951), and Hamilton
andiewless (1956) all found a reduction in the number of creative hobbies
Empondents had or the range of activities they participated in after
tmaintroduction of television. Himmelweit (1958) described her "addict"
vhnmrs as being less active and having fewer interests. Thus,

hypothesis 35 is:

 

37

H35: The number of reSpondent hobbies will be
negatively related to frequency of television
viewing.

Number of parent hobbies. By applying the parental imitation

 

generalization to the evidence supporting hypothesis 35 above,
hypothesis 36 becomes:
H36: The number of parent hobbies will be negatively

related to frequency of child television
viewing.

Respondent knowledge levels. The evidence generally suggests

 

that television neither helps nor hurts a child's general information
level (Himmelweit 1958, Schramm 1961), when intelligence is con-
Q

trolled for. Intelligence is an important control here for several
researchers agree that intelligence is one of the single most
important determinants of amount of viewing (Himmelweit 1958,

:fitty 1966, Maccoby 1964, Schramm 1961, Bailyn 1959). When the focus
milearning is changed to Specific classes of infermation, however,
several researchers offer support for predicting significant relation-
ships between knowledge levels and television exposure. Schramm (1961),
fin~example, found that heavy viewers of television were more able

to name bandleaders and pop singers -- figures often represented on
television. The heavy viewers were less able to name faraway places
anistatesmen -- places and people represented more often in the other

media. Some evidence is available on knowledge of math (a measure

tore directly related to intelligence). Scott (1958) found heavy

38

viewers got lower arithmetic scores than light viewers. This
finding is contradicted by Witty (1966), who feund no difference in
math grades between respondents at different viewing levels. DeSpite
some contradictions in the evidence, hypothesis 37 is:
H37: Information levels on advertising slogans

and television characters will be positively

related to frequency of television viewing

while information levels on math will be

negatively related to frequency of television
viewing.

Re3pondent's outside home employment. None of the literature

 

cited here has looked at the relationship of this variable to amount
of television viewing although logic would suggest that the more a
child parks, the less time he should have for television. Work done
in the child development area adds more substance. Clausen (1966)
indicates that boys (at least) who work outside the home are more
peer-oriented, assertive, and high in drive for recOgnition. These
attributes sound much like those used by Himmelweit (1958) and.others
to describe the child who is less likely to be a heavy television
\dewer. Thus, hypothesis 38 is:

H38: The amount of the reSpondent'S outside

home employment will be negatively related
to frequency of television viewing.

 

Number of hours reSpondent studies. Most of the evidence here
comes from comparisons of non-television vs television users and has
generally found no relationship between this variable and television
use (Maccoby 1951, Himmelweit 1958, Campbell 1962, Schramm 1961).

lbne of the literature cited here, however, has looked at the

relationship of the variable to amount of television viewing and
logic would suggest that at some high level of viewing there would be
less time available for study. Thus, hypothesis 39 is:

H39: The number of hours reSpondents study will

be negatively related to frequency of
television viewing.

Frequency reSpondent studies in library. Support for a pre-

 

diction on this variable is derived from that offered for hypothesis 39
above and for hypothesis 16 predicting a negative relationship between
general library use and amoxmt of exposure. Hypothesis #0 is:

H40: The frequency with which respondents study
‘in the library will be negatively related
to frequency of television viewing.

I

Consumer Orientation

 

The sixth category of variables in this study deals with the
reSpondent's family as a consumption unit. Describing American
society as a "consumer" society has almost become a triteism.l Yet,
few of the studies dealing with correlates of amount of television
viewing in children have tapped this dimension of their lives. A
conceptual rationale for tapping this area is provided by discussions
by Schneider and Lysgaard (1953) on deferred gratification patterns
and Schramm (1961) on use of the immediate vs delayed reward
model. For Schneider and Lysgaard, deferred gratification referred

generally to the ability to postpone gratifications or satisfactions.

 

11950 is often used as the baseline date at which consumer
credit began to grow in the 0.8. Huber (1965), for example, states
that since 1950 consumer credit has increased 165% in this country.

 

....

v
I. II

a; -
u.

... -

cv-

   

Inn-i."-

 

 

 

 

 

<~n
- ..
h L;'
'-.‘ ‘

 

 

 

 

 

 

a“
‘ .

'. .‘n
'\‘
3‘

‘

 

 

 

140

Among the attributes of people who can defer are a wish to save
money for the future and wise expenditure of money in the present.
h‘hile they applied their variable mainly to socio-economic class
differences in the early 1950's, it seems lOgical that deferred
gratification may vary widely both within and between socio-economic
class levels. The application of the deferred gratification notion
to media use was first made by Schramm who sees television, in part at
least, as a satisfaction of a need for fantasy on the part of some
children.

In making predictions on consumer variables, a number of the

generalizations outlined in Chapter I seem to be relevant. The

3

frustration generalization, for example, seems applicable if the child
sees his'parents as having a large number of money worries. The
information void generalization might apply to situations in which

the child has picked up an attitude on consumer credit, a topic which
is not usually one which children know about but which television
expounds at great length. The parental imitation hypothesis would
apply, also, in the sense that if the child indicates his parents have
a high consumption orientation this suggests lack of the deferred

gratification pattern and increased television eXposure for both

parents and child.

Number of money worries. In addition to the rationale above,
one study (Huber 1965) indicated that, in comparison with control

C
l

anilies, overextended debtor families viewed television much more.

Ml

Thus, hypothesis #1 is:

H41: The number of money worries the child
reports he has and his parents have
will be positively related to frequency
of child television viewing.

Spend—save orientations. In addition to the basic notions on

 

the deferred gratification pattern, Himmelweit (1958) offer evidence
supporting predictions for this variable. She found, in her‘
absence vs presence of television comparison, that viewers became more
materialistic the longer they had television -- i.e. they became more
interested in things they would like to own. Thus, hypothesis #2 is:
H42: Children who report that they would rather
spend-than-save and that their parents
'would rather Spend-than-Save will be

heavier television viewers than other
A respondents.

Family use of credit. Using the support offered for hypothesis

 

41, hypothesis #3 is:

Hu3: Frequency of family use of credit will be
positively related to frequency of child
television viewing.

A

Attitude toward credit. A combination of the rationales used

 

above and the infermation void generalization suggests a prediction
fin~this variable. If one can assume that the use of credit is not a
'umic which young children often talk about either to adults or peers,
'umn the large amount of consumer credit type advertising which is
Inesented on television should, via the information void generalization

have some impact. Thus, hypothesis ”H is:

42

H44: Respondent attitudes toward credit will be
positively related to frequency of television
viewing.

Variety of respondent Spending. On the basis of the evidence

 

chum (particularly the support for hypothesis #2 on page #1),
impothesis 45 is:
Hus: Variety of reSpondent Spending will be positively

' related to frequency of television viewing.

Family shopping orientation. This variable is descriptive of

 

um geographic width of the family's shopping activities as perceived
the respondent. A restricted range has been shown by such

I
cumumer researchers as Caplovitz (1963) to be a characteristic of the

humping methods of low-income families. It has also been suggested
nun more consumption-oriented families do less comparison shopping.
hum, hypothesis #6 is:

H46: The range of the family's shopping

orientation will be negatively related to
frequency of child television viewing.

Demography

The seventh and last category of variables include demographic-
tmxacharacteristics of the respondent and his family. White (1964)
andcmhers have often termed the use of this class of variables in
Egflaining mass media exposure as the "social aggregates approach".
hathe early work on television and children, this was the most
Pqndar class of variables and, significantly, was often among the

snxmg predictors of behavior. For example, much evidence supports

43

a strong relationship between socio-economic class and various
television behaviors (Schramm, 1961, Bailyn 1959, Samuelson, Carter,
Ruggels 1963, and others). Host researchers have acknowledged, however,
that demography can account for only a certain percentage of the
variance (e.g. Schramm 1961) after which personality, social, and
Other variables must account for differences. Recent evidence
further suggests that the usual differences in media exposure pre-
dicted by demography may be disappearing (Edelstein 1966). Television,
in particular, has become so ubiquitous that long held assumptions
about the relationship of demographic variables to television usage
need reevaluation. The variables listed below are included in the

t
interest of this reevaluation.

I
Socio-economic status. This variable has been termed by

 

various researchers (e.g. Kahl and Davis 1955) as at best a clumsy
variable, a composite of some social and economic attributes that
tend to cluster together. Income, they said, is a poor measure while
occupational status is the best single index. Their factor analytic
study of nineteen socio-economic class measures yielded two major
factors -- one including occupational and educational measures, the
other including ecological measures (e.g. rental value of property).
Income loaded on both factors. This account eXplains some of the
nflict in findings on the relationship of various socio-economic
class measures to television use. DeSpite some conflict in the
findings, however, most evidence suggests that socio-economic class

variables relate negatively to amount of television viewing for both

1+4

ahflts and children. With occupational status as the measure, Scott
(1958) and Bailyn (1959) both found negative relationships with
Himmelweit (1958) finding no significant difference. With education
of household adults as the measure, Steiner (1963), Schramm (1961)

F and Samuelson, Carter, and Ruggels (1963) all fbund negative relation-

ships while Westley and Severin (1954) found no significant difference.

with income as a measure, most studies up to 1955 found a negative

; relationship (Coffin 1955) while Westley and Severin (1964) again
found no difference. One researcher (Geiger 1959) used the more
ecological variable, rent, and again found a negative relationship.
Since most of the evidence agrees, hypothesis #7 is:

hs7: Socio-economic status of parents

. will be negatively related to
frequency of child television
viewing.

Family size. The evidence from television studies for this

 

wndable applies mainly to the introduction of television and
huficates that larger families were more eager for the new medium
(Cmflin 1955, Himmelweit, 1958). LOgic, of course, would suggest
thatin.terms of amount of usage, television would more often be
158d as a child distractor in larger families. Child develOpment
researchers (Clausen 1966) also offer evidence that large families
Emnenate more rules for member behavior and provide less parent-
<fiild interaction -— characteristics which various television

researchers have used to describe the families of the "addict"

4S

television-user child (himnelweit 1958, Naccoby 1954, and others).
Thus, hypothesis 48 is:
H48: Family size will be positively related

to frequency of child television viewing.

Mother's employment status. None of the literature cited

 

here has tapped this variable. However, popular beliefs that maternal
employment provides less opportunity for child-parent interaction and
leads to more anxious and insecure children suggests that a test should
be made of whether the child whose mother works is a heavier television
viewer. Thus, hypothesis 49 is:

H49: The frequency of the mother's employment

‘ will be positively related to frequency
of child television viewing.

I

This hypothesis is offered while acknowledging that recent work on
the correlates of maternal employment shows no relationship to parent
attitudes toward child rearing or quality of parent-child interaction
(Yarrow 1964).

Family type, Whether respondents come from one or two parent

 

families has not been tapped in the literature cited here. With
increasing divorce rates, however, the variable may be more relevant
today than it was 10 to 15 years ago when most television-children
research was done. On the basis of an inference that the child from a
one parent family may, either through social comparison with peers or
through the trauma of separation from one of his parents, be more

anxious and insecure, hypothesis 50 is:

 

 

 

 

 

 

”A‘..-___ ”-i ‘4‘!“ 4%“ -___._._.__._____ <-‘_—-‘

 

46

'f‘
U]

I O: Fespondents from other than two-parent
amilies will be more frequent television
users than respondents from two-parent
camilies.

Birth order. Whether the respondent is an only child, first

 

Irrn, or later born is another variable which has not been specifically
tapped in television and children studies. The rationale for including
it here is the increasing reference to the variable as having some
significance in communication behaviors (e.g. Bdelstein 1966).
Schachter (1959) introduced the variable as a mediator of both anxiety
and gregariousness. His rationale indicated that as parents have more
children, they devote less attention to the latest child, and become
less permishive, less restrictive, and more consistent in their
interactions with the child. Schachter sees the first born child as
being raised in a manner that would reinforce dependent behavior,
requiring more social approval and support. Here again is a list of
attributes which sound like those such researchers as Himmelweit
(1958) have used to describe their addict television viewers. While
the leap from Schachter's work is, perhaps, tenuous, hypothesis 51 is:

H51: Only and first born children will be more

frequent television users than later-
born children.

ng, The last variable in the demographic category is respondent
sex and is the one variable on which all available evidence agrees
one hundred per cent. No significant difference has been found
between boys and girls on the amount of their television viewing

(Merrill 1961, Baxter 1960, Himmelweit 1958, Schramm 1961).

L+7

Bailyn (1959) did find that boys were significantly more eXposed

to ”the pictorial media” but this difference was accounted for mainly

by attention to comics not television. Thus, sex is included here

in the interests of replicating an often-tested variable but no Specific

hypothesis is formulated .

CHAPTER III

METHODOLOGY

This study is drawn from a large set of available data originally
mfllected by the Center for'COnsumer Affairs, the University of
ifisconsin Extension, University of Wisconsin - Milwaukee. Purpose of
the original survey was to collect some eXploratory data on children's
knowledge of consumer and homemaking practices as well as to tap various
aspects f their media use and life styles. The original questionnaire
its developed in cooperation with Milwaukee area public school home
economics teachers and pre-tested on'a judgmentally selected sample
@650 junior high school students attending public schools in the
bulwaukee metropolitan area.‘ The questionnaire was revised based on
pre-test results and then administered to all the 7th and 8th grade
students in one suburban school system outside the city of Milwaukee.
ihdle analysis of marginals was completed in the summer of 1965, no

prior reports of the survey have been published.

Questionnaire administration

The questionnaire was administered by teachers in the suburban
school system in three phases -- one in October 1964, one in
Zbcember 1964, and one in February 1965. During each phase,
approximately 90 students -- half from the 7th grade and half from

the 8th grade -- completed the self-administered questionnaire.

48

 

49

he questionnaire took two 50—minute class periods to complete and

included 212 dif erent items, many containing multiple sections.

'Rm respondents and their community

 

ReSpondents included all the students in the suburban
ammunity's public seventh and eighth grades. Eliminating absences
hmtimated at less than 5%), 270 students completed the original
<pmstionnaire: 122 seventh graders and 148 eighth graders; 143 boys
aw1127 girls. The respondents ranged in age from 11 to 14 years - 4%
vmre 11 years old; 36% were 12; 49% were 13; and 11% were 14.

In 1960, the suburban community in which these respondents
lived was pOpulated (U.S. Bureau of Census figures, 1961) by some
36,000 persons, of whom more than 99% were Caucasian. Median family
hxmme for the four census tracts covered by the community ranged
from $7646 to $12,414 and the median value of a housing unit ranged
from $17,400 to $25,500. The median education obtained by adults in

'nm community ranged from 12.4 to 14.2 years.

Item selection

 

From the original set of 212 items in the Wisconsin survey,
filwere used for the purposes of this study. Since many of the
mfiginal items contained more than one Specific reSponse from each

mspondent, a total of 147 different responses were actually coded

l—J.

Ilth€ first phase of the present study. These 147 responses were
selected to tap the variables for which predictions are made in

Gmpter II. The 147 responses were then coded, punched and verified

—"-_. v.

50

onto IBM cards, and then reduced (with indexing and deletions) to the
final set of 62 variables. Complete details on these Operations

,. a 1 "
IOllOw o

Inmzmeasurement

 

 

 

The original Wisconsin survey included both open and closed items.
About one-quarter of the actual items selected for the present study
involved content analysis of open-ended reSponses. For the rest of
the items, the original survey had detailed reSponse categories from
which codes were easily derived.

In developing a coding scheme for all items, reSponse codes
were cast in a continuous fashion with as extended scales as possible.

4

A few variables are measured as dichotomies on "yes vs no" scales. An
1

e:-:anple of such a variable is the reSpondent's rating of media

credibility which was coded as "checked television vs did not check
television." Most variables may be seen as on the ordinal level of
measurement. An example is parent television viewing for which
response codes were "often, sometimes, once in a while, or never."
Some variables in this study have extended scales that meet interval
level of measurement assumptions. An example is the criterion
variable -- amount of respondent television viewing -- which taps
respondent viewing of up to 26 television programs.

Strictly speaking, the rcplication purpose of this study did not
require the kind of measurement outlined above. Most of the past
television studies have used contingency analysis for their data and

have included many nominal or category-type variables. However, the

51

;-esent study intends to go beyond a straight replication to a
multivariate, correlational analysis. In order to meet the assumptions
of a correlational analysis, an attempt was made to cast each variable
in a continuous manner (or, at least, so underlying continuity could

be assumed) and to extend the range of the variable as far as possible.

Item coding

 

Appendix A lists the original wordings and codes for the
humpendent or predictor variables used in this study. Full details
mithe dependent or criterion variable are reported in a later section
ofthis chapter. In Appendix A, variables which required content

amlysis are starred with an asterisk (*). The original coding for
Q

q

dds study was done by a team of four coders, working in pairs. Each
ofthe 147 hriginal items was coded.twice, once by each member of a
;nder pair. Iterjudge coding reliabilities were calculated for all
items across all reSpondents. The measure of reliability used was

'nm percentage agreement index:

Total # of respondents - # of coder disagreements

 

Total # of reSpondents

The formula produces a percentage, for which the a priori
muterion of acceptability was set at .90. Appendix A lists the
percentage agreement indexes for all items used in this study -- both

:mose that were used alone or those that were collapsed into indexes.

52

hfiiabilities ranged f.om a low of 89% to a high of 100% with most
ring in the range of 97 to 99%. For most items, the agreement
afiterion was set for agreement on an exact code. For example, in
ommting the total number of television programs a reSpondent said
maviewed regularly, one coder might have calculated 1?, the other 18.
as discrepancy was counted as a coder disagreement. On one variable,
Um criterion was relaxed to agreement within one point. That
vafiable was occupational prestige, measured on a 12 point scale. If
mm coder rated a family's occupational prestige as 8 and the other
mned it as 9, this was not counted as a disagreement. However,
rmfings of 8 and 10 by two coders was counted as a disagreement. The
mason for relaxing the agreement criterion for this single variable
vasthe high.degree of subjective judgment involved in coding

ocmmational prestige.

Etuction of items by deletion and indexing_

 

From the set of 147 different reSponses coded in the first
pmme of this study, 103 were finally used in this study. The remaining
M+were deleted for one of two reasons: 1) the item lacked sufficient
vadance for statistical analysis; or 2) the item did not interrelate
widzother items for the purposes of index construction and was deleted.

Complete details on the operations involved in constructing
imhxes for this study are outlined in Appendix B. Briefly, the
omnetions involved first making a priori decision of what items seemed
tobe tapping similar behaviors. This a priori decision was then

tanked by running Pearson product moment correlations between the

 

 

53

items for each set of potential indexes. Items which were
significantly intercorrelated were then indexed by simply summing
scores across the set. In addition to giving details on the general
indexing procedures, ppendix 3 lists the operations involved for
constructing each of the final 13 indexes and the correlations for
items summed for each index.

After item deletion and indexing, the final 62 predictor
variables used in this study remained. Appendix C lists these variables
with their means, standard deviations, non-reSponse counts, and code

ranges.

Respondent deletion

r

 

Before indexes were constructed, 18 reSpondents were deleted
from the o;igina1 set of 270 respondents who completed the Wisconsin
"rvey. Each of these respondents was excluded because he had a non-
response level above 10% for the original 147 reSponses coded. An
ana-ysis was made of the effect of the deletion of these respondents
on the criterion variable and ten other randomly selected variables.
Results showed that reSponses of the deleted respondents were randomly
distributed -- i.e. they represented fairly equally all levels of

television viewing, both sexes, all levels of socio-economic class,

and so on. The final n for this study, then,lis 252.

Description of the criterion variable

 

The original Wisconsin survey asked reSpondents to check off which

H1
M

o- 6 different television programs they watched regularly. The 26

different shows represented a total of 26 hours of television viewing

54

a week or almost 4 hours a day on the average. The original items were
worded as follows:

”Which of the following TV shows do you watch regularly?"

 

 

Andy Griffith Petticoat Junction
—-—Dick Van Dyke -_—Uack Benny

Ed Sullivan ———Donna Reed

Patty Duke ._—Beverly Hillbillies
__—Danny Thomas -__Lucy Show
'__—Red Skelton __—I've Got a Secret
—__Ny Favorite Martian :::thale's Navy

Lassie Bonanza

"Do you watch any of the following shows?"

Ben Casey Dr. Kildare
—_—Breaking Point -——Richard Boone
——_Fast Side West Side -_—Bob Hope
‘-_—Defenders -__Bleventh Hour
::::Danny Kaye —_-Late movie

I

The first 16 shows listed include the top 16 shows of Fall 1964
vmen the Wisconsin survey was first launched. The second 10 shows
lsted were a judgmental selection of adult drama shows that were
nxeiving considerable attention in the Fall of 1964 but were not in
the top viewing ratings.

While these two lists of television shows are unlike any
neasures of television viewing Specifically used in past studies, they
seem most like the "aided recall" type of measure that Schramm (1961)
naked about in the discussion of measuring exposure time to television.
Khua11y, of course, aided recall measures list all the television
programs for a specific time period and ask the reSpondent to check

amich ones he watched. Instead the measure available for the present

55

study lists a selection of the more popular shows. According to
Schramm (1961), the aided recall method of tapping television viewing
was one of the better measures in terms of its correspondence with
a check of actual time spent viewing.

Original coding of the amount of television viewing measure for
the present study was simply a count of the number of different programs

. . l
checked With codes ranging from O to 26. One major problem was

readily apparent, because the original Wisconsin survey had been
administered over a five month period with one administration in
October 1964, one in December 1964 and one in February 1965. While
the list of shows in the questionnaire checklist remained constant,
actual television programs changed as the television season progressed.
is expecged, a check of the means of respondent scores on the two
television checklists showed a significant difference between the
October, December, and February administration times. Tables 1 and

2 on the fellowing page report the mean scores for the first checklist
(the checklist of the top 16 shows) and the results of the analysis

of variance used to test whether these means were significantly
different. Tables 3 and 4 (on page 57) report the same information

'for the second checklist of 10 adult drama shows.

 

The interjudge coding reliability fer the criterion variable
was 97%. See page 51 of this chapter for an explanation of coding
operations and reliability checks.

56

Table 1.

Means for respondent scores on the top 16 television

show checklist, reported by questionnaire administration date.

 

October 1964 administration

December 1964 administration
February 1965 administration

Mean
7‘3???
6.68
5.45

 

hmle 2. One-way analysis of variance table testing the difference
between scores on the top 16 television show checklist
by questionnaire administration date

A
V

 

Source of variance df ss ms F p
I
Between 2 238 119.00 9.31 p < .001
Within 262 3347 12.77
Total 2641 3585

 

1
Total n for this test was 265.

The test was made befOre the

respondent deletion procedure described earlier in Chapter II. Five
of the original 270 reSpondents did not answer these two television

use items.

57

 

 

Table 3. Teens for respondent scores on the 10 miscellaneous
adult drama television show checklist, reported by
questionnaire administration date

Mean
October 1964 administration 5.78
December 1964 administration , 5.15
February 1965 administration 4.65

 

Table 4. One-way analysis of variance table testing the difference
between scores on the 10 miscellaneous adult drama
television show checklist by administration date.

 

 

Source of variance df 33 ms F p
Between 2 94 47 5.78 p< .01
IEthin 262 2129 8.13

Total ~ 264 2223

 

Results of this analysis show that, for the top 16 show checklist,
neans for the three time periods declined from 7.67 for the first
adunistration to 5.45 for the last administration. The difference

hemmen the three groups is significant (p (2001). For the second

58

checklist of 10 adult drama shows, again the means declined from

5.78 in the first adminis ration to 4.65 in the last administration.
The difference between the three administration groups is significant
(p (.01).

To make sure that this difference between administration dates
zesulted only from television prOgramming changes as the year progressed
auiwas not an inherent difference between respondents, analysis of
wndance tests were also made on three additional variables: 1) variety
ofreSpondent newspaper reading; 2) frequency of reSpondent radio use;
and 3) frequency of respondent movie attendance. For those three

wriables, the differences between mean scores of reSpondents in the

. 4 . . ‘ o o o o o
sree administration time groups were not Significant.

rt-

To compensate for the difference that administration times made

I hirespondent television use scores, the scores (which resulted from
smmung the total number of shows checked on both the checklists described
amve) were standardized within administration groups. A check of

mm shapes of the distribution of the television use raw scores

hxucated that these distributions closely approximated normality and,
'dmreby, met the criterion for using standard scores (McNemar 1962).

kmher than use the usual standard z score which involves positive
auinegative values, scores were standardized to distributions with

:eans of 50 and standard deviations of 10. The standard scores ranged

firm 24.2 to 81.4.

59

Statistical analyses

 

In review, the basic purposes of the present study are:

1) To replicate much of the past work on correlates of frequency
of child television viewing;

2) To go beyond a replication by including some variables
which have been partly or wholly overlooked in past works;

3) To then extend the analysis to a multivariate method; and

4) To Specifically look at the types of relationships which
exist between the criterion variable and its predictors.

To accommodate these purposes, fOur different types of
statistics are reported in Chapter IV. In analyzing the relationship

I

of any one predictor variable to the criterion variable, three measures
are used: 1) contingency crossbreaks with chi-square values;

2) Pearson product moment correlations; and 3) Eta, the curvilinear
correlation ratio. For the multivariate analysis, one measure is

used: multiple regression or multiple linear correlations. Full
details on the operations, assumptions, and rationales for these

analyses follow. All reported Significance tests are two-tailed.

Chi-square contingency analysis. With rare exceptions, the

 

technique used for analysis in most of the past television and children
studies has been contingency analysis of frequencies. For this
reason, contingency analysis is used here to Study the relationship

between each of the predictor variables and the criterion variable.

60

Since most of the variables in this study were measured on as

extended continuous scales as possible, the use of contingency
analysis required that variables be collapsed. In all cases,
variables were collapsed empirically rather than judgmentally -- i.e.
after examining the marginal distributions for each variable,
respondents were divided into groups by a quintile, quartile, tertiary,
<u~median Split. The maximum Split used was one which created five
categories on a variable Since any larger number of categories reduced
cell expecteds to a point where the use of the chi-square distribution
becomes suSpect (McNemar 1962). The criteria used for collapsing

each variable involved two Operations: 1) an attempt to have as large

I
a number of categories (up to 5) as possible; and 2) an attempt to have

rt

he total‘number of respondents as equally divided among categories

3 possible.

m

Special mention should be made of the reduction of the criterion
variable to four categories. The variable in its standard score form
tad scores ranging from 24.2 to 81.4, approximately normally distri-
buted. The variable was collapsed into quartiles, which have been
labelled: 1) low frequency of television viewing; 2) moderately low;
3) moderately high; and 4) high. In the result tables in Chapter IV,
these four levels of viewing will often be referred to by initial as:

L (low); ML (Moderately low); MH (Moderately high); and H (high).

In splitting the criterion varhble, ties at quartile deviation
points were allocated to viewing levels so that n's would be as nearly
equal as possible. The actual split produced the following distribution

A: w-I—aA—r. 57-. ,I -‘ 0 ‘1’... fl 0 - . U ”.-
9- - 3;: Jae-eats: ‘1, do , .-.J, 0", ban-1’ 07, Eric 4., 03 o

61

Because the quartile Split on the criterion was done on the
standardized score version of the variable rather than the raw scores,
the resulting levels of viewing are not directly translatable into
number of shows watched regularly. The significant difference in
frequency of viewing times across administration groups makes a precise
translation impossible. As a rough guideline, however, the levels of
viewing correspond roughly with the following number of Shows checked
on a 26 TV program checklist: L, 0-7 shows; ML, 8-10 shows; MH, 11-15 '
shows; and H, 16-26 Shows.

All contingency analyses reported in Chapter IV involve the

pitting of one predictor variable against the criterion variable.

31
And-‘-

OJ

nalysefs were done by computer. Significance criterion was

preset at p<’.05; relationships reaching p<‘.10 are also reported.
Tabled n's will vary somewhat because for this aspect of the analysis,
respondents with non-responses on the predictor variables were
deleted. The number of non-responses for each variable is in

Appendix C.

Pearson product moment correlations. Since oneof the purposes of
this study is to look at the nature of the relationship between each pre-
dictor variable and the criterion, Pearson product moment correlations
51‘9- reported. These will be analyzed in two ways: 1) in terms of the
direction of the relationships; and 2) in terms of the variance accounted
for in the criterion by the predictor. Variance accounted for

17‘] this linear model will be compared with variance accounted for by a

62

curvilinear model (the Eta correlation ratio described below).
Computation of the product moment correlations for the results
section of the present Study was done by computer. As noted earlier,
all variables were cast on as continuous and extended scales as
possible in order to meet the measurement assumptions of product
moment correlations. Whenever non-responses occurred on items, the
:ean value of scores for that item was filled in. Appendix C lists
allitems, their means, and the number of'non-responses. USing the r
«>2 transformation test of significance (McNemar 1962), the critical
‘mlue of r at p<(.05 fer an n of 252 is .12. All correlations
:flgnificant at or beyond the .05 level will be noted in the findings

sections. I

Eta curvilinear correlation ratio. Eta or the correlation ratio

 

is a measure which taps degrees of relationships in general, whether they

U'

o linear or curvilinear. If a relationship is actually linear, the Eta
auithe product moment correlation will be the same. If the relationship
hsactually curvilinear, the Eta will be larger than the product moment
anrelation. (McNemar 1962).

On the basis of this reasoning, Etas are reported for the
mflationship of each of the predictor variables to the criterion

vanumle. Since there are two Etas for each relationship, the

63

specific Eta reported here is the one which taps the accuracy with
which the criterion variable (Y) can be predicted by the predictor
variable (X).1

One question that might be asked is why use Eta when the
contingency analysis described earlier taps curvilinearity. One
problem with contingency analysis and the use of the chi-square
distribution, however, is the constraint of needing adequate cell us.
This requires collapsing variables, often to such a degree that.
there is not sufficient Spread for tapping curvilinearity. Since Eta
:3 computed from an analysis of variance model, the constraint of
sufficient ns is not as restricting -- i.e. for each value of a
wuiable, a minimum number of reSpondents is needed in order to
enable cabculating a measure of variance. By this logic, Eta is a
:easure more directly comparable to the linear product moment
unmelation which taps all values of each variable in its computations.

The calculation of Etas for this study was done by computer.
h1the process, each variable was standardized and then broken into
muegories, each 1/u standard deviations in width. Purpose of this
grocedrre is to enable the calculation of the one-way analysis of
vafience from which Eta is derived. Ti atzent of non-reSponses for
'fie Eta analysis is like that for the product moment correlations --

amen values for each variable replaced all non-responses. Thus, the

 

l - .
The other Eta taps the accuracy of predicting in the opp031te
ditxmion -- the accuracy with which X may be predicted by Y._

64

n for each Eta was 252.1

In order to compare the predictive power of a linear versus
a curvilinear model, the procedures used here was to square the r and
the Eta for the relationship of each predictor to the criterion. The
squaring process produces comparable figures -- percentage of
variance accounted for in the criterion by the predictor. With such a
comparison, the question must be asked: by how much must Eta exceed r
before we reject the notion of linearity and accept the notion of
curvilinearity? McNemar (1962) reports an analysis of variance method

er the significance of the difference between product moment

A
5‘)

C)
'5
d
(D
U)
r f
F:

correlations and Eta correlation ratios, These tests have not been

performed for the present study and comparisons are made on an intuitive
I

One note of caution must be added. To the extent that Eta is
larger than r, we obtain an indication that a curvilinear model may
better fit the data. However, Eta says nothing about the nature of
the relationship. For the present study, gleanings about the nature

of the relationships may be obtained from the contingency tables.

Multiple correlation. One purpose is to use a multivariate

 

 

1Unlike the computation of r, the computation of Eta does require
some collapsing of the variables involved. The question may be raised
whether the Eta derived from slightly collapsed scores is directly
comparable to the r derived from actual raw scores. (McNemar 1962)
a specific check on this problem was made by comparing r's derived from
raw scores with r's derived from the categories used in the computation
of Eta. In all cases, the two r's were equal.

65

technique to determine how much variance the 62 predictors account for
in the criterion variable. The multivariate method used here is the
multiple linear correlation or multiple R. The reason fOr the choice
of this particular method is, first, that it is one of few
multivariate techniques available. Secondly, it is the most efficient
in terms of ease of prediction for it assumes a linear model. While
not all predictor variables will be related in a linear fashion to the
criterion variable, the multiple R analysis delineates which variables
predict ”best" assuming linearity. The loss, of course, is that some
variables with a non-linear but high relationship to the criterion
variable are necessarily deleted in the multiple R analysis. (McNemar 1962)
.

Several multiple R analyses were done for the present study.
First, multiple correlations were run within each variable category
(as outlined in Chapter I and II). This operation answered the
question: which of the variables within each variable category accounts
fbr most of the variance in amount of child television viewing? In
the second phase, the best predictors from each category were analyzed-
to determine which of the category predictors were "best" overall.

The multiple correlation analyses were done by computer. The
procedure involved first computing a multiple R using all the variables
for a particular analysis as predictors of the criterion variable. The
question of which variables are the "best" predictors was then answered by
a least squares deletion routine. In this routine, predictor variables

are deleted one by one according to which of the variables contributes

CHAPTER IV

RESULTS

In presenting results, the following format is used. Results
are presented first within variable categories. For each variable
category, the discussion begins with a table presenting the relation-
ships of all variables to the criterion variable. Results are then
discussed in terms of those variables for which hypotheses were
confirmed and those which were not. The report for eadh variable

catesory ends with a comparison of the linear correlations. model

g?

to the curvilinear model and a report on the multiple correlation
.I . .. .
analySis within that category. The chapter ends With a report of the

multivariate analysis between variable categories.

ReSpondent_perceptions of parent media use

 

Table 5 summarizes the results fer individual hypotheses on
the relationship of parent media use variables to frequency of child.

television viewing. The results indicate that of ten variables, four

m

re significantly related to frequency of child television viewing --
three in the direction predicted by hypotheses in Chapter II and one
in the opposite direction., The variables which are not significantly
related to child television viewing are: number of television sets,
amount of parental radio listening, number of newspapers: dailies,

number of newspapers: weeklies, number of magazines parents read,

67

66

the least to eXplaining variance in the criterion variable. The
process allows variables to remain only if they account for a
significant proportion of the variance in the criterion variable over
that accounted for by the other predictor variables and the mean of
the criterion variable. Every variable which does not contribute to
explanation at p<f.05 is deleted.

In reporting the results of the multiple correlation analyses,
the best predictors for each analysis will be listed along with the
significance level reached by each multiple R. These significance
tests were completed by computer with an analysis of variance for
overall regression technique. ReSpondent ns fer the multiple
correlation analyses is constant at 252. Non-responses on any items

were rc;laced with the mean value of the scores on that item.

In presenting results, the following format is used. Results
are presented first within variable categories. For each variable
category, the discussion begins with a table presenting the relation-
ships of all variables to the criterion variable. Results are then
discussed in terms of those variables for which hypotheSes were
confirmed and those which were not. The report for each variable
category ends with a comparison of the linear correlations. model

!
to the curvilinear model and a report on the multiple correlation
analysis within that category. The chapter ends with a report of the

multivariate analysis between variable categories.

if]

respondent perceptions of parent media use

 

Table 5 summarizes the results for individual hypotheses on
the relationship of parent media use variables to frequency of child .
television viewing. The results indicate that of ten variables, four
are significantly related to frequency of child television viewing --
three in the direction predicted by hypotheses in Chapter II and one
in the opposite direction. The variables which are not significantly
related to child television viewing are: number of television sets,
amount of parental radio listening, number of newspapers: dailies,

number of newspapers: weeklies, number of magazines parents read,

67

68

Table 5. Summary of results for individual hypotheses on the
relationship of parent media use variables to frequency
of child television viewing.

 

 

 

Variable Pre- Variance

dic— 2 accounted for

tion X p r. . Eta

P2 Eta2

Amount Of P TV viewing 1' (.05 .153 .30 .02 .09
Variety of P TV viewing + (.05 .10 .33 .01 .lO
Number of TV sets + n.s. .07 .26 .00 .07
Amount of P radio listening - n.s. b-.02 .32 .00 .10
Variety of P radio use ' (.001 .31a .U3 .09 .18
Number of newspapers: dailies - n.s. -.02 .30 .00 .09
Number of newSpapers: weeklies ' n.s. ~03 .28 .00 .08
Number of magazines P read - n.s. .05 .27 .00 .07
Parental sources of news + (.10 .15a .38 .02 .14
Number of phonographs + n.s. .01 .32 .00 .10

I

 

aSignificant at least at p<1.05. Critical values of r's, by r to
transformation test of significance, are: p.(.05 = .12; p‘<.01 = .15;
p (.001 = .19.

DWhile results for this variable are significant, they are in a direction
Opposite to that predicted.

 

and number of phonographs. For each of these variables, neither the

X2 nor the Pearson product moment correlations reached a significance

level of p< .05. Table 7 (starting on page 73 ) presents the contingency
crossbreaks and chi-square results for each of the predictor variables

in this category. Details on the variables which are significantly related

to the criterion variable follow.

a- . I I I 2 0
amount or parental teleViSion Viewing. The X for the contingency

 

table comparing ”mount of parental television viewing against fre-

quency of child viewing is significant at p<1.05, with amount of

parental viewing positively related to frequency of child viewing.

This positive relationship is confirmed by the significant (p<(.01)correlation
of .15. The hypothesis for this variable is confirmed: respondents

who see their parents as viewing TV less than "often" are significantly

more likely to be lighter television viewers than reSpondents who see

their parents as viewing TV "often". I

. 2 .
Variety of parental television viewing. The X for this

 

variable readhes significance at p<:.05. While the r of .10 does not
reach significance at p.;.05, the direction is positive. The positive
direction of the relationship of variety of parental television
viewing to frequency of child television viewing is supported by the
distribution of percentages in the contingenCy table. Respondents

who see their parents as viewing less variety in television are
significantly more likely to be light television viewers than reSpondents
who see their parents as viewing more variety. Thus, the hypothesis
predicting a positive relationship between variety of parental
television viewing and frequency of Child television viewing is
mnfirmed.

Variety of parental radio use. The X2 for this variable is

 

highly significant (p.(.001). The direction of the relationship

between this variable and the criterion is positive as indicated by

70

the significant (p4;.001) r of .31. The positive relationship is
supported by the distribution of responses in the contingency cross-
break: higher levels of child television usage are significantly
associated with higher variety of parental radio usage. While results
show a significant positive relationship, the hypothesis for this
variable predicted a negative relationship to frequency of child
television viewing. Two other radio use predictor variables in this
study show significant relationships to the criterion variable in a
direction Opposite to that predicted. A possible rationale for these
unexpected findings is in the next section of this chapter in the
discussion of‘the variable, variety of respondent radio usage.

Parental sources of news. The X2 for the relationship of

 

this variable to the criterion is not significant according to the
a priori p(.05 criterion. However, the X“!2 does reach the p<.10 level
of significance and the r of .15 is significant at p‘(,01. The
crossbreak itself suggests the positive nature of the relationship:
respondents who indicate that TV is their parents' source of news
tend to be heavier television viewers than respondents who indicate
other media are their parents major source of news. On the basis of
the significant r, the hypothesis predicting parental use of television
as a source of news would be positively related to frequency of child
television viewing is confirmed.

Comparison of linear and curvilinear models. The summary table

2 and Eta2 or

 

for this category of variables (Table 5) reports the r

'wndance accounted fOr in the criterion variable by each predictor

variable with a linear versus a curvilinear model. Without ex-

ception, the variance accounted for by Eta, the curvilinear

O

orrelation ratio, is greater than that accounted fer by r. This

discrepancy in variance accounted for ranges from a low of 7% to a

high of 12%. While the linear r's account fer from 0 to 9% of the

variance, the curvilinear Etas account for from 7 to 18% of the

variance.

Multiple linear correlation analysis. While the above would

 

suggest that a linear model might not be the best fer these data,

nevertheless, a linear model still remains the most efficient multi-

variate prediction method. For this reason, a multiple R.was computed
‘

among variables in this category.
a
computation.

Table 6 reports the results of that

 

Table 6. Results of multiple linear correlation analysis for
the parent media use variable category.

 

Value of Multiple Correlation (R)

Variables retained after least
squares deletion

 

With all 10 With variables Variable r to DV

variables retained after

included in least squares

eguation deletion

a a b

37 ..3u Amount of parent TV .15
viewing
Variety of parent b
radio use .31

*—

aSignificant at p ( .0005.

b . . . . . . . .
The partial r's for tne relationship of both variables to the criterion

variable equal the r's.

72

Assuming a linear model, the 10 parent media use variables in
combination accounted for about 14% of the variance in frequency of
child television viewing. Variables were then deleted from the

multiple R equation by the least squares criterion. At the end of

1

the deleti . process, the resulting multiple R equalled .34, accounting

V

for 12% of the variance in the criterion variable. The two variables
retained after the deletion were: amount of parental television viewing

and variety of parent radio use.

 

0mm

mm 0: am ma ma awn:
am mm mm mm mm . can; samumnmeo:
am am on ma mm 30H samumsoeoz
8. v m 8.: 2. n: an in am as on :3

 

mewsow> cowmfl>caop
Hmucopma mo hpowmm>

 

 

73

 

 

 

 

 

mmm

Hm mm mm mm ma coumo

onaonOm

mo.uv m ma.m Had “seam am am saw .maaa: m an mono .em>mz
mcmsow> COMmm>maou
Hepconma mo pesoe<

a he ax : as a: a

memsonuwzo mewsow> :Qwa>oaow Udaeo mo >UCmsvmnh mabmwnm> mouowomnm

 

cownowwnu may 0% moanmwem>
on: wanes unseen 029 MO mangprmHmn use new wonam> osssvmlwco ocm.moanm#_ho:oucflpcoo .b adamh

7w

E

 

 

 

 

 

 

 

 

 

a: on mm as a cane
am am am am ea awn; sambmpmeoz
mo ea mm mm .mw 20H sampmnmeoz
89V m 3.3 am "a 2 mm mm ram :3
mm: oflomn
Hmpcmemm mo zvowmm>
«mm
oma nu ma am am embmo
as an on Hm ea washbasom
.m.: a am.e a: u: as am am sad manage an mono .eosmz
meadowmwd
Owens Howsmmmm mo pcsoe<
men
and mm mm mm om pew one can» 20:
.m.: m mm. add": mm mm as an“ new coawa>mamb mco
oaocomso: Qfi meow
coflmw>mamp mo nonesz
a we ax : mm a: a

one: mlwco

measow> cofluw>oaop tawzo mo %o:oswonm

OHQW ..nrHfluxr ..HOPU MU gnu

Umicwuazoo I... N. Gang

h:m

 

 

 

 

 

 

 

 

 

me mm mm as ea names on o>wd
a: ma on ma mm 950m
we Hm mm hm ma omega
.w.: m m>.mH mm u: mm mm om rm“ 039 ow ocoz
one,» we comma
mocwwmmme mo Lobesz
mmm
mma mm mm mm mm 990% ow 0:0
.m.: m mm.m man: am. am Hm saw meoz
.3
7. moflaxoos "0p cobweombsm
whomwmmzo: mo nonesz
mmw
med um mm mm mm m>ww op 039
.m.c m mm. m:au: mm mm hm wan oco no 0:02
wowawme “ow cobwbombsm
mememawsoc we nmbEsz
a an «x z :m z a

onmsomzwco

mcflsow> coflmw>oao+ pawzo we homosdoem

oaemfimm> mowofioenm

Unscwwcoolxh vague

 

 

.moabmp one mmonom wcwomom waoa pew mm coospmn Hmpop hoe .onomononp .ocm mmo omecsom coon o>mc wommpcoonom

.30 .m ”we .mz was .42 mam .q "mam maagmw.mmmau an wao>ma

wcwzow> mew mom m.: Enamcwz .mm .: mum .32 Maw .42 mom .4 one wcwzow> mo mao>oa mcwmnm> any mom m:
.Aouwm ransom Eseflxmev .mmm u : Hmuoy cons .omyoamo ones moabmwmm> nouowooem new so momcoemmnncon
omsmoon moanm> means“ one mom a Houoe .cmwn u : .zmfln mampmnmeoe u m: msoa maopmnmeoe u a:
amnesmw> 30H u A .HHH ampameo cw ombflpomop mm meaflunmsw Opr oomamaaoo mm: oabmwem> cowmovwmo mean

A

 

 

 

 

 

 

 

0mm
am am mm am an names on wages
mm :m :m an :N 039
.w.c w mh.w hm u: mm mm mm wnm . mco ow ocoz
% @056 mcmmnwocozm
mo Lonesz
Hmw
hb mm mm mm pH >9
Aoa.vv m :o.w :bau: Hm mm mm ram >9 cosy mozwo
O m 0 c
wSGG
mo moonsom Hopcwpmm
a we ax ii use as a
oemsvmufico mewsow> newmw>®H0p UHHLO we mocmsvohm manmwpm> powowomgm

UUZCflaccoilh mfladh

 

 

77

-espondent Media ‘se

 

Table 8 summarizes the resu ts for individual hypotheses on
the relationship of respondent media use variables to frequency of

child television viewing.

 

Table 8. Summary of results for individual hypotheses on the relation-
ship of reSpondent media use variables to frequency of
child television viewing.

 

 

 

Variance

Variable Pre- 2 acgounted fpr»

diction X D r Eta r‘ Eta[_
Anemit of a radio use - an: .1251 .35 .01 .12
Variety of. R radio use - <.01 .25“1 .uo .06 .16
Etriety of R newspaper use - n.s. .ou .32 .00 .10
Number of magazines R reads - n.s. .05 .31 .00 .10
lariety of R book preferences + (305 .18a .41 .03 .17
Eipreference for comics + <,02 .213 .3” .04 .12
R library use , - n.s. -.08 .32 .01 .10
Frequency of R mov1e attendance + n.s. .173 .36 .03 .13
Variety of R record preferences + n.s. .ou .31 .00 .10
redia R would miss most + n.s. .15a .28 .02 .08
2V3 most believed media + n.s. .09 .33 .01 .ll
Rls least believed media - n.s. -.08 .3“ .01 .12

 

5Significant at least at p.<.05. Critical values for r's, by r to z
transformation test of significance, are: p.<.05 = .12; p‘<.01 = .15;
p< .001 = .19.

DWhile results for this variable are significant, they are in a
direction opposite to that predicted.

 

78

km results indicate that of 12 variables, six are significantly related
n>frequency of child television viewing. Four of these are significant

mzboth the X2 and r analyses; two have non-significant X25 but

significant rs. The variables which are not significantly related to
we criterion variable by either statistical analysis are: variety of
impendent newspaper use, number of magazines reSpondent reads,
nepondent library use, variety of respondent record preferences, and
:espondent media credibility ratings (most believe and least believed
India). Table 10 (starting on page an) presents the contingency cross-
meaks and chi-square values for each of the predictor variables in

ads category. Details on the variables which are significantly related
tithe criterion variable follow.

amount of respondent radio use. The X2 for this variable is

 

25.;fnificant at p4 .01. The r of .12 is also significant (pC .05).

Any

lationship is positive with frequency of child television

a
’3
(b

fiewing rising significantly as amount of reSpondent radio use rises.
Rule the results are significant, the hypothesis for this variable
medicted a negative relationship between amount of reSpondent radio
we and frequency of television viewing. This is the second radio use
uniable (variety of parental radio use in the preceding section is the
onmx0 which has related to frequency of child television viewing in
adirection opposite to that predicted. A possible rationale for

its counter finding is offered below in the discussion on variety of

remxmdent radio use which also relates significantly to the criterion

lfifi‘ble in a direction Opposite to that hypothesized.

79

Variety of respondent radio use. The X2 for this variable is
significant at p*<.0l with the distribution of reSponses again showing
a positive relationship. The r Of .25 is significant at pv(.001. In
_general, then, as variety of reSpondent radio use rises there is a
significant tendency for television usage to rise also. As with amount
of reSpondent radio usage, the hypothesis for variety of radio usage
suggested a negative relationship. In fact, the hypotheses fOr all
radio usage predictor variables suggested a negative relationship
between them and child viewing. Yet, three of the fOur radio usage
variables in this study show a positive relationship -- variety of
parental radio use, amount of reSpondent radio use, and variety Of
reSpondent radio use. Amount of parental radio use was non-significantly
related.to the criterion.

An explanation of these counter findings is difficult to draw.
As the review of past research in Chapter II indicated, early
television studies consistently showed that television displaced radio
usage -- i.e. consistently negative relationships were fOund between-
the variables, absence vs presence of television and amount of radio
usage. Variety of radio usage was, itself, seldom tested but the
few available findings also suggested that variety became more
restricted with the introduction of television. These findings, however,
pertain mainly to the early introduction of television and the old-
style dramatic radio. In the intervening years, radio has changed as
a medium to the currently accepted music-news fOrmat. The most

recently reported correlations Of children's radio use and television

__——_. __~_..4 4~_—. 4

80

viewing have been non—significant (Parker 1960 and Troldahl and

Costello 1966). One study (Bailyn 1959) found a small but significant
negative correlation between amount of usage for both variables. In

the context of these most recent findings, the results for the present
study seem incongruous. Analysis of the contingency crossbreaks for

the three radio use variables shows that very high and very low levels

Of radio usage and radio variety seem to generally go with very high and
very low levels Of television usage, reSpectively. This distribution of
reSponses for the three variables may suggest a new kind of "all or none"
principle applying only to electronic media usage.

Variety of respondent book preferences. The X? for this

 

variable is significant at p(.05. The distribution of responses
indicates a generally linear trend with increased variety in book
pre‘erences going with increased television usage. The significant
(p<<.01) r of .18 supports this positive relationship. The hypothesis
predicting a positive relationship between variety of reSpondent book

preferences and frequency of child television viewing is confirmed.

 

Respondent preference for comics. The X2 for this variable is
also significant (p (.02) as is the r of .21 (p(.001). The hypothesis
predicting a positive relationship between preference for comics and
frequency Of Child television viewing is confirmed.

2
Frequency of respondent movie attendance. While the X for

 

this relationship is not significant, the r of .17 is (p<:.01). The

discrepancy between the two statistics is probably accounted for by
the extensive collapsing required for the contingency analysis. While
the crossbreak is not significant, the distribution of reSponses does
show the hypothesized positive trend supported by the r -- i.e. higher
levels of child viewing go with higher frequency of movie attendance.

Nedia respondent would miss most.' Again, this is a variable

 

for which the X2 is not significant but the r of .15 is (p 4,01).

The distribution of reSponses is the crossbreak suggests the source
of the positive correlation. While reSpondents indicating television
as their most missed medium are slightly more likely to be heavier
television viewers, respondents who indicate another medium as their
most missed are slightly less likely to be heavier viewers. Given
the significant r, the hypothesis predicting a positive relationship

between reSpondent reports of television as their most missed medium

nd

H
I”)

f

—requency of television viewing is considered confirmed.

Comparison of linear and curvilinear models. The summary table
2

 

for this category of variables (Table 8) reports the r2 and Eta

or variance accounted for in the criterion by the predictor with a
linear versus curvilinear model. As with the parent media use
category of variables, respondent media use variables account for more
variance in the criterion with the curvilinear model. The discrepancy
in variance accounted for ranges from a low of 6% to a high of 14%.
While the linear r's account for from O to 6% of the variance in the
criterion variable, the curvilinear Etas account for from 8 to 17%

of the variance.

82

Eultiple linear correlation analysis. DeSpite indications that

 

a curvilinear model better fit the present data, the multiple linear
correlation analysis was applied to reSpondent media use variables.

r1“ 5' ‘ '
Tile 1111(111".

s below, then, apply to a linear prediction.
Table 9 reports the results of the multiple R analysis within

this variable category.

 

Table 9. Results of multiple linear correlation analysis for the
reSpondent media use variable category.

 

Value of Multiple Correlation (R) Variables retained after_
' least squares deletion

 

 

With all 12 With variables
variables retained after Variable r to DV
included in least squares
equation deletion
.423 .39a Freougncv R movie .17b
atten an'ce

Amount R radio

listening .12b
Variety R radio

b
use .25
Media missed most .le
Variety R book b
preferences .18

 

aSignificant at p ( .0005.

b
The partial r's for these variables all approximately equal the r's.

The largest discrepancy between an r and a partial r is .03.

 

83

Assuming a linear model, the 12 respondent_media use variables in
combination, account for approximately 18% of the variance in frequency
of child television viewing. At the end of the least squares deletion
process, the resulting Multiple R equalled .39, accounting for 15% of the

v riance in the dependent variable. The variables retained after the

{U

deletion process included: frequency of respondent movie attendance,
amount of reSpondent radio use, variety of reSpondent radio use,

media missed most, and variety of respondent book preferences.

 

84

 

ozm

 

mm a: mm mm Ha awn:

mm mm mm mm on mpmnaeoz

.8.v a «as: man. a 3 . mm am new 33
mm: oflomn

 

pcwocoamon mo mpowbm>

 

 

 

mam -
on ma a: ma ma see a mason m away one:
om am mm on ad saw a mason m on a:
on mm :H . m: we see m neon H on a:
HotV m mm.mm on u s :m mm Hm wnom xmms m mason
: cmcu wwoa .9o>mz
a
pcoocoamop mo pesos<
a we ax : . as a: a
onmsownflco mwcflsow> CmeH>oHep pawco mo homosvosm wannanm> novowoonm

ma

.oagmwam> sewnowwfio one ow modnmwho> om:
oHUoE.uc0pcoameA mo chmcoflymaoh mzw hoe woSHo> m#ad@caw£o pcm moanew kocoucwwcoo .OH oanna

 

:zm

 

 

 

 

 

 

 

an em mm mm ma bgmnm op anon
we mm mm mm ma means
mm mm pm me am 029
.m.a m mo.s om u a mm mm em emu one on ocoz
momma Homocommmfi
mocmumwme mo amossz
Hmm
.3
8
am am om mm on awn:
m: as am am me new; sambwnmeoz
as am am mm mm god sambmnoeoz
.m.: m mm.m me u : am mm mm mam zoo
om: nomwamsm:
pcoocommmn we muownw>
a me «x : :2 a2 a

oemzwmuflco

measofl> cofimw>oaop paflzo mo zocosvonm

‘

essences paponemnm

I

0::H#5001\OH OHde

 

51m

no mm on ma mm copwo
med bu om om . mm moefivmeom
.w.: m ma.m mm u : mm :m 5H wow nm>oz

 

om: mnmnnwa pcoocoomom

 

 

 

 

86

 

 

 

.HMMI
:HH 3m mm Hm 5H moweoo mnemonm
mo.v. m no.0H swan : ma mm mm whw mowsoo
nomona p0: moon
womsoo now
monogamone pceocommom
Hmw
“mm. mm om mm ma awn:
a: mm am am as awn; snobmnoeoz
mm mm mm :H mm 20H sambmnoeoz
mo.v. m m>.ma :m u c :H mm mm wmm . 304
woocobomond xoon
pomocoamon mo xpownm>
a me. «x x :2 oz 4
onmsvunflco msflsofl> CmeH>oHoe oaazo to monograph wannabe> nobowoosm

C 0: C we swam! : OH O -... n1» .H.

87

 

 

 

 

 

 

 

owe mm mm :N am >9
.m.c m m:.m so a ma om om em“ >9 amen amnpo
pmos mmwm
UHDOZ pcoocoameh mace:
Hmm
as mm mm mm ma swam
He mm Hm mm as awn: sambmnmeoz
as am am Hm :a 30H sambmnmeoz
.m.c m :m.m a: a as an an sea son
moocosommnm onooon
pceoCOmen mo meownm>
.omm
em on om an we genes m 00:0 swap who:
as mm mm mm ad space m mono psog<
.m.: a :s.o add u a ma mm mm new eoeamm .nm>mz
mocwocopum mw>OE
vcooeoomon mo >ocosomhm
a ea mx 2 m: a: a

obmsomafico

 

 

 

mzwsom> commw>oaow SAHzo k0 hocozvonm

oaoewnm> pepofloonm

lli-

C $35.. u. :00 I I OH OH. on;

 

88

.mmaamp

ecu mmonom mcwomon .waoa one mm coozpmn kuou has .mnommnmcp .ocm mmo raccoon coon m>mn mmmmpeoonomb

.mm .2 mom .:2 ”no .az mam .4 "ohm madame mama» an

mam>ma wcfizmfi> on“ now we EDEHCw: .mo .2 who .:2 mew .42 mom .4 "ohm weflsmw> mo mam>ma wew>nm> may
now we Amman mamemm ESEmeEv .mmm u a Hobo“ cos: .oopmamo one: mmHQMan> pouowponm ecu co newcommop
use: enamomn wofinm> madam“ may now : Hmu09 .zmmn u : “swag zaopmnmooe u z: msoa mampmnmooa u A:
mmnmzmw> 30H u a .HHH ampmwcu cw ownapowmp mm mmafiunmsc Opr roommaaoo was oanmwnm> :oflnmpflno mn9m

 

 

 

 

 

 

 

 

- 0mm
9m rm ma :w mm >9
.m.: m mm.m mam" : mm mm mm wow >9 away nmcvo
wwomE oo>omaon unwed
”mucwpwa >pflawnflomno
warms pemocommmm
0mm
mzd mm mm mm ma . >9
.m.c m 9:.H med u a mm mm on row >9 swap nonuo
. mHomE om>ofiaon umoE “mmcwpmn
>uflamnwpmno warms pcmocoamwm
a he ax lms, rel a: a
ommncmuflco msfl30w> cownfl>oaob peace we mocozwenh odomflbm> nowowoonm

I‘lli‘l

UUZCH#COOIXCH CHASE

89

Family Cohesiveness

 

Table ll summarizes the results for individual hypotheses on
the relationship of family cohesiveness variables to frequency of child

television viewing. The results indicate that of 10 variables, none

 

Table 11. Summary of results for individual hypotheses on the relation-
ship of family cohesiveness variables to frequency of child
television viewing.

 

Variable Pre- é Variance
diction X p r Eta accounted for
2
P Eta

 

Parental permissiveness:

Knowledge R whereabouts - n.s. .02 .30 .00 .09
Parental permissiveness:

Restriction on hours - n.s. .06 .32 .00 .10
Child-parent communication - n.s. .ll .34 .01 .12
Family togetherness - n.s. .07 .26 .00 .07
Parent orientation - (.10 .11 .29 .01 .08

Parent decision making:

Who pays bills + n.s. .06 .32 .00 .10
Parent decision making:

Who pays allowance + n.s. .06 .26 .00 .07
R home responsibilities + n.s. .09 .32 .Ol .10
R knowledge family Operation + n.s. -.07 .31 .00 -.10
P-R agreement on TV programs + n.s. -.O2 .24 .00 .06

 

are significantly related to frequency of child television viewing.
Table 12 (starting on pagesu.) presents the contingency crossbreaks
and chi-square values for each of the 10 predictor variables in this

category.

90

Comnarison of linear and curvilinear models. The summary table

2 2

 

or

or this category of variables (Table ll) reports the r and Eta

HI

variance accounted for in the criterion by each predictor, using a
'near versus curvilinear model. As with the two preceding categories
of variables, again we find that a curvilinear model seems to have a
better fit with the data. The discrepancy in variance accounted for
ranges from a low of 6% to a high of 11%. While the linear r's
account for from 0 to 1% of the variance in the criterion, the
curvilinear Etas account for from 6 to 12%.

Hultiple linear correlation analysis. As would be expected

 

fsom the non—significant results of the individual hypothesis tests for

J
(J

Iables in this category, the multiple R within the category is not
significant (R = .22). Assuming a linear model, the 10 family
cohesiveness variables either alone in in combination show no pre-

dictive power for the criterion variable.

 

 

 

 

 

 

 

 

as am mm mm . an seem
as om an Ha Ha emae sampmnmeoz
an an mm an ad god sambmgmeoz
.m.a m mm.ma 90 u a ma 9m Hm ram 304
sewpmowcseeoo pampmmlpawco
mam
mo 9m ma 9m om mmoco>fimwwenmm swam
am on om as ea ewes sambmnmeoz
om om mm om ma- 20H snobmnmeoz
.m.: m mo.oa m: n a :H mm mm wmm mmmcm>wmmwenmm 30a
meson co coauoflnumon
mu Homeow>flmwwenmmtamucmnmm
Hmm
mm mm mm ma :m wmocm>wmmwssom mem
.m. m as.a mod n a mm am am has“ mmmco>nmmweaoa so;
musonmmnmcs m emooasonx
"mwmcm>flmmwemma Hopcmnmm
a we «x x :2 a: a

oncsvauflcu

 

mwcw30w> :onfl>mfioy pawzo mo mononwonu

maowwnm> nobowoonm

 

eanuHLm> COwLopflbo 02w ow medotwnm>

neoco>flmozoo xdflSom mo awznc0wpodoh 0:9 new cases» eaczouawzo new comedy xoceuzwpcoo .mH oan9

92

arm

 

 

 

 

 

 

 

am am nu on pH mbcmema zoom
.m.: m no.“ mean a mm am am an“ panama mco
waawo m>mo 0:3
mewxmfi sewwwomo pcosmm
mam
am am mm am am mecmnma seem
8a mm mm S cm £83 25
ea.v m mm.aa a: u a m an om mom panama nmebnmz
0 w 0 a
COMHwMuFHmWHO chfimm
mmm
mm on mm on ma sea:
on ma mm mm Hm awn: sampmhmeoz
m: mm mm Hm ma assume
:3 mm mm mm on sea >Hmpmnmooz
.m.: ma mm.@ mm u a an mm Hm wan 30A
mmmcnmnwomov >Hwemm
a he ax : -2: a: a
ontzwmxfizo mcwzmw> COwwfl>oaow ramco wo xcsosvohm ofinmwam> Loyowwmhm

 

I‘ll

Unscwncoozlmd Odao9

93

m:m

 

 

 

 

 

 

 

mm am an Ad ad new:

a: mm an em 5H new: saobmnoooz

mm mm am am ma esnooz

mm on mm as :m 30H sHobmnoeoz

.m.a «a mm.~a o: u a mm mm mm mom monpnannnmcoomon zoo
mowuwawnflmcoamop
oEoc m.pcoocommmm

cam

.mmm om mm mm ma wbcoooa abom

Add mm on am am econmo oco

.m.c o no.0 :m u c mm Hm 9H wow bamboo socpwmz
mocmsoaam m>oa 0:3
. mcwxms cofimflooo reopen

a me x : .m: a: a

Gnoscwxw;o

(\J

m:930w> coflmw>oace peace wo >0Coswonu

canowbo> nobofloona

 

C 03 Z ......w ECO! INH wuflnflwa.

.mmaomp ocp mmObom mcflommn .waoa one mm cmospmo Hmpou mos .mnomonocv .ocm mmo raccoon coon o>w: mowopcoULOAA
A .sm .x mam .mz mmm .az mam .4
"ohm moaomu omega aw mao>oa mcflsmw> one new we ESEHsz .mo .: m9m .mz mam .42 mom .q "one mcfisow>
mo mam>md mafi>nm> ecu pom m.c .Aonww oaasmm EDEmeEV «mm H : Hmpov cmc:.omuoaop one; moabmfl9m> bouowooae
co mowcommopuaoc omsmoon moflnw> moanmw may now : Hmp09 .cwmc u : ”cw“: >Hovmnoooe n ma ”30H >Howmnoooe
u a: meoSmM> 30H u A .HHH nopnmzo cw omoflnommo wmmvafiunwsd oucw powdmaaoo mm: wanmwpm> cownmuwso 0:98

 

 

 

 

 

 

 

arm
>3 mm am on ma new:
o9 om mm an :m opmnooo:
.m.: o m9.m mma u : mm an em WAN 30A
moofiono Emnwona comwm>oaow
co vcmEmonmw pHHLonpconcm
u.
9
mmm
em om mm an am can:
a: am am om Hm some sambmnoooz
as am am am an 20a saopmnoooz
.m.c m mm.: m9 u a an :m mm wow so;
cofipmnmeo >Hflsem
mo ompmazocx ucoococcox
a me x : :2 a: a

‘l
“-CHRZCDI

H

"-1
L)

I.
mewzofl> coflmw>olob cawzo mo >ocoswonh oaomflnm> powofioong

'llll'li

ooSprcooncma oenfi9

CO 553-11133 ITY IN TEGRAT ION

 

95

Table 13 summarizes the results for individual hypotheses on

the relationship of community integration variables to frequency of

czhild television viewing.

The results indicate that of seven variables,

 

Table 13. Summary of results for individual hypotheses on the relation-
ship of community integration variables to frequency of
child television viewing.

 

 

Variance
Variable 2 accounted for
diction X p r Eta r2 Eta2

gregariousness n.s. -.Ol .31 .00 .10
R gregariousness:

organizations 4 .02b .11 . 30 .01 .09
R gre ga.r iousness: peers n.s. .04 .30 .00 .09
Jute id e home orientation n.s. .03 .30 .00 .09
Len t1 of time in community n.s. -.07 .23 .00 .05
Fiequency of attendance at

church (.10b .123 .34 .01 .12
R ..c lac e local 8 state

figures (.10 -.04 .22 .00 .05

 

aSignificant at p<f.05, by r to z transformation test of significance.

b".. .. . . . . . . .
Walle results for this variable are Significant or near Significant,
they are in a direction opposite to that predicted.

 

96

h

l

live do not reach the criterion level of significance of p( .05 in
either the X2 or correlational analysis. These five variables are:
parent gregariousness, reSpondent gregariousness (peers), outside

home orientation, length of time in community, and reSpondent knowledge
of local and state public figures. The latter variable nears
significance in the X2 analysis but the distribution of reSponses is

so unclear that no trend conclusions may be drawn. Table 1“ (starting
on page 99) presents the contingency crossbreaks and chi-square values
for each of the 7 predictor variables in this category. Details on
the two variables which are significant follow.

. . . 2
Respondent gregariousness: organizations. The X for this

 

‘Jariablesis significant at p<<.02 and the r.of .11 is near significant
at p<:.lO. The correlation suggests the direction of the relationship
which is generally positive with higher levels of gregariousness
associated with high levels of television usage. However, analysis of
the contingency crossbreak indicates that the relationship is not as
clear cut as a linear prediction would indicate. The direction of
the results is contrary to the hypothesis which suggested that
reSpondent gregariousness would be negatively related to frequency of
child television viewing.

As the review of past literature in Chapter II indicated,
evidence on the relationship of gregariousness to television viewing is
contradictory. Early work suggested that the introduction of television

reduced the informal gregariousness of reSpondents but not the formal

 

 

97

,—

tregariousness or participation in organizations. More recent work

0

11‘
(I)
0)

found only non-significant relationships between gregariousness

d TV 'iewing. This history makes the present finding even more in-
congruous. Unfortunately, an analysis of the distribution of responses
in the contingency cross-break (Table l“) is not very helpful. Most
of the cell contribution to the significant X2 comes from only four
cells in the 12 cell table, Moderate gregarious reSpondents are more
likely to be very light viewers while high gregarious respondents are
more likely to be moderately light viewers. Otherwise, both moderate
and high gregarious respondents seem equally likely to be moderately
high and high viewers. And, low gregarious reSpondents are equally
likely to fall in any of the four viewing levels.

1

2 . . .
Erecuency of attendance at church. The X for this variable 13

r

 

near-significant at p(.lO and the r of .12 is significant at p (.05.
Again we have a result that runs counter to the hypothesis. While

the prediction was that increased attendance at church should go

with decreased viewing, the results show the opposite. Greater
frequency of church attendance generally goes with greater frequency

of television viewing. An explanation of this counter-finding is

also difficult to draw. The early television research produced
negative and non-significant relationships between absence-presence

of television and church attendance. The most recent work has resulted
in entirely non-significant findings between church going and amount

of viewing. Of course, the significant r is so low (a r of .12 accounts

98

for only 1% of tn variance in the criterion) that an elaborate attempt

(7)

to rationalize this counter-finding lacks merit.

Comparison of linear and curvilinear models. The summary table

- .. . . 2 2
:or this category of variables (Table 13) reports the r and Eta

 

or variance accounted for in the criterion by the predictor, using a
linear versus curvilinear model. As with preceding categories of
variables, we again find that a curvilinear model better fits the data.
The discrepancy in variance accounted for ranges from a low of 5% to

a high of 11%. While the linear r's account for from 0 to 1% of the

variance in the criterion, the curvilinear Etas account for from S to

r.)
IJ'
O

‘

I,_I

Multiple linear correlation analysis. With all 7 community

 

integration variables included in the Multiple R equation, the resulting
R equals .18 and is not-significant. Thus, assuming a linear model,
the variables in this category do not show significant predictive power

for the criterion variable.

99

 

 

 

 

 

 

 

 

 

0: mm mm mm 9H LmH:
:9 mm mm :w :m :uH: maopmnmco:
99 :m mm ma mm 30H >aoumnmpoz
.m.: m NH.HH mm u a ma ma mm wxm 304
whoom
"mmocmsofibmmobm peoUCOQmom
new
mm mm mm on oa swam
mm mm mm ma mm opmaopoz
mo.V o m9.0~ mm H a nu mm Hm won 304
m:0wpmxflcmmso
“mmchSOHsmuonm poocCOQmom
- mew
9m 9N Hm om mm gmw:
:9 om on ma mm mumbooox
.w.: o Ho.m «9 n a ma mm mm Awma 30a
meccmsowbomenm Happen
o to «x z . :2 a: a

monasomuwco

emcwsow> conw>ony camxo mo >oco5donh

I

vabmwnm> nepowwmnm

magmwam> COMLOPMbo 0:9 09 noanmahm>
:cHquwOPCw kHHCSESOU 9o awszOHHtaon 3:; boy me:am> Qacnwwnw;o Use woascu 9023::awcov .:H cand9

 

mmm

 

 

 

 

 

 

 

 

 

 

mad mm am am am new:
a: mm on am am oregano:
Aoa.vv o Hm.aa me u a «a am am emu so;
0 m. a
SUQDLU Pm. QOCmGEQPPm
mo >ocosvmnh
mam
mm am am mm am whoa no NH
mo am am am mm memos Ha . a
mm mm mm am am _ mnmos o u a
o .m.c o :o.m om u a mm mm om row memos m - o
Mm >wfiCdEEoo
ow oaww mo cpmcmq
mam
am am am am am swam
as am am mm Hm awn; maoumnoeoz
as am am om an god adobmnoooz
.a.: m Hm.m om u a mu ea mm ram eon
conmucoflpo
. mac: crampso
a no ax ;m| . :2 a: a
obmscnxflco mewsofl> coaww>eaow peace we mocobvoam odomwnm> nobowoonm

 

 

 

U mozcfiwnouuisd 0.93:.

.moanmp esp mmonom wcfiommn waoa one am coozpmb HmPOp mos .obowmnocp .ocm mmo coocnon coon e>mc mommpcoonomn

.oo .: Mme .22 new .az mam .a "one madame omens an mam>oa

wowsow> esp new we ESEwcH: .mo .2 m9m .mz mew .42 mom .4 ”one wsflsmw> mo mao>oa mcwmnm> may new
we .AmNflw ransom ESEmeEV «mm u a Hero“ crazy .oopmamp one: woaomfibm> bosowoona may no womcoamop
Icon omsmomn weanm> mebMp may new we Hep09 .£m«; u I mama: >Hopmnooos n :2 mSOH maopmbocoe u a:
mmnmzow> 30H u a .HHH nopamco ow coownomoo mm meaflunmnw Opcw oomamHHoo mos oabwwnm> sewsopwno mc9m

 

 

 

 

omm

.i mmm. an ma mm mm awn:

m“ Hm Hm mm am ma cane snobmnoeoz

m: mm ma ea mm assume

a: an mm mm an god snobwnoeo:

leauvv NH om.om mm u a mm mm ma and . zoo
m c memmwmw
owansa women one Hmooa
mo earmasocx pcmocommmm

o me x z :2 a: a

ohmsoonwzo wowsow> cowhepwaop pawco mo mucosvobm mabmwbw> nOHUfioonm

toncflpcoonzsd GHAM9

102

“-1-: h,“'f“"\'r‘\fil"lT.e-\\
“ v- ‘

c ,_ _I
0.414- UL\~AA.\‘AAJL..\/e\

 

Table 15 summarizes the results for individual hypotheses on
the relation of self orientation variables to frequency of child

television viewing. The results indicate that of eight variables,

 

Table 15. Summary of results for individual hypotheses on the relation-
ship of self orientation variables to frequency of child
television viewing.

 

 

Variance

Variable Pre- 2 accounted for

diction X p r Eta r2 Eta2
hunber R hobbies - n.s. -.05 .36 .00 .13
Number P hobbies - n.s. -.10 .25 .Ol .06
R.knowledge: ad slogans + n.s. -.08 .30 .Ol .09
Eiknowledge: TV characters + n.s. .10 .30 .01 .09
Eiknowledge: math - n.s. -.12a .23 .01 .05
Zloutside home employment - (.01b .19a .37 .04 .14
Either hours R studies - n.s. .01 .27 .00 .07
Frequency R studies in
library - n.s. -.02 .30 .00 .09

 

hfignificant at least at p<.05, by r to z transformation test of
significance. Critical values for r's are: p (.05 = .12; p<.Ol = .15;.
p<.001 = .19.

DWhile results of this variable are significant, they are in a
direction opposite to that predicted.

 

six do not reach the criterion level of significance of p.(.05 in

.,_ , 2 . .
Either tne X or r analyses. These SlX variables are: number of

respondent hobbies, number of parent hobbies, respondent knowledge of

103

ad slogans, reSpondent knowledge of TV Characters, number of hours
respondent studies, and frequency respondent studies in library.
Table 17 (starting on page 107) presents the contingency crossbreaks
and chi—square values for each of the 8 predictor variables in this
category. Details on the two variables which reach significance

follow.

Respondent knowledge levels: math. While the X2 for this

 

variable is not significant, the r of -.12 is at p (.05. If the con—
tingency table for the variable is collapsed to a 4 x 2 (four levels

of math knowledge x two levels -- light and heavy -- of viewing),

the resulting X2 is significant at p< .05 (X2 = 9.34, df = 3). The
distribution of reSponses in the crossbreak as well as the negative
correlation concur that generally lower math knowledge levels go with
higher television usage. The hypothesis predicting a negative relation-
Lip between these two variables is confirmed.

Respondent's outside home emplpyment. Both the X2 and the r
of .19 for this variable are significant -- the X2 at P<-Ol: the r
at p<§001. The direction of the relationship indicates that more
frequent employment generally goes with more frequent television
viewing. This finding runs counter to the hypothesis which suggested
a negative relationship.

Since none of the past television work cited here has looked
Specifically at this variable it is difficult to place this counter-

finding within a context. Child development literature (as noted in

Chapter II) has suggested that the child who works outside his home

104

prediction was made. It is possible to conjecture, however, that
outside home employment for a child goes with such attributes only up
to a point. At some high level of outside work, the child may be too
lausy to make peer contacts and is burdened with reSponsibilities so
that television then becomes a solace. The distribution of reSponses
in the crossbreak for this variable (Table 17) supports this kind of
curvilinear relationship. Respondents who don't work at all seem to
be higher vie*ers than respondents who work "when work is available"

or occasionally. The reSpondent who has a steady job, however, is

b

“he most likely to be a heavy viewer.

Comparison of linear and curvilinear models. The summary table
for this category of variables (Table 15) reports both the r2 and
Eta2 or variance accounted for in the criterion by the predictor,
using a linear versus curvilinear model. Again we find that a
urvilinear model fits the data better. The discrepancy in variance
accounted for ranges from a low of 4% to a high of 12%. While the
linear r's account for from O to 4% of the variance in the criterion,
the curvilinear Etas account for from 5 to 13%.

Multiple linear correlation analysis. Table 16 on-the next page
reports the results of the multiple R analysis within this variable

category.

 

 

 

Tasle l5. Results of rultiple linear correlation analysis for the
self orientation variable category
Value of Kultiple Correlation (R) Variables retained after least
squares deletion
With all 8 With variables
variables retained after
included in least squares
equation deletion Variable r to DV
29a a ' t '
. v .23 Child 8 outSide
home employment .19
. Knowledge level: b

aThe R of .28 is significant at p =(.009; the R of .23 is significant
at p =<ZOOl.

The partial r's for these variables are: .20 for child's outside
home employment and -.13 for knowledge level: math.

 

Assuming a linear model, the eight self orientation variables
in combination account for approximately 8% of the variance in frequency
of child television viewing. Variables were then deleted from the
multiple R equation by the least squares deletion criterion. At the
end of the deletion process, the resulting Multiple R equalled .23,

accounting for 5% of the variance in the dependent variable.

'red after the deletion process included:

-‘oyment and knowledge level: math. These

clilu's outSice here empl
CAGE are the ”nest” self-orientation predictor variables of frequency

of child television viewing, assuming a linear model.

 

 

 

 

 

 

 

m: :H mm mm mm xwm ow 03%
an mm Hm mm ma 0:0
.m.a o m:.o mod a mm mm 3N wmm ocoz
moflnbon
pcwnma mo ambesz
bzm
mm 5H mm mm mm xwm ow coach
7 mm om mm mm am 039
O
.1.—
mm mm mm aw mm mco
.m. c m m: . a :m a mm mm on £me meoz
moflbbo;
weepcommmfi
mo noQES~
a me mx 2 :2 .E A
Osmscmz.zo mwcfismfl> :ofimfi>eaop passe mo zocosvoah manmwnm> nowowpopm

I

 

:owumbcowpc wave m0

.Qagmw9m> coflgmwwhu onv ow moancfigw>
afizmcowpcacp 0:? new ecumw> vgmzcwlwzo was mudflmp zozom:M#:Oo .bd omaah

 

1mm

 

 

 

 

 

 

 

 

 

 

5 mm 3 so an :3:
mm mm mm mm om zmwg mamumnmpox
on mm mm mm mm 30H hampmhwpoz
6.: m 3.: a: u c 0: mm ma wad 33
gyms ”mam>md
mmpoazocx envenommwm
0mm
a: mm mm 8 S :3:
.m.c m mo.: and" : mm mm mm wen 304
W mampomamco >e "mam>0H
a. omnmazocx unoccommom
0mm
mm mm om mm mm :3:
oo mm om um Hm :wwn hampmnmvoz
Hm mm ma on an 30H mamuwnowoz
.w.: m Ho.m on H mm mm :m wsa 304
mammoam p< umHm>mH
emphasocx pampcoammm
- a we wx : E ..E a
mcflzow> coflufl>oawp UHH;Q we >;:;:co&h wabmwnm> nouowomam

opmscmumzu

 

 

 

 

poscwwcoouuna oases

 

:mm

mm mm mm mm mm see m meson : . m\a N
mm am mm am as see m meson m . N\H a
o mm.o as u e an am am sad see m 9:0: a u o

 

weepsum pampcoawwp
mnsoz mo embasz

 

 

109

 

 

 

0mm

H: :: mm ma NH AOm zemwum mmm

mod mm ma mm mm magmawm>m

ma x903 cog: mxno:

Ho..V Aw mo.ha moa u a ma mm :m wmm x903 #0: when
pcmE>oagEo @Eoc
opwmwso m.pcmpcoawmm

me x : :2 42 a

MCH395> ceaaa>meow paazo %o mocozdoam canmAhw> nowadfimhm
ohmsomxflzo . . . . . .

It

casemwzooxxsa magma

llO

.meAMp one weepom mempmmn waoa new mm cemsumb Hmpou >mE amnommnmzp .Uem mmo pmpcson amen w>mn mommpcwonmmn

.sm .: law a: mom .az mm: .q "mam mmfinmu amaze an .mam>mfi

mewzww> may now we ESEHCHE .mm “2 who .32 mxm .42 mom .4 "mam mcwzmfl> wo wam>ma mammnw> mnw

pm we .Amnww mamemm ESwamEV mmm n : ampov cons .pwpoamp mum: moanmflnm> QOPOflpmnm map so newcommma
1:0: mmsmown wwwpm> mmabmp exp 90% : Hmpoe .zmfl: u I mnww: mamumnmpos u :2 mzoa hamuwhmpoe u a:
mmnozmw> 30H u A .HHH ampmmzo ca omnwnomwp mm mmawwnmsv ova“ pummmaaoo mm: manmwnm> cownmvwno mesa

 

 

 

 

0mm

«2 mm mm :H :m xmmz m coco cozy @902

mm :m mm Hm 5H xooz m moco

mm :m mm mm mm xem: m mono cmzp wmoq

.m.: m mm.m me u a mm mm mm mam . no>mz
>am9bfia a“ mewpzpw
ecopcomwmh >ocmswonm

a me «x z . :2 a: a

mhuzvmchv mcflzefl> :oflmfl>waop pafizo mo xososvonu mabmfipm> mouowwwnm

poszwwcoozasd vanes

lll

. \?‘flv-‘fh*x p‘v‘Q-H-um. . ‘T
. _ .g: .' a. n n H!"
be. -‘uC'V.;lJ-\ L-\.L._a.\lfl ..Oh

 

Table 18 summarizes the results for individual hypotheses on the
relation of consumer orientation variables to frequency of child

television viewing. The results indicate that of eight variables, four

 

 

 

Table 18. Summary of results for individual hypotheses on the relation-
ship of consumer orientation variables to frequency of ch 1d
teleVision viewing.

Variance

Variable Pre- accounted for

diction X2 p r Eta r2 Etaz
L

hutner of P money worries + (.05 .lua .33 .02 .11

Hurler oi R money worries + n.s. .ll .32 .Ol .10

P Spend-save orientation - n.s.-.13a .33 .02 .11

I spend-save orientation - n.s.-.04 .27 .00 .07

Fanily use of credit + n.s. .08 .26 .Ol .07

R attitude toward credit + n.s. .15a .27 .02 .07

Variety of R Spending + .<.05 .16a .32 .03 .10

Family shopping orientation — n.s.-.09 .25 .Ol .06

 

aSignificant at least at p<.05, by r to z transbrmation test of significance.
Critical values for r's are: p(.05 = .12, p4.Ol = .15, p<.OOl = .19.

 

do not reach the criterion level of significance of p(.05 in either the
X2 or r analyses. These four variables are: number of reSpondent money
worries, reSpondent Spend—save orientation, family use of credit, and

family shopping orientation. Table 20 (starting on page 115) presents

112

4.1‘ 3

the contingency crossbreaks and chi-square values for each of the

8 redictor variables in this category. Details on the four variables

' {“1

which reach significance follow.

. . 2 . .
Number of parent money worries. Both the X for this variable

 

and the r of .14 are significant at p(.05. The direction of the re-
lationship is positive with more parental money worries being associated
with more frequent child television viewing. The hypothesis predicting
a positive relationship is confirmed.

Parent spend-save orientation. While the X2 for this variable is

 

not significant, the r of -.13 is significant at p<.05. The dis-

crepancy between the two statistics is best accounted for by the extensive
collapsihg necessary fer the contingency analysis. DeSpite the
discrepancy, the contingency crossbreak does show a distribution of
reSponses indicative of a negative relationship. There is a tendency

for reSpondents who indicate their parents would "Spend" or "Spend and
save” a windfall of money to be heavier television viewers. On the

basis of the significant r, the hypothesis predicting a negative
correlation between parental spend-save orientation and frequency of

child television viewing is considered confirmed.

Attitude toward credit. 'Again, the X2 for this variable is not

 

significant but the r of .15 is at p(.Ol. Despite the lack of

2

significance for the X , the contingency table of responses does show

a trend toward a positive relationship -- a more favorable attitude toward

113

credit goes with heavier use of television by reSpondents. Given the
significant correlation, the hypothesis predicting a positive relation-
ship between attitude toward credit and frequency of child television
viewing is considered confirmed.

Variety of reSpondent Spending. The X2 for this variable is

 

significant at p<.05 and the r of .16 is significant at p(.Ol. The
direction of the relationship is positive, as predicted, with greater
variety of respondent Spending being associated with more frequent
television viewing. The hypothesis for this variable is, therefore,
confirmed.

'

Comparison of linear and curvilinear models. The summary table

 

for this tategory of variables (Table 18) reports both the r2 and
31a‘4 or 'ariance accounted for in the criterion by the predictor,
using a linear versus curvilinear model. Again we find a curvilinear
model fits better. The discrepancy in variance accounted for by the
linear versus curvilinear models ranges from a low of 5% to a high
of 9%. While the linear r's accounted for from 0 to 3% of the
variance in the criterion, the curvilinear Etas account for from 6 to
11%.

Multiple linear correlation analysis. Table 19 reports the
results of the multiple R analysis within this variable category.

Assuming a linear model, the 8 consumer orientation variables in

combination account for about 8% of the variance in the criterion

114

Table 19. Results of multiple linear correlation analysis for the
consumer orientation variable category.

 

Value of Eultiple Correlation (R) Variables retained after least
' squares deletion

 

 

with all 8 With variables

varinbles retained after

includes in least squares Variable r to DV

equation deletion

.28a .21a Variety of b
respondent spending .16
Attitude toward credit .le

I
The p. of .23 is Significant at p =(.009; the R of .21 is significant
at p =(.003.

b
...

:ne -art'al rs for he two vari

'(J

1 ables are .16 (variety of respondent
Spending) and .l# (attitude on credit).

 

variable. Variables were then deleted from the multiple R equation by
the least squares deletion criterion. At the end of the deletion
gnecess, the resulting multiple R accounted for H% of the variance in
'ne dependent variable.

The variables retained in the final R equation were variety of

nxmondent Spending and reSpondent attitude on credit.

115

mrm

 

 

 

 

 

 

 

 

 

ms mm mm mm mm m>sm
mos am mm mm ma ssmm was ssmam
.m.: o om.m mm u a mm mm am sad esmsm
coflpmpoommo
o>mmaocoem Hambwm
mam
as am mm mm ma Hmsm>om
mod ma em Hm mm mso
.m.: e ms.o as u : om mm ma new . msoz
. moannoz zmcoe
Homecoamob mo bones:
mam
was as am mu ma shoe so see
mo.v m eo.m me. u a ma mm om new“ ssoz
moflanos wocoe
pcmnmm Mo ambasz
s we mx 2 I: a: a

onmswmnwzo

mmCHSow> nommw>waow mawzo wo monoscmph

 

cowwmpcoflbo noenmnoo

 

manmfinm> Loyofiombm

 

 

.oaboflbm> cownopwno one On moabeabm>

m0 Lazmcoflpmaob egg Low ueDHm> obnzvnlwfio new moans“ %o:om:wpnoo .ow oases

 

mrm

am am mm mm ma saassosmn
smH am am am sm mesossa
m mm.s an n s an am am sea . seamsosmme:

 

pwpmbo onmzou opswwpu<

 

116

 

 

 

 

 

 

::m
son mm mm mm om mmebmeom .saamsm:
Hm mm mm mm :H no>oc meEa<
o mo.m mm u c an :N 5w wmm no>oz
“moose %0 mm: hamamu
Nisw
mmH mm mm mm mm . o>mm
m: mm mm ma NH e>mm pom ocomm
s os.m os u s as ma mm as“ esssm
ooflpmpcmwno o>mw
. :pcwow pcopcommom
es ax ii inn as - a
mewzow> newefi>0mop cameo mo xccosombu manownm> QOpUMpobm

onmscmlwcq

—

Concepcoousom oabtfi

.mmanmu may mmonom mnncmmn «waoa new mm embanmb annoy zma nonomonesu .pcm wwo popcson comb o>mn momensoonomn

.mw .m Mme .:2 was .42 mam .4 "was

moanmp women an mam>oa wnwzmw> any new m: ESEHcHz .mo .: who .:2 m:m .4: mom «4 “mnm mewzow> mo mao>ma
mcwmnm> esp now me .Amwnm oaaemm Esenxmev mmw u a Hmvov nos: .pwnmamp one: moabmfinm> nononomnm esp co
momcoammnuco: mmsmomn monnm> moanmp may now : amnoe .cmnz u : mama: maopmnoooe m2 msoa maonwnopos u a:
mmnmzmw> 30H u a .HHH nmvmwzo an pmbnnomop mm moaflnnmzw oven pmmamaaoo mm: manwwnw> cownwnflno scam

 

 

 

 

 

 

33m
hm ma mm mm mm ooflz
mm mm - on an mm sens sampmnoeoz
7 so . mm mm as am senses snapssoeo:
.1
ii .m.s a as.m as u c am mm :m sea smash sonssz
cowpmvcowno wammonm mwnsmm
he
ms nu am as ma sen:
mo.v m mo.m . ms u n :m mm :m wow . 204
mcwocomw
. “concoamon mo >nownm>
s we ax 11 gel as a
mewson> cowmw>caon mango Lo xocosvonm sabofinm> nOHUHponm

onwnonlwco

pozcwnc001|cm canoe

118

D1:“'-’\~"" ‘ATZ‘LTTC
“J.i‘l.\,‘t\n’|L $3.}.

 

The final category of predictor variables in this study include
the demographic characteristics of the child and his family. Table 21
s‘mmarizes the results of individual hypotheses on the relation of

chamcgraphic variables to frequency of child television viewing.

 

Table 21. Summary of results for individual hypotheses on the relation-
ship of demographic variables to frequency of child television

 

 

viewing.

Variable Pre- Variance
diction X2 p r Eta Accounted for
P2 Eta2
3:: status: occupation - (.01 -.21a . 34 .ou .12
83 st: ts: possessions - (.05 -.O2 .28 .OO .08
Family size + n.s. -.09 .36 .01 .13
Fother's employment + n.s. .03 .35 .00 .12
Family type - n.s. -.06 .34 .00 .12
Birth order - n.s. -.07 .26 .OO .07
Sex none n.s. .14a .27 .02 .07

 

aSignificant at least at p{.05, by r to z transfermation test of .
significance. Critical values of r's are: p(.05 = .12, p(,01 = .15;
p{.001 = .19

 

The results indicate that of seven variables, four do not reach
the criterion level of significance of p(.OS in either the X2 or
correlation analyses. These four variables are: family size, mother's
employment, family type, and birth order. Table 23 (starting on page 123)

presents the contingency crossbreaks and chi-square values for each

119

of the seven predictor variables in this category. Details on the
three variables which reach significance follow.

. . . . 2
SOCio-economic status: occupational prestige. Both the X

 

for this variable and the 4 of -.21 are significant at p<.Ol. As
predicted, the direction of the relationship is negative with lower
levels of status being associated with more frequent child viewing.
The hypothesis predicting a negative relationship is confirmed.

O I O O 2 .
Boeio-economic status; posseSSions index. The X for this

 

variable is significant at p(.05 but the r of -.02 is insignificant.
Analyses of the contingency crossbreak for the variable suggests the
reason for'the discrepancy between the two statistics. The dis-
tribution of reSponses shows marked curvilinearity. Low levels of
status seem to go with higher levels of viewing; moderate levels of
status go with lower viewing; and high levels of status go with higher
viewing. The hypothesis predicting a negative relationship between
this measure of status and frequency of child television viewing is not
confirmed. The review of literature in Chapter II provides some basis
for explaining the discrepancy in relationship to the criterion
variable between this possession index of socio—economic status and
occupational prestige which shows a significant negative relationship.
A possession index of status lOgically seems more like an income or
ecological measure which past work has shown does not tap the same kind
of status as occupational prestige. While the discrepancy in results

between the two socio-economic status measures can be partially

120

explained, the reason for the curvilinear relationship of the possession
index to frequency of child viewing is not clear at all. One might
conjecture that reSpondents at the high possession level are more
likely to have their own television sets and therefOre have more
opportunity to view TV. Or, in line with the rationale offered for
the consumer orientation category of variables, one might conjecture
that a possession index of socio-economic status is a measure which
confounds both a family's purely economic buying ability and a family's
tendency toward satisfaction of immediate reward needs. This rationale
would eXplain the higher viewing at higher possession levels.

63x: While the X2 is not significant, the distribution of responses
in the contingency crossbreak supports the positive relationship
suggested by the r of .14 (p<LOS). Boys are more likely to be heavy
television viewers than girls. This is an unanticipated finding.

No hypothesis was formulated for this variable because prior work has
<:onsistently shown non—significant relationships between sex and child
viewing (see Chapter II for literature review). The only prior study
showing any sex difference was Bailyn (1959). She found boys
significantly more exposed to the pictorial media (movies, comics,
television) than girls but attributed the difference mainly to greater
exposure to comics.

Comparison of linear and curvilinear models. The summary table

 

‘ 2
for this category is variables (Table 21) reports both the r and

2 o o o o q 0
Sta or variance accounted fer in the criterion by tne predictors

121

using a linear versus a curvilinear model. Again we find the
curvilinear model provides a better fit. The discrepancy in variance

antes from a low of % to a high of 12%. While the

m
( J
O
Q
h
rt
0)
0
F1,
0
*3
"5

(H

u CI
“r ‘o

rt

O;
H)
O
'5
H1
8
B
0
d
0

e of the variance in the criterion,
the curvilinear Etas accounted for from 7 to 12% of the variance.

Hultiple linear correlation analysis. Table 22 reports the

 

results of the multiple R analysis within this variable category.

 

Table 22. Results of multiple linear correlation analysis fer the
demographic variablecntegory.

 

Value of Multiple Correlation (R) Variables retained after least
squares deletion

 

With all 7 With variables
variables retained after Variable r to DV
included in least squares
equation deletion
.25a .21a Socio-economic status: -.21

occupational prestige

 

aThe R of 25 is significant at p =(.O25; the R of .21 is significant

 

Assuming a linear model, the seven demographic variables in
combination account for about 6% of the variance in the criterion
variable. Variables were then deleted from the R equation by the

least squares deletion criterion. At the end of the deletion process,

9

the resulting multiple R accounted for no of the variance in the
criterion variable. The one variable remaining at the end of the
deletion process was the occupational prestige index of socio-economic

status.

123

 

 

 

 

 

 

as as am an an sen:
om om mm on mm swan maowmnmpoz
mm mm rm mm ma 30H happenewoz
mo.v m 8.: a: u c mm 2 mm a: so;
xeecw eonmommom
“msnmnw oneocoooIOnoom
manmwnm> nonowpmnm
.mmm
mm ma mm ma mm swam
on ma om N: ha openepoz
-S.v s 3.8 mm uc mm en am at: 33
omnnmonm HmCOflnmdbooo
”msnmum unaccoomnOHoom
s we ax : ii: a: a

mmcwsom> newwn>smon mango no >onosvonm

 

nil

mabmflnm> nononponm

 

GHCCflLHQr :OHPHmva/HU 0:“. Cu.

sombmflnm> UnzamnuOEot we enzxcofinmaon 0:9 non mozmo> onesonlflno t2o

“canon kocmunwnceo .mm cane

 

mom mm as mm mm snnamn ncmnms can

mo.a as u c mm mm mm sen snnesn bashes mco

 

waxy hawEmh

 

124

 

 

 

 

.mmm as an mm as manpunnsn misc:

mm mm am am an . sann-nnms expo:

sm.s can" c as am an as“ ago: no: mass
msvmnm pcoemoamem m.no;noz

.mmm

Hm mm me an em meownos an . s

as om mm mm mm meomnms m

mn.m so u c mm mm mm awn . mcomnmc : u m
mmnm zawemm

ax : lei . a: a

mcnzon> conen>o~in mango mo zocosdonm mabmwnm> novowwonm

pmzcwnnoollmm vdbmh

.mmabmn may mmonom monommn “waoa one am coosnob HmnOp zme wonomononp .pcm mmo popcoon comb o>m£ momencoonmmn

.mm .: Moo .:2 mmo .az mam .a "one monoou women

cw mao>od wcnzmn> may now we Ezeficwz .mm .m who .:2 mzm «a: mmm .4 "one mcwzow> mo mam>oa wnnmnm> one
now we .Aownw wamemm ESEnmeV mmm u c annoy con: .oonoawo one: woaemnnm> nouonoonm can no momcoamon
Icon omsmoon wowno> weapon may now a Hmnoe .nmn: u z mnmfln mamnmnopoe u :2 mzoa xaonmnmpoe u a:
Munozow> 30H u A .HHH nonemzo on wobnnomop mm moannnmSW oncw pomnmaaoo mos oabmnnm> connopfino echo

 

 

 

 

 

mom
«on no as so on msom
.m.: m os.m and u c on om mm . sow manna
No. II
1i xom
an“
as am am cm as shop been
as om om an em anon onoenz
.m.: o ma.s as u c mm mm mm no“ egos pmnnn
nmeno npnnm
a me x : z: a: a

onzsomxwgo msnzuw> :Ownw>cflon Udfizo no >occsomnh sabeflno> nouowemnm

 

Ilv Iii ll“

Tebcw930011mm Udbfla

COX ARISON OF EULTIPLE LINEAR CORRELATIONS WITHIN CATEGORIES

'U

 

Table 24 summarizes the results of the within category multiple
linear correlation analyses in terms of variance accounted for by each

category of variables.

 

Table 24. Summary of within variable category multiple linear
correlation analyses

 

Variable category Variance accounted for

By R with all variables By R.with variables

 

within a category retained after least
included in equation squares deletion
Parent media use .14 .12
Respondent media use .18 .15
Family cohesiveness .05 .00
Community integration .03 .00
Self orientation .08 .05
Consumer orientation .08 .04
Demographic .06 .04

 

From the table above, we find that the two strong classes of pre-
dictors are respondent media use and parent media use. Not as

strong but still contributing some predictive power are variables in
the self orientation, consumer orientation, and demographic categories.
Two categories -- family cohesiveness and community integration --

contribute very little to variance accounted for.

F.)
l‘x.)
\1

RESULTS OF MULTIPLE LIICE.";R CORRELATION ANALYSIS ACROSS VARIABLE CATEGORIES

Two multiple R analyses were done across variable categories.
The first took the ”best” predictors from each variable category and
included them in one multiple R equation. The second multiple R was
computed on all 62 predictor varuables regardless of category.

Hultiple R of "best" within category'predictors. Each of the

 

12 variables retained from the within category multiple correlations
was included in one multiple R equation. The resulting multiple R
equals .50, accounting for 25% of the variance in the criterion
variable. ITable 25 reports the results of this "best" predictor
anllysis. Variables were then deleted from the equation by the
Suares deletion criterion fully explained in Chapter III. At
the end of the deletion process, the resulting multiple R equalled
.48, accounting for 23% of the variance in the criterion variable.
Of the original 12 "best" predictors, 8 remained. The feur which
were deleted include: frequency of reSpondent radio listening,
variety of respondent book preferences, reSpondent knowledge level:
math, and variety of respondent Spending. The variables retained

in the deletion process are listed in Table 25.

.mooo.V a pm nnmonmwcwflmb

.momzamcm mnowonmo :wcpns ecu cw pocwmnon moabmflnm>

men Mo mnmna now mm pom .ma .wa .m .o modem“ mom .COMHmSGo oawcwm m oncw momzawom mnomwnwo oabmnnm> canvas

men now mmmoona cemwoaoo monmsom nwmoa esp nonmo poowmnon moabmnnm> can Ham pom mnmxamcm m mamwuHsE weeks

 

128

 

 

ma.o Hm.n omwnmonm COnnmmsooo "wonmvm om
.ma. ma. . unpono co opzpwpn<
2H. ma. pcoemoameo 050; opnwnso m.pango
ma. ma. pmoe commas mace:
ma. mm. om: owpmn m >pownm>
on. ma. oocopconno ow>0E m mocoswonu
om. am. we: enemn agenda anownm>

ma. ma. mcfisow> >9 neonmm ucooe< mm. Ame. mm._ pom.

mm m . . mm m

cownoaoo connmsuo

monsoon nmmoa nonmm on oopsaonm moanmwnm>

n- >2 0n n one: .uscnmson moabmflnm> new: nonowoona :nmobz Ham new;

one
anon o eonescm newoa nonno cocnonon woabmnno> s Amv :0nnoaonnoo oaannasz mo osam>
mownomondo

dancers; STEM: .2an unOQUfioon; :Hmon: .37.... .....Cir.n.i.t..: :Cmumaonnoo nosing. oaawwabz uno baboon .nm muddled.

11 62 predictor variables. When all 62

v-
r)
y.)
0
7.)
O
"h
Cu

 

predictor var

A

bles, regardless of category, were included in one

F J
(J

Multiple R equation, the resulting R equalled .65, accounting for 42%

of the variance in the criterion variable. (See Table 26 on the

next page). When variables were deleted by the least squares criterion
described in Chapter III, 11 variables remained. The resulting

multiple R equalled .54, accounting for 29% of the variance in the
criterion variable.

A comparison of this overall multiple R with the analysis of the
”best" within category predictors shows a great similarity between
them. The differences between the two analyses are most apparent in
terms of the size of the multiple R before the least squares deletion

I
fprocess. The 12 ”best" predictors from the within category analyses account

)

'1

for 25% of the variance in the criterion while the entire set of 62
predictors regardless of category account for 42% of the variance.
A-ter the deletion process, however, the "best" predictor R accounts
of the variance while the "overall" R accounts for 23%.

So, in terms of the least squares criterion, the predictive power
resulting from either analysis is somewhat comparable.

The other difference between the two analyses is the larger
number of variables retained after the least Squares deletion in the
overall analysis. For the I‘best" predictors analysis, 8,of the
original 12 variables were retained. For the "overall" analysis, 11

i'.he original 62 variables were retained. Seven of the variables
retained in both analysis are identical. These variables are

1

Sunscripted in Table 26 with a small "a."

pom .moHnnos memos peonmm mo nonesn .QOHvoncoHno mnHamosw :: omonn mo oonns .mHmmHmcm nonunponm :nmob:

onp CH poononon Hon ono: :w: pmHnombsm m Ann: poxnme poo moHannm> one .mHmmHocm mng cH non nob mHthoom
pony oH oonHmuon mm; 1: wcHecoaw m.oHHno mo knoHno> In pooHno> one Amm oHeoe oomv .moHnowonoo pooHno> :«
thaw: mnOHoHponm :pmob: one no mammHmoo m onHnHSE one now ooCHwnon onomHnm> m one mo 5 one moHoonno> omoasm

 

mH.a HN.|0 .
ocHnmona HmcoHnoooooo “msnmnm omo
mH.u mo.u eoHyonooHnomtmmmonw
mH. mH. unpono no opoann<m
:H. :H. onnnoz moooE m nobezz
mH. NH. oocmpnonpm Lonszo >ocoswonh
mH. mH. nmoE oomeE oHoozm
mH. mm. om: OHpmn m xnoHnm>o
mH. NH. oms oHpon m posos<
mw mH. SH. oocmooonpo oH>OE m mooosoonhm
1
pH. Hm. om: Onoon “condo monno>m
:H. mH. mcHon> >9 peonom ncsoe< . . . .
m mm 03m me One
we x mm m
cOHpoHov ooprovo
n HmHnnoe >2 on n pooHnm> monsoon unmoH nopmm :H popsHocH mpomHnm>
oosnoson moaooneo> sun: sononoono mo Ham can:
coHnoHow monmzoo unmoH nopmm UoCHonon mpozflnm> . Amv cowumHonnoo oHaflstz mo ode>

ll

1

mnemonoo oHnoHno>
mo commenomcn epoeHno> nononeone nHe we mHthocm :oHnoHonnoo noocHH onHPHSE no pstom .mm pooa

 

Mi-
-

he fact that both analyses retain seven ”common” variables suggests
that the logical categorization of variables utilized for this study

has some empirical merit.

 

frequency of church attendance -- were not among the "best" predictors as
determined by the within category multiple Rs. The other -- frequency

of reSpondent radio use -- was among the "best" predictors but was
deleted from the final "best predictors" R.

bThis multiple R put all 62 predictor variables into a multiple R
equation regardless of their variable categories.

CSignificant at p (.0005.

 

CHAPTER V

SUMMARY AND CONCLUSIONS

Summary

Of the 62 predictor variables in this study, 21 were significantly
related to the criterion variable -- frequency of child television
viewing. These 21 variables, listed by variable category, were:

Parent media use: The best1 predictors were amount of parental

 

television viewing and variety of parental radio use. Other significant
correlates were variety of parental television viewing and parental

News sources. All four variables were positively related to the

s . l .
“asunncent media use: The best predictors were amount of

 

respondent radio use, variety of respondent radio use, variety of
reSpondent book preferences, frequency of respondent movie attendance,
and media reSpondent would miss most. One additional variable --
reSpondent preference for comics -- was a significant correlate. All
six variables were positively related to the criterion.

Family cohesiveness: None of the variables in this category

 

were significantly related to the criterion.

 

Community integration: Significant correlates were respondent

gregariousness: organizations and frequency of family church attendance.

1These variables are "best” predictors within categories as de—
termined by the multiple R analyses. See later section of this summary
for explanation.

132

Both variables were positively related to the criterion. This
a. I" ”2 '
category had no ‘best predictors.

. . . 2 .
Self-orientation: The nest predictors were reSpondent knowledge

 

level: math and reSpondent outside home employment. These two variables
were also the only significant correlates in this variable category.

The knowledge level variable was negatively related to the criterion;
respondent outside home employment was positively related.

0 O 2 O 0
Consumer orientation: The best predictors were respondent attitude

 

on credit and variety of respondent Spending. Other significant
correlates were number of parental money worries and parent Spend-save
orientation. Parent spend-save orientation was negatively related to

ne criterion variable; the other three variables were positively

1 2 O I .
emography: Tne best predictor was SOCio-economic status:

 

occupational prestige. Other significant correlates were socio-economic
status: possessions index and sex. The best predictor was negatively
related to the criterion and the possessions index was curvilinearly
related. On the third variable (sex), boys were significantly heavier
viewers than girls.

Two statistical measures were used to tap the relationship of
each predictor to the criterion variable -- chi-squares and Pearson
product moment correlations. Reasons for using both measures were

detailed in Chapter II. The variables listed above are those that

 

2These variables are ”best” predictors within categories as de-
termined by the multiple R analyses. See later section of this summary
for €Xample .

134

were significant on one or both of the statistical analyses. The two

rt

analyses agreed that ll of he variables were significantly related

to the criterion. Of the remaining 10 variables, 3 were significant

only in he chi—square analyses and 7 were significant only in the r
analyses. Discrepancies between the two analyses were for the most

part small. Thus, if a variable was significantly related in one
analysis, the trend was clearly evident in the other analysis. The
reasons for the discrepancies between the two analyses were: 1) extensive

collapsing, particularly on the criterion variable, for the chi-square

lyse , depressing the chi-squares; 2) restriction of range and

OJ
{J
in
0)

skewed distributions on predictor variables, depressing the rs; and 3)
curvilinear relationships between predictors and the criterion,
depressing the rs.

In Chapter II, 51 different hypotheses were formulated. The
results confirmed 13 of these and showed significant findings in a
direction Opposite to that predicted for 6 hypotheses. The variables
involved in the 6 counter-findings were:

1) variety of parental radio use;

2) amount of reSpondent radio use;

3) variety of reSpondent radio use;

u) reSpondent gregariousness: organizations;

5) frequency of family attendance at church; and

6) respondent outside home employment.
For all six, hypotheses predicted negative relationships to the criterion
variable while findings indicated positive relationships. The hy-
pothesis for one additional variable -- socio-economic status: possessions

index -- predicted a negative relationship while findings indicated a

:significant curvilinear relationship.

O
:J
C)
O
rt
:3
(L3
'3
'U
a
O
(I)
O
O
l”h
r f
.7
H
0')
U]

tudy was to analyze the nature of the
relationships existing between the predictor variables and the criterion.
For this, correlation ratios or Etas were computed. Variance accounted

‘

or in the criterion by tne ‘ta curvilinear correlation was compared to

['1

ccounted for by the linear r. While significance tests

4
m
’5
it

,
g,
0
(D
m

were not computed to test the significance of the difference in variance
accounted for, inSpection of the results shows clearly the Etas account
for more variance than the rs. While the linear rs accounted for

from O to 9% of the variance in the criterion, the curvilinear Etas
accounted for from 5 to 18% of the variance. The discrepancy in
variance accounted for by the two meaSures ranged from H% to 14%. As

tcrfiative conclusion, then, the predictor variables in this study

(I)

explain more of the variance in the criterion variable with a curvilinear
model than a linear model.

Despite the indications of curvilinearity noted above, one purpose
of the present study was to do a multivariate prediction of frequency
of child television viewing. Since the easiest multivariate method
available is the linear multiple regression, multiple Rs were run .
both within and.between variable categories. Results showed that the

respondent media use category of variables accounted for the most

 

variance in the criterion (18% with an R of .42). Parent media use
variables accounted for the second greatest amount of variance (14%

with an R of .37), followed by self-orientation variables and consumer

 

for 8% of the variance with Rs of

firs:r:a"*-fb: '5 r-av-n.-"*‘v q) -r~."‘ - ”5+!“
\y-—VL‘£C—;Coa ‘a--C.-J-eu ecbdd QCCOL‘L-‘&

UQ

 

.28). Demographic variables accounted for 6% of the variance (with an

 

AA
(JO

R of .25). The multiple Rs for both family cohesiveness and

 

communitv orientation variables were not significant.

 

When the least squares deletion criterion was applied to each
of the multiple Rs within categories, the variance accounted for was

reduced a maximum of 49. Of the 21 variables significantly related to the
criterion by the X2 and r analyses, 12 remained as "best" predictors
within categories after the least squares deletion process. These 12
variables were then included in one multiple R equation. The resulting
multiple R equalled .50, accounting for 25% of the variance in the
criterion. The least squares criterion was also applied to this
multiple R. After the deletion process, the resulting multiple R
I

equalled .us, accounting for 23% of the variance in the criterion. Of
the 12 ”best" predictor variables included in the equation, 8 remained
after the deletion process. These variables were:

1) amount of parental television viewing;

2) variety of parental radio use;

3) variety of reSpondent radio use;

u) frequency of respondent movie attendance; -

5) media respondent misses most;

6) respondent outside home employment;

7) reSpondent attitude toward credit; and
8) socio-economic status: occupational prestige.

Conclusions

 

The four purposes of this study were: 1) to replicate much of
the past work on correlates of frequency of child television viewing;
2) to go beyond a replication by including some variables which have
been partly or wholly overlooked in past work; 3) to extend the analysis

to a multivariate method; and 4) to Specifically look at the types of

137

’2
J
p:
0»
H
,4
O
S
0

"Lips which exist between the criterion variable and its

this study will be discussed in terms of these four

ptxrpcses. First, in terms of the replication function, results here

semen much like results from the host of prior studies done in the area.

at the collective results of the many television—children

'es, several researchers (e.g. Meyersohn 1957, Schramm 1961, Himmelweit

=958) termed the overall picture somewhat confusing and inconclusive.

Considering the impact that television is popularly thought to have,

researcn re

sults generally show relatively few significant relationships

V

Jeen predictor variables and television use. The present study

I

.3ea:s to be no exception. Of 62 predictor variables, 21 are significantly

And, as has been the case in past work, the

vari”nce explained by any one predictor is small. Given the best

offered by the curvilinear correlation ratio,
Eta) the most variance any one predictor variable accounts for in the
criterion is 18%. Since the size of an Eta only indicates variance
potentially explainable and says nothing of the nature of the relation-
ship, knowing that 18% of the variance in the criterion may be accounted

for by a predictor is not a great deal of information. Some complex

, of curve-fitting operation would be needed to fully use this pre-

dictive power. With 82% of the variance still unaccounted for, a

complex curve fitting procedure doesn't seem to have merit. Given a

linear prediction model, the situation is even worse. The best single

predictor accounts for only 9% of the variance in the criterion by the

ranalysis.

138

Such results really are logical, if not encouraging statistically.
Television viewing is obviously a complex behavior and one that can
only be viewed within the context of the respondent's total life

situation. As researcher after researcher has noted, complex behaviors
do not have simple or single causes (e.g. Schramm 1961). One should

not expect that any one variable will account for an overwhelming amount

of variance in television behavior.
If the problem were just one of low predictive power, the

:ntuation might be clearer than it is. But, as one looks at past

:findings, several confounding trends emerge. First, there are a great

0

many contradictory findings in the literature.

“A

A review of the

~ ‘ O O
hypotheses stated in Chapter II shows that only one-third of the 51

otheses were based on clear-cut evidence -- e.g. evidence

(J)

-~ 13 ‘* v“
tatc \A .11“. k)

agreed on the direction a variable would relate to frequency

-

of child television viewing. For another third of the variables,

evidence was contradictory with past work showing a complex of
signif'cant and non-significant findings and, in some cases, significant

indings in two different directions. For another third of the

H1

hypotheses, evidence from past research was Sparse or not available.

he present study does little to clarify the total picture.
For example, only Six of the 21 significant results in the present

study agree completely with past work. Thus, this study found these

variables related to child television viewing in the same direction as

ts from prior studies: amount of parental television viewing,

”1’3 _
‘93

D
(a
U)
'd
O
I)
(A.
0
:3
d
’O
'5
(I)
"h
(D
t
a
:3
0
0
F1»)
0
*i
O
O
53
p.
O
0')
*3
G)
0)
’U
0
:3
Li.
(D
:3
d
X.
:3
O
2
[..J
(D
Q;
m
(D
...:
1:"
(D
l-‘
(n
5
w
d-
{3"
U

eaconomic status: occupational prestige.

.1
.)

ile past work agreed that such variables as parental per-
:nissiveness, family togetherness, respondent library use, reSpondent
tiobbies, and reSpondent knowledge levels Should be significantly related
‘to amount of child television viewing, the present results were all
Iion—significant. Clouding the picture even more are the contradictory
‘results obtained in the present study. Most notable of these are the
tiesults for radio uSe variables. The present results Show amount and
variety cf respondent radio use and variety of parent radio use

v related to frequency of child viewing. Yet, past work
points to either negative or (according to the most recent findings)
non-significant relationships. The current study also finds reSpondent
_ gregariousness: organizations, and family church attendance positively
related to the criterion while past work suggests either negative or
non-significant relationships.

Another trend from the past research is the emergence in recent
studies of non-Significant relationships for variables which in earlier
studies were reasonably good correlates of television usage.‘ As an
example, we find that recent studies agree that the relationship among
various aspects of media usage are tending to be non—significant
(Edelstein 1966). Results from the present study generally agree.

In addition, the present study showed nothing but non-Significant

relationships for two major classes of variables -- family cohesiveness

 

1.

community integration -— which had, in the past, been somewhat fruit-

.9.)

'1']
’4

’1'!

predictors 0 television viewing.

Of course, confounding all predictions is the problem of the leap
frxxn he many television versus no television studies to correlates of
anxount of television viewing. Relatively few findings are available
trust eXplicitly looked at correlates of amount of viewing. Even

firmer findings are available from data collected in the mid-sixties
witji'television at almost 100 percent saturation.

The probable impact of the changing media environment is seen

znost clearly in terms of a review of the five generalizations derived
frcn past reseirch (outlined in Chapter I). As a brief review, these

rive generalizations were:

l. :he parental imitation generalization: children tend to do

 

rt‘

what heir parents do, all other things being equal.

2. The demographic attributes of the child generalization: certain

 

demographic attributes of the reSpondent predict his television usage.

3. The functional displacement generalization: the child

 

will sacrifice in lieu of television those activities which satisfy
the same needs as television but do so less effectively.

4. The frustration generalization: the more frustrated a child

 

is, the more time he will Spend in front of his television set.

5. The information void generalization: television is likely

 

to have an effect on a child's values and outlooks if a) the values and
views recur over and over again in TV content; b) the content is linked

to the child's needs and interests; c) the form of presentation is

11+].

tiranatic; and d) the child is not presented through peers or relatives
Iiith a standard against which to assess the views offered on television.
nesults from this study generally support the parental imitation
ggeneralization with amount of parental television viewing and parental
:socio-economic being two of th ”best” predictors. In this reSpect,
the present findings are in agreement with the major studies of the
Ipast (e.g. Schramm 1961, Bailyn 1959, Himmelweit 1958).

While generalization #2 —- the demographic attributes of the
child generalization -- applies more to television usage changes through-
che childhood years, the present study offers some support. The

V

one variable closest to a measure of intelligence —- reSpondent
knowledge‘level: math -- is one of the strong predictors, in agreement
witn past work (Himmelweit 1958, Bailyn 1959). In addition, sex was
significantly related to viewing in the current study while all past
work found non-significant relationships.

The two generalizations which are least supported by the current
study are the functional displacement and frustration generalizations.'

While the functional displacement generalization should predict results

for most of the reSpondent media use variables, as well as some of the

H)

anily cohesiveness, community integration, and self orientation
variables, we find little support for the generalization in the present
study. Past work would suggest that use of such media as radio would
be displaced by increased television usage. The present study finds
the Opposite -- a positive relationship. Past work would suggest that

family activity levels and family use of outside home entertainment

 

Should.be displaced by increased television usage. The present study
inds non-significant relationships for these variables.

These results bring the whole question of what is "functional” to
“fine fore. It appears as if activities which once were "functional"
equivalents to television usage may not be any longer. This is most clear
with radio usage since radio itself has undergone major changes as a medium
since the introduction of tievision. Within the teenage culture in which
pepular music is a major focus of concern, there seems no reason to
expect that amount of radio use would be diSplaced by increasing television

q

ularly since popular music is the only radio content for

t
u
'L‘}

L:

9

rt

’4

O

"Lny stations. various researchers (e.g. Himmelweit 1958, Katz and
have warned that inferences about "functional equivalents"
can not be safely made without some empirical support. What may be
functionally equivalent for one group of reSpondents may not be for
another.

The frustration generalization receives even less support from the
present study. As a predictive tool, it should apply to most of the
variables measuring family cohesiveness and community integration in
this study. Yet, the results show that not a single family cohesiveness

variable is significantly related to the criterion. Further, the two

significant community integration variables -- respondent gregariousness:

O

rganizations, and frequency of family church attendance -- are related
positively to frequency of child television viewing while the frustration
generalization predicts a negative relationship.

The present results for these two classes of variables are not

geatly different from those in past research. In general, variables

nithese two classes have not yielded a great deal of

n..-

 

 

 

 

 

- I

u

.1

..‘

'-
a

...

 

 

 

 

 

[.1
.r:
(0

predictive power and have resulted in conflicting findings.l

ticularly in the most recent work, we find warnings that the very
behaviors may be indicative of both frustration and lack of
fmstration. Katz and Foulkes (l962) emphasize the point. The child
up is in conflict with his parents may escape to television; the child
zip has a harmonious relationship with his parents may watch more
nflevision in order to be with his parents. The child with few friends

my escape to tievision; the child with many friends may use television

{u

ssa means of relating to them. As with the diSplacement generalization,
itseems erroneous to make inferences about underlying reasons for
vaious television'behaviors just from the behaviors themselves.
Interestingly, the relationships to the criterion of some of the
uniables in the consumer orientation category may also be predicted
inaxtne =rustration hypothesis. For example, to the extent that the
rapcndent sees his parents as worried about money problems, the
fnmtration hypothesis would suggest television viewing would increase.
he problem of whether level of parent money worries is an indicant of
fmstration still remains. Consumer variables, in general, turned out
tdbe among the better predictors in the present study. An equally

gum.rationale based on deferred vs immediate gratification patterns may

Jacrrived to explain results. Again, some sort of explicit check of

‘h—

1
LvVO

nimmelweit (1962) noted the lack of order in results from studies
taxing the relationship of various family variables to frequency of
dfildtelevision viewing. She suggested that television may act as a
anahst showing up the characteristic mode of relationships within the
Ififily. She concluded, however, that the unclear results in the area
Stanin part from lack of adequate measures of family variables. Currently

mednmasures seem too superficial to be related to the "core of family
life.“

inferences about underlying behaviors is called for.

Tne final ”information void" generalization is really tangentially
related to the present study. The one variable which seems most

clearly related to it -- reSpondent attitudes on credit -- was significantly
-elated to the criterion variable in the positive direction predicted.
Again, however, any inference that the reSpondent's attitude toward

credit is related to an "information void" on credit in his family or

peer group might be erroneous. Attitudes toward credit may simply be
another indicant of a basic immediate vs deferred gratification pattern.

In sum, the 'iscussion above suggests a major theoretical obstacle
nets to explain and predict frequency of child usage. Support for
the :encralibations seems hampered by lack of clarity on the "meaning"
cf television viewing to the reSpondent. More will be said about the
question of the ”meaning" of television at the end of this chapter.

Despite the confusion indicated by the individual hypothesis
tests, one encouraging consequent of the present study was the results
on some of the "new" variables which were added to it in addition to
the variables included mainly for replication purposes. All the
xeriables in the "consumer orientation" category were "new". Four of
them were among the 21 variables significantly related to the criterion
and one was among the "best" predictors from the overall multiple regression
analysis. Another "new" variable which turned up among the strong predictors
in this study was reSpondent outside home employment. These results
suggest fruitful areas of eXpansion. The consumer dimension of the

family’s behavior, in particular, seems a lucrative area of focus because

1145

1.

t

H)

0
*h

.e o sible relationship 0 immediate vs deferred gratification

' C}
m

'U

atterns to both consumption behaviors and media behaviors.

Another encouraging result of this study was the fact that the
:nultivariate analysis isolated in its group of "strong predictors"
the two predictor variables most often found in past work -- amount
of parent television viewing and family socio-economic status

(occupational prestige). To this extent, then, the present analysis

agrees with past work.

3-1

n general, of course, the "strong" predictors in the multiple R
analyses a:e those predictors which were most highly related to the
ion variable in the individual hypothesis tests. Thus, comments

I
made earlier on the results to the individual hypothesis tests apply

Another encouraging aspect of the multiple R analyses is the
amount of variance that can be accounted for in the criterion given the
restrictions of a linear model. The eight "best" predictors account for
23% of the variance. This cerainly is not overwhelming but is a sizable
increase over the 9% accounted for by the one single variable with the
highest correlation (r) to the criterion.

In terms of full explanation of television viewing, however, the
restriction of a linear model appears to be a serious one. The analyses
of the Eta curvilinear correlation ratios showed that the relationships
between the predictors and the criterion are accounted for more fully
by a curvilinear than a linear function. As suggested earlier, however,

the amount of variance accounted for even with a curvilinear function

is; at no point remarkable. The best variable accounts fer 18% of the
*vardence in terms of Eta. Thus, any extended foray into plotting curves
for~the relationships of each predictor to the criterion is certainly
not suggested.

However, these results do suggest potential for the use of a
uniltivariate technique that makes no assumptions about the nature of
the functional relationship between predictors and the criterion. The
size of the Etas obtained suggests that no one variableoffers great
3predictive power but that some complex combination of predictors might.
ssentially, this is the kind of question which the multiple R analysis
aspts to answer. But, the multiple R assumes linear functions and
what is needed is a multivariate technique that makes no such assumption.
Tne possible technique would be a configurational approach, such as the
Automatic Interaction Detection (AID) method developed.by Sonquist and
Morgan (1964).

All the discussion above has suggested different analytic
techniques and inclusion of new variables in the attempt to explain
and predict the frequency of a child's television viewing. Throughout
the discussion an underlying difficulty has been the problem of what
’generalizations or inferences may be made about the results of a
correlational analysis of frequency of child viewing.

The results of this study as well as the more recent discussions
by researchers in the area suggest that inferences about the "meaning"
of television viewing need to be explicitly checked.‘ As noted earlier,

two of the more often cited generalizations on television behavior

147

require the researcher to make inferences about frustration states in
respondents and about functionally equivalent activities. These

'nferences require that assumptions be made about the "meaning" of

}_l

(’1’

elevision as an activity to reSpondents. Is television an escape from
real world frustration, such as having too few friends? Is television

a means of entertaining friends? Is television a substitute for missing
communication with parents? Or, is television a means of sharing an
activity with parents?

A major difficulty seems to be that, in today's ubiquitous
television environment, the very same "end" behavior -- frequency of
television viewing -- may have very different meanings for different
respondents. This author does not intend to suggest that frequency
is not an important or relevant variable. Rather, looking at television
viewing as an end unto itself seems incongruous when one considers that
television is one artifact in an environment filled with artifacts.
Television usage might more lOgically be seen as a type of intervening
variable in the reSpondent's life. This idea is certainly not new as
such researchers as Meyersohn (1957), Himmelweit (1962), Bauer (196u),
Schramm (1961), and Troldahl (1965) have all called for an explicit
attempt to look at the "meaning" of television exPosure. This study
seems to make the need even more pressing.

The question would become more complex than what are the correlates
of amount of television viewing. Rather, the question would become
what are the correlates of various needs of the reSpondent and how does
television (as well as other environmental artifacts) intervene in

these relationships.

148

This approach would require that previously made assumptions

about th

(I)

nderlying reasons for behaviors be checked. For example,

it would not be assumed that reduced child-parent communication leads

to frustration on the part of the child. Either some measure of anxiety
would be used or reSpondent introspections would be collected. These
data would allow the develOpment of various typologies of the relation-
ship of child-parent communication to anxiety. Television usage would
then be analyzed within the context of these typologies. As another
example: varying levels of child gregariousness could be related to
varying levels of reSpondent reports of need for affiliation. Television
e could then be seen within the context of the relationship of actual
ariouénass to need for affiliation.

Once these relationships have been established between various
reSpondent and family attributes and resulting needs or states,
television usage may be analyzed as it "intervenes" in these relation-
ships.

Frequency of television viewing would still be a meaningful
measure. However, given that the very same end behavior -- frequency
of viewing - may serve different functions-for different respondents,
an attempt to tap the "meaning" of television in the child's life
would need to go beyond pure quantity measures. Other possible
measures include:

1) Respondent intrOSpections on why television is wanted. This

 

ap_roadh -- simply asking reSpondents why they like television in
general and why they like certain shows —- has been suggested by Schramm

(1961). Troldahl (1965), and others.

149

2) Quality of espondent viewing. Some work has been done

 

andlyzin*r the content of child viewing. Such a content orientation
would be crucial to understanding how the same amount of exposure may
serve different purposes for different respondents.

3) Respondent perceptions of TV content. Not only might the same

 

amount of exposure have different meaning for reSpondents but the same
content of exposure might have different meaning. Various researchers
(e.g. Schramm 1961, Bailyn 1959) have suggested that two different
children viewing the same program will select content which suits their
own particular needs.

important aspect of the search for the "meaning" of television

E;

would be 118 emphasis on the medium as one of many artifacts in the

environment. As Katz and Foulkes (1962) have pointed out, if a child

has a need to escape a certain kind of frustration, there are many

alternative routes for doing so. Analyzing television as one of many

intervening artifacts or activities in the childs life seems like one

way in which the problem of "functional equivalency" can be attacked.

To the extent that different media continually serve as "intervening

variables" in the same way, they may well be "functionally equivalent."
Before concluding this chapter, a few cautions must be applied.

Conclusions based on the present study face several serious limitations.

The most constraining are: l) the uncertainty of just how the questionnaire

was administered to the reSpondents; 2) the form of measurement of

*mndables in the original questionnaire; and 3) the fact that reSpondents

mere all students in one rather homogeneous suburban school system.

“‘1'!
*U

In conclusion, the major directions suggested for future research

ave included:

:34

discussed in this chapter

1. The need to include classes of variables (e.g. consumer
behavior) previously ignored in attempts to predict child television
viewing;

2. The need to use multivariate techniques to predict television

3. The need to use statistical techniques which make no assumptions
about the nature of the functional relationships between predictors
and the criterion;

Q

a. [he need to look at television viewing as an "intervening
behavi r” whibh mediates various respondent and family attributes and
-eSpondent needs; and

5. The need to look at aspects of television usage otherithan

quantity -- content of exPosure, reSpondent perceptions of content,

and respondent introspections on why they watch television.

BIBLIOGRAPHY

Abrams, Hark. ”Child Audiences for Television in Great Britain," Journalism

 

Quarterly, 33, 1 (Winter l956), pp. 35-u1,

Bailyn, Lotte. "Kass Media and Children: A Study of Exposure Habits
and COgnitive Effects," Psychological MonOgraphs, 73, l (1959),
Whole No. 471.

 

"P“

Bauer, Raymond A. The Obstinate Audience: The Influence Process from
the Point of View of Social Communication," American PsychOIOgist,

 

 

'lliam S. "The Mass Media and Young PeOple," Journal of
.adcasting, 51, 1 (Winter 1960-61), pp. #9-58.

 

Foundation, 196%, I, pp. 169-208.
Belson, William A. "The Effects of Television on the Reading and the
Buying of Newspapers and Magazines," Public Opinion Quarterly,
25, 3 (Fall 1961), pp. 366—381.

 

"Television and Family Life," Advancement of Science, XVI,
54 (1960), pp. ans-353.

 

. ”The Effects of Television upon Family Life," Discovery, XXI,
l0 (1960), reprinted by the Research Techniques Division of
London School of Economics and Political Sciences, Reprint #127.

. ”Effects of Television on the Interests and Initiative of
Adult Viewers in Greater London," British Journal of Psychology,
50 (May 1959). pp. 145—158.

 

Bogart, Leo. "American Television: A Brief Survey of Findings,"
Journal of Social Issues, 18, 2 (March 1962), pp. 36-42.

 

. The Age of Television. New York: Frederick Ungar, 1958.

_

Cambell, W.J. Television and the Australian Adolescent. Sydney,
A'stralia: Angus and Robertson, 1962.

 

Caplovitz, David. The Poor Pay More. New York: The Free Press, 1963.

 

Quter,1tichard F. and Bradley S. Greenberg. "Newspapers or Television:
[dutch do you Believe?" Journalism Quarterly, 42 (Winter 1965),

 

lSl

152

 

Clarke, Peter. ”Parental Socialization vehes and Children's Newspaper
Reading,” Journalism Quarterly, 42, H (Autumn 1965), pp. 539—546.
Clausen, J hn A. ”Family Struct are, Socialization, and Personality, "

o
n Lois liladis .offman and Ifartin L. Hoffman, ed., Review of

Develonm ent .-seuroh New York: Russell Sage Foundation,
966, II, pp. 1-53. ‘

 

Coffin, Thomas E. ”Television's Impact on Society," American
Psychologist, 10 (1955), pp. 630-6ul.

 

Coleman, James S. The Adolescent Society. New York: The Free Press,
l96l.

 

Douvan, E. and Adelson, J. The Adolescent Experience.' New York: Wiley,

 

 

1966.

and} Iartin Gold. "Modal Patterns in American Adolescence,"
i. Lois Eladis Hoffman and Martin L. Hoffman, ed., Review of
Child Development Research. New York: Russell Sage Foundation,
1966, II, pp. #69- 528.

n, Alex. PerSpectives in Mass Communication. Kobenhavn:
Liner har cks Forlag, 1966.

 

Eine, B.J. and N. Maccoby. Television and Family Life (mimeo). Boston:
Boston University School of Public Relations and Communication,
1952.

 

Freedman, Lawrence Zelic. "Daydream in a Vacuum Tube: A Psychiatrist's
Comment on the Effects of Television," in Wilbur Schramm,
Jack Lyle, and Edwin B. Parker, Television in the Lives of Our
Children. Stanford: Stanford University Press, 1961, pp. 189-194.

 

Eteiw‘ on, Eliot. "The Relation of the Social Situation of Contact to
the Media in Mass Communication," Public Opinion Quarterly, 17
(1953), pp. 230-238.

 

. "Communications Research and the Concept of the Mass," in
Wilbur Schramm, ed., The Process and Effects of Mass Communication.
erana: University of Illinois Press, 1955, pp. 380-388.

 

(biger, Kent and Robert Sokol. "Social Norms in Television Watching,"
American Journal of Sociology, 65 (1959), pp. 174-181.

 

Hafilton, Robert V. and Richard H. Lawless. "Television within the
Social Matrix," Public Opinion Quarterly, XX, 2 (Summer 1956),
pp. 393-404.

 

153

 

innolweit, Hilde T. “A The retical Framework for the Consideration
of the Effects of Television: A British Report," Journal of
Social Issues, 18, 2 (IMa ch 1962). pp. 16- 28.

 

, A.N. Oppenheim, and Pamela Vince. Television and the Child.
London: Oxford University Press, 1958.

 

Huber, Milton J. A Study of Financially Over-extended Families (mimeo).
Detroit: Merrill-Palmer Institute, 1965.

Kahl, Joseph A. and James A. Davis. "A Comparison of Indexes of Socio-

Economic Status," American Sociological Review, 20, 3 (June 1955),

pp. 317- 325.

Eatz, Elihu and David Foulkes. "On the Use of the Mass Media as 'Escape':
Clarification of a Concept," Public Opinion Quarterly, XXVI, 3
(Fall 1962), pp. 377-388.

 

 

d oper, Joseph T. The Effects of Mass Communication. New York:
The Free Press, 1960.
:;.:arofeld, Laul F. "Some Reflections on Past and Future Research on

Broadcasting," in Gary Steiner, The People Look at Television.
New York: Alfred A. Knopf, 1963, pp. MO9-422.

, and Patricia Kendall. "The Communication Behavior of the
Average American," in Paul F. Lazarsfeld and Patricia Kendall,
Radio Listening in America. New York: Prentice-Hall, 19u8.

Lyle, Jack. "Immediate vs Delayed Reward Use of Newspapers by
Adolescents," Journalism Quarterly, 39, 1 (Winter 1962),
Pp. 83-850

inccoby, Eleanor E. "Effects of the Mass Media," in Martin L. Hoffman
and Lois Wladis Hoffman, ed., Review of Child Development
Research. New York: Russell Sage Foundation, 1964, I, pp. 323-348.

 

. "The Effects of Television on Children," in Wilbur Schramm,
ed., The Science of Human Communication. New York: Basic
Books, Inc., 1963, pp. 116-127.

"Why Do Children watch Television,"-Public Opinion Quarterly,
13, 3 (Fall 195u), pp. 239-244.

 

________3 "Television, its Impact on School Children," Public Opinion
Quarterly, 15 (1951), pp. u21-4uu.

 

Kdmmar, Quinn. Psychological Statistics. New York: John Wiley and
Sons, Inc., 1962.

Heycnssohn. Rolf B. ”Social Research in Television," in Bernard
Rosenberg and David Manning White, Mass Culture. New York:

The Free Press, 1957, pp. 345-357.

 

 

, .win B. "The Effects of Television on Public Library
Circulation,” Public Opinion Quarterly, 27, 4 (Winter 1963),
pp. 578—589.

”’1'

. Changes in the Function of Radio with the Adoption of
Television," Journal of Broadcasting, 5, 1 (Winter 1960-61),
pp. 39-48.

 

 

Beirmuater, Lee, Richard P. Coleman, and Gerald Handel. Workingman's Wife.
New York: MacFadden Books, Inc., 1959.

 

Remmexms, H.H. and D. H. Radler, The American Teenager. New York:
Bobbs-Merrill, 1957.

 

Riley, iflJJ. and S. Flowerman. "Group Relations as a Variable in
Communication Research," American Sociological Review, 16, 2
(1851), pp. 178-179.

 

and'John W. Riley Jr. "A Sociological Approach to Communications
Research," Public Opinion Quarterly, 15 (1951), pp. 445-460.

 

Samuelson, Merrill, Richard P. Carter, and Lee Ruggels. "Education,
Available Time, and Use of Mass Media," Journalism Quarterly,
90 (Fall 1963), pp. #91-u96.

 

Stanley. The Psychology of Affiliation: Experimental Studies
0 the Sources of Gregariousness. Stanford: Stanford University
Press, 1959.

Schneider, Louis and Sverre Lysgaard, "The Deferred Gratification
Pattern: A Preliminary Study," American Sociological Review, 18,

Sfuemm, Wilbur, ed., The Effects of Television on Children and
Adolescents: An Annotated Bibliography with an Introductory
Overview of Research Results. Paris: UNESCO Mass Communication
Techniques Division, 1964.

. "Mass Communication," Annual Review of Psychology, 13 (1962),

 

, Jack Lyle, and Edwin Parker. Television in the Lives of Our
Children. Stanford: Stanford University Press, 1961.

 

, Jack Lyle, and Edwin Parker. "Children's Learning from
Te evision," Studies in Public Communication, 3 (Summer 1961),
DD. 86-98.

 

 

fScott, Lloyd. ”Relationships between Elementary School Children and
Television," Journal of Educational Research, 52, 8 (December
1958), pp. 13%-1370 '

Siegel, Sidney. Nonparametric Statistics for the Behavioral Sciences.
New York: McGraw-hill Book Co., 1956.

 

Sonquist, John A. and James N. Morgan. The Detection of Interaction
Effects. Ann Arbor: Survey Research Center, Institute for Social
Research, The University of Michigan, Monograph No. 35, 1964.

 

Steiner, Gary A. The People Look at Television. New York: Alfred A.
Knopf, 1963.

 

Troldahl, Verling C. and Daniel E. Costello. Mass Media Exposure Patterns

 

and Interpersonal Community Behavior of Teenagers. Paper
read at the Theory and Methodology Division Sessions of the
Aspociation for Education in Journalism Convention, University
of Iowa, Iowa City, Iowa, August 1966.

 

r7“ "

.roidahl, Verling C. "Studies of Consumption of Mass Media Content,"
Journalism Quarterly, #2 (Autumn 1965), pp. 596-606.

 

. ”Occupational Prestige Scale," in Verling C. Troldahl and
Robert Van Dam, Public Affairs Information-Seeking from
Expert Institutionalized Sources. East Lansing: Department of
Communication, Michigan State University (mimeo), 1965,
Appendix E.

 

 

U.S. Bureau of Census. U.S. Censuses of Population and Housing,
Census Tracts, Final Report PHC(l)-92. Washington, D.C.:
U.S. Government Printing Office, 1961.

 

 

Westley, Bruce H. and Werner J. Severin. "Some Correlates of Media
Credibility," Journalism Quarterly, 41 (Summer 1964),
pp. 325-335.

 

I

White, David Manning. "Mass Communication Research: A View in
PerSpective," in Lewis Anthony Dexter and David Manning White,
People, Society, and Mass Communication. New York: The Free
Press, 1964, pp. 521-5H6.

 

Witty, Paul A., Paul Kinsella, and Anne Coomer. "A Summary of Yearly
Studies of Televiewing, 1989-1963," Elementary English,
October 1963, pp. 590-597.

 

:itty, Paul. ”Studies of the Mass Media, 1949-1965," Science Education,
March 1966, pp. 119-126.

 

 

 

Yarexng {eon J. “Separation from Parents d'ring Early Childhood,"
in Martin L. Hoffman and Lois Wladis Hoffman, Review of Child
DSVGlOpTQMt Research. New York: Russell Sage Foundation, 196%,
I, pp. 89-136. '

APPENDIX A

ITEMS, CODES, AND INTERJUDGE CODING RELIABILITIES

LISTED BY VARIABLE CATEGORY AND VARIABLE NAME

 
 
 
 

158

econ mo xmeaa ca

.coprcmmeo mom WHH nopdmno mom
.xmocH pooEmoemm owmpcmono

.eOHumcmmeo now HHH novmmso new
.momcommon ooocoueoao peopeommon mo mHmmHmem peoweoo poanvmn Heep omonp one mebmem> Aev poeemwm

.meoooo wooem HomEmmewm mponEoo mopmodoCH
mcv mm; com: hHHHHAmHHon mcHooo oprHLoHCH mo enammme ones

A

 

mmsonHHmnlwmzoc‘wmoHooeoo‘dwmemno

 

 

 

                           

 

 

.meomo .Hmonse .m.coeoz .mmmodo meow om: oHomn

mm nmvoHnm> cmH: .zoxOOH omHo .munomm "ow comeH myconma Hmuaonma

ow BOHV mushy m o 930% or memnmona oHomn mo peHx “we: mo huoHnm>

no>oc u o meHeoumHH

mm oHst :H mono u H «hope: no BoHHns oHomn

moEHHmeom n m m CH mono mem>o .moEHuoEom .coumo Hmueonmm

coumo u m oHemn one o» coumHH wwconmd each on mo pesoe<

oHocomzo:

pom >e o: u o moHnmupom mew meme so; .oco emcu oboe cH mpow

am you H u H o>mn :0» MH moHnmunom w you any mH :onH>onu

meow onoe no N u m meow :OHmH>oHov m :30 mHHemm poo» moon mo nonesz

mmmemev .menopm>E .

.msoa .onomEoo .eoHpon ooemHom wcHsmH>

.muco>o HmHooam .munoam .menovmos conH>oHou

mm AHHoHec> ech .mmnodo meow ”Hume mxHH mesonma Hmuewnmm

ow :oHv m:oznm u o pro» on mzonw mo oeHx was: mo >pmHem>

nope: u o weHon>

rem oHst :H mono u H meo>oe conH>oHow
moEHHmuo u N no .oHHcs.m cH mono revue .m)a_H«m com Hmpcwnmm om: MHooE
coumo u m .copmo conH>oHow nouns mezohme poem on we pesoe< Humane
mnoHHHHHroHH oooo EoHH pomHem> know.owmo
sci ec..ooo poaHem>

oucsH; ozH

m_i:efibeuaL > new xwoemmwwluwmmflmmmndflWMHWflHmfiiflflfiflflrmmfiwwfim tempos ocpsHeopsH pom motoo .wHVHH .nm m:;mmN

159

um

AapoHpm> :an

maumHonmnov eonpo .memnwoem
meme .msoem mpoeoo .mHmHAom no mmeomo
doom .wunOQm .mzozm moxUOH opr "been

om: oHomn
peoocommee

 

on soHv moose m s o oxHH so» on memewoem oHomn we ocHx “we: we xpoHem>
no>me n o
xooz m 2 deep mmoH u H
mm amp m 9302 H u N ano>oe .xoms m
moo m mason mam u m meson : emzv won mmmp m 990: H “roam mm: OHpma
zoo m . whee m meson mnm when m meson mu: HomeGOQmom om: mHeme
meson @905 no m u : "oHomn one ow copmHH 50% or oowmo so: mo pcsoe< pceocoamom
posse
mm mpmmemoeonm m z o mono meamnwoeona
hHHEmm poo» meow mnamewoeonm home 30: mo noberz
«90:90 no .woeHumwme
some .mnodmdmsoc Eonm .:0HmH>oHov Eoem
mm cOHmH>oHow #0: u H .oHdooa ov meHmev Eonm :1 pHeoz may meme
conH>oHou n m eH so mcHom m.pmn3 prone use: eHonp we no mooeoom
pmoe pow mucoema prom xcHnw so» on when: Hmpcopmm
mmoeo pegs "mom mH .wooHwame mom or omen mueoemm
em mocHnommE m I o oanomnom upcoema each on «mHQMHzmme mmeHwamE
omen mueonme 9:02 on mocHnmme pus: mo embesz
mm unease : u o needed szomz peonowHHo.noAE:z
wmm wedded m a o needed mmocsw
paw shame pampmemae pmaesz
mmoHonoz cam ow woanomnsm
.meoemd seesaw .meoema HHHmo In mHHEmm mnemmamzoc
Leo» CH you :o% on whoammmso: Hun; mo nonesz
maHHHHHLmHH owou EouH pomHnm> anomoymo
gem mcHroo oHannm>

ouoszUHCH

Ii

 

 

tr;:Hrzooxism pomb

..Hmurog n O

EOUHmm u H
abooe m mono u m mem>oc .EOpHow .bpeoe w some p50bm
mm cpeoe m mOHsv u m .zpcoe m ooHsp usonm .xoms Agape
xmoz >Ho>o n : "moH>OE may 09 ow so» or emvmo 30m

 

oocmocmupm

.oH>oe ueoocoamme

zoomsoonm

 

eo>oe u 0
mm moEHpoeom n H «copmo so: .mo» mH «semenHH oHHbsm ems memebHH
:oymo u m on“ Eoem pro meOA zooso 50% om Homecoamome
OOH we: oHo u H m>oem mm 05mm moHEoo mom
moHeoo poxooeo u m mononomoea
HemoGOQmom
expand
uxoonov nonpo .mxoon oHeoo .wxoon Ho>mny
.moHpopmme .mmHnomnwOHn .moHeovm HmEHcm moocopomonm
.moHnoum am: .eonon ooeoHom .COHHonnco: goon
O AHHoHem> ean ow .mHo>oc HMUHLOPmH: .onQOHm o>0H .oesuco>om Homecomwon
”m cm ROHV mommy H n o "Hmon oxHH so» on mxoon mo mweHx pee: mo hpoHem>
momma
unopeommoh
mocHNmmme
mm mocHnmmoe a u o mmHansmon omen pom op mocHnmmme wee: mo embesz
mapmHonoeov moHeoo .poonm
comma .msm: huwHoow .aoHuoow >Bn0Homa
.momma w.cmEoz .mwmm HMHocmch .mHchesHoo
arm HHHOHem> .mom Hem: .HmHQOpHom .mzm: eonuo .owma om: nommmwsoa
an; ow sOHV peonm .ocHwamE .munOQm "omen so» Homecommon
moomwoom m I o op nonmawzom mew mo mopooom ems: mo upwHem>
eoHpHHHLoHH ooou 1 1;:ltla _aouH memHnm> Anomopmo
toe acfleoo oHanpm>

omosmnowcH

.epoempcooa-sm passe

pro ow v.emo .E.m m

 

 

 

u m
poéieua .sdmu:
.ed a ... m and 0H ... m
am .ad .2 u m
uancoHE u b meson Hons .mo> HH mmzmexmoz no 950:
ocHHomop o: u m :Hmpnmo m pm cH on ou o>m£ so» on Am
memo: co
II Homoeo>HmmHEeom nww: 204mm mzth weOHuoHeuwmn
ow zOHV mH I Ho 039 mom mmmoom wszzzm rm ammezH "H xmazH
mmmeo>meHenom mmocopHmonoo
awn-age... H353
wo>mHHon ow hHoxHH oo>oHHob ummma
wm o>0bm mm msmm HumoH on 50% oHsos eOHweo> gone... "m eonnm>
mm conH>oHoy Hoe u H mo>oHHob ow hHoxHH popoHHoA “mo:
conH>oHow u w umoe on so» oHsoz cOHunm> :och... “H conne>
11 ...mocHNmmmE .
um no mnodmmmso: :H .conH>mHow no OHpma mwerme
may no HHOPm meow on“ mo meonno> szHHnHomno
HomeommHo omen no oemon :0» MH memE uaooeommom
mmomx so» oHsoz nOHQS
mmoeHNmmmE no .wemdmmmzoe .oHomn pmoe mmHE oHsos
mm conH>oHov no: u H w .pom QOHwH>oHoH m "MCHsoHHom mew Homoeoamee
conH>oHoH u m mo one cho moox pHsoo 30> omomazm . MHpoz
mApmHonmnov
peeve .NhMm .momeoo Hmonze
u m .mpOHmHmon .mHmoo> .mHepeoespwm woodenomonm
mom AmHoHnm> :mH; ICH .HHos c noon .chmE XHom pnoooe Homoaoawoa
op sOHV moemw s u o "oxHH so» or mfinooo_ we oeHx Hob: mo honnm>
co ..m 3.3 an H .H 5me0 ii- -. I..-I----..II. JimmpIH 82... ES Hg
Iom aCHwoo eHanbe>
oaemmnobcH

 

 

pochHcoozlbm QHQUH

«on

no mdb ow MCHow ohm menu wcHevaOm

 

 

mm o>0bm mm oemm gnome :0HnHmo prom 90m 50% xmm
eo>o mHHEmH poo» CH unannonw any on HA
OGHH
mm onHpoeom u m mepoonm 950% psonm
me» u m unnamed prom op mev 50% emu Am
Amesz
I- Acoaumoaaseeoo emu: 204mm memeH memes coapmoaesaeoo
ow onv p I H mze mom mmmoom oszzsm rm ammeZH peonmmIoHHnoe
so o>onm mm mEmm mmpeoxmoz
co pro on.so» con: on HH.soz
opens mweonmd 950% HHoH 50% on no
m“ so o>onm mm meow wmmcho>o
:1 may cH woo $9.50% cos: we HH.:om
when: mucoemm 9:0» HHop 50% on HA
«Hoonom
no» u H popwm poo on.so> dogs on HH.:O%
we eoeHbmeow no 0: n m memes mpemnmm poo» HHow so» on Am masonmoeonz
m.veoocommoa mo
304mm memeH enema mwepazoex heaped
mom mmmoom wszzam wm memQZH "a xmmzH
meson
Hogs amo%_mH wwocoxoos so n50:
msm mpore no meow :Hpueoo m Hm :H on OH one; so» on An
-wamwiHbeHH coco IIII-:I:III!I EoHH pomHem> mnemoumo
In; demoeo GHQmem>
nwmsmhcwcH

IIIIII.

ncesHHCQOIIbm pooe

30an £00 cme: u N
#00900 H 00%: n H

 

 

 

om #00900 «50% pom pmoe cOvapcoHpo
oocmec @080: n o mzw 0000 mm: xcH:# 50% 00 common awn: vcmpmma
Qm>wc u N
mm :mpmo mpozpoWOH muoowoaa
was moEHpmfiom u H CO xaoz pm>m hHHEmm poo» moon AA
wfimeononov ooHo mum>Hpm ow 0m «mm>HpmHoa
no 0am mocoHpm yHmH> .Amo 0H 000090 00H“
Hm .roch OH #50 cm .mmew mmHa .wcHaEmo
.050: um chwmmpam nwchaonw .on>oe
.moEmw HHmn "ponummoy mwcng wcHBOHHOM
mm moHpH>Hvom OH I Ho may m0 how on 90>0 hHHEmm 930% moon Am
Axmasz
-- Aommcpmspomoo smog zoqmm mzmeH mmmzmmzemmoe
0w 30HV HH 1 Ho 02H are mom mm¢oom oszmDm wm nmxmazH quz<m
.090 .mconoom
“#cofizmwcsm u H .
woo cowpwcHgEoo u m
coHumcmHaxo woo mpcogmo 030% on pas: n0H8.
ow copmHH n m 0%.:oz Cozy popmH.flH oeoc 50% MH A0
memengrHH 0mmwxlls ....... EoyH mHAmem> mgoaopmo

oflcsfigoch

oHanLm>

ooscwwcou::wm 04009

 

mm
mm
mm
mm
OCH
OCH
OCH
mm
mm
hm
mm

mm

o>oom mm
o>onm mm
9500 mm
m>oom mm
m>oom mm
o>onm mm
m>oom mm
m>onm mm
m>oom mm
m>oom mm
9000 mm
9m>ma
moEHumeom
cmumo

05mm mm30©5H3 may :mwz AH
050m «900 may 2mm: Ax
memo mmHmmE x000 Am
05mm mx903 090% 00 AH
0500 wmwnpOHo :30 co9H A:
05mm «mongHo w.%HHEmm 009H Am
memo mmmngHo :30 cwwz Am
memo mmonpoHo m.%HHEmm 5mm: A0
memo mwm5on mHosz :mmHo mHms Au
mama «6009 530 cmoHo A0
0500 mcm9wHHno 9owc50% Mo 09mo 0x0» an

n o

u H

u a mooom 90% monm Am

0 o o oOHv 50% CC. PM£3 «0.0.. “H

 

 

 

@050; pm X9030m5oc 0:9 59H: #50 5H0; 50% on Axmasz
M moHyHHHnHmcoam¢9
x. --Am09999909mcogmm9 nmflg ”040m mEmHH zmmemaom mac:
09 2090 mm - 00 9:9 moo mmmoom oszzam 9m omxmazH m.u:mccoammm
w5H%mm
mm 0>opm mm 050m m50% ow pH mm>Hw on: .50% 09 mocmonHm
co>Hw mH %050E wchcomm 950% mo 0500 MH "N onmmm>
u wc99mm 9995
wmm 05000 00 000m m%HHEmm 950% CH mHHHA on“ m%mm on: ”H onmmm>
. :0HpmcH80o
9509mm«
muflfl So I. 208 vii :. 58,9 3 0m 28> fiom 38
to: mammou mHQmH9m>
cm©5m9owcH

UQSCszoozzbm oHan

 

50H90>0 05 u o
H 50 u009w0 u H
m 50 0009w0 u N 030:0
m 50 0009w0 u m 0%H90H5wm9 50903 050909050 90 50909>0H09 50
H0 050990055 3900 90% 0950905 950% 00 030:0 9509 9053 003050 9505009w0
0050: 030:0 : 0500 n : 50909>0H09 09990>0m 9509 950% 090 90:: 0H9£0|9509050
mm 0>000 00 0500 000509505H 59H00:
0>0c %H950m 950% 5H 000%50 0005 A0
00 0>000 00 0500 wmcwonHom 0:9 mo %:0 90m %0505 030
05090905m 90 0950905 950% #5959 50% 05 no
5
”m 00599 50 00H905 0
90>o 09505%05 %H:9505 m0 009900 0 :9
0%05 50:9 050 000509H550 90 095995959
mm 0>090 00 0500 0%59 %H950w 950% MH zocx 50% 05 AA
mm 903050 50 0>0m u o 050H0095 %0505 95000
:20:x 9.500: 0900 n H %9903 0950905 950% x5959 50% 05 A0 Axmosz
509909050 %H950m
5000092000 30000 2099 0000 me 000092009
-- :09 09 50050 3 u 0 0:9 000 000000 0292000 90 0090029 90005000000
am 0:090 0 0500 0:050 H0>0£0 A:
$00 0>on0 00 0500 . 5000090m 050 0:09 A5
memewnuwH ., 0000 x 509H 0H00H90> %9om0900
1.3; 552000 waflmwflm>
ow70m9095H

’lli'i

50:599mooszm 0H205

 

 

 

 

mm 05 u H «5993 055090 w505
00% u m 50% 0050999 90 50555 0 0>05 50% 05 A0
9x00290
:3 90005050990w09m 309mm 02099 09005
0090 09 3090 0 - 0 00009 009 000 000000 0292200 90 0090029 “0 209000>
00050 900: .00% 99 909505
H00500 90H5m09 0090950 59 09059099905
050 09 m5OH05 50% 50953 0509900950w9o
9o 055Ho H00500 09059 09< 59909Hx0050v
90590 .0H99o 3055905 .090950505 0.500
.05509w 5950% 0509m9H0m .55Ho .0%0m
90005050990009m 5095 .095000 %om n<oz% fizz: .095000 H99w
09 20Hv 0509900950w90 .909500 5950% .<oz% “0509900950w90 0509900950w90
00 900900090 9 - 9 009309900 029 0o 9:0 09 000900.009 00 n9 209000>
ho 0005050990w09w
”m 9500505005
90>05 u H «0050: 99059 59 00>990H09
mm 005990500 u m 050 0050999 9909>. . .m5930HHom
50990 n m 059 90 %50 00 90>0 %H9509 950% 0005 AA
QCOS u H
mm 309 %90> u N 000999>9900
0500 u 0 %99555500 90 90H 0 59 09059099905
90H 0 u 0 050909050 90 0950905 950% 05 90
9000290
90005050990w09m 205mm mzmeH 0005050990W09w 509909w0959
.. 0090 09 2090 9 - 9 009 029 000 000000 0290200 90 0000029 900900 299005000
00999995099
'09 059000 0000 : 5099 0950990> %9000900

000509095H

090a9am>

’tl‘! III-"l

0055995001105 0H909

o 050 0 0%99555500 0959 59 00>99

50% 0>05 0509 305 05905 50% 0903 0905:

00095095 50 cm Aw
000959009 099555 09 0m 99
0099050 909090050 09 0m 90
0905590 09 950 om 90
009900500 09055 09 0m 90
00%095 00090 09 OM A5

0059>990 950 cm A0

. . .m59309909
059 mo %50 00 90>0 %99509 950% 0005

"305mm mEMBH zm>mm
mmb mom mmmoom UZHEEDm %m memmzH

000505 .0050999 950%
90 9509590>0 %090 90>0 50% 05 90

000505
.0050999 950% 90 9909> 90>0 50% 05 A5

.5.

 

900% 9 5059 0009 0 oo

.0900% m 5059 0009 u 90

mm 09 5059 0009 u 09

.09% 09 5059 0905 u 09

0905 5905 u :9

0m 05 n 9

wmh u N

00 0>050 00 0500

mm 0>050 00 0500

mm 0>050 00 0500

mm 0>050 00 0500

mm 0>050 00 0500
fim>0fi u H

W 00 005990500 u N
10 fimPMO u m
950990950990

:1 0505 0090950 5&95

09 2090 09 1 00

am 0>050 00 0500
90>05 u 9

who 005990500 u m
309%0 u m

.09999990099 0000

:05 559000

0000950909

!l| 3', '. .| .v 0 ‘1'!

%99555500 59
0599 90 090000

AxmazH
50990950990
0505 0090950

 

0099

0950990> %9000900
0950990>

 

 

0055995oozl0m 09500

 

 

«Anemmmm Mo macaw coymy>mawy may
my vagmyompmzo 0H Mo ymwav "mymmyoyypmm
maoyow no mpwyowamzo mayzoyaom mLy guys:

 

mamyowpmno >9

 

mm yomanoo wmma no m u a Cy msocm >e mzy mo wmemc any 09m ymn: “H onmmm>
yoonaoo Ham u m
mam>ma mmwmaxoax
ycmwcomwom
mm vmem: wmwnaoz o I o wymsz .mmm mH wwmynnos mam mmynno: ycmpmm
m>mn mcmwmmmsw mo mycmamm 950% on mo hwnssz
.. mmwaaoc ycmwcommma coyymycmyno
mm cmEmc mmflnno; w a o wymnz .wmm mH wwmwppos %cw m>mz 50% on we LmnEyz mamm
«AyamEcam>ow uymyw go
a n a
‘ ycmEc9m>om mywo Emymzm Hoosow may .
.% nywz vmyowccoo wcwmn mm >MHycmow oy mmpsmym owansm
1. wmxmw mum: mycwvcommmn soaps wmaswyw mymym cam Hmooa
:m mayomaaoo wowwwycwvw HmUOA wcm mymym mafia mo ymwaxomno mo mwvmazocx
amaswmm m I o mv “onmQ vamya mamomm msy 09m 0:: ycmvcommmm
am on u a wagonm m mm %Hamaswmm mamey no
mum u m cognac ncoyym mawemm 9:0» mmom an
«8.5: u H mcogno
who mmeflymaom u m 0y om...a@£ymmoy mcmzoyaoy may
ccymo u m Mo mam ow Qo>o mawamm 930% mwoa Aw Axmasz
. 509330
:. Amocmocmppm noqum mzmeH pm moqmucoupm
guy; ow zeav m I m 03y m:y mop mmmoow wgflmusm ym amxmmzH mo mocmsvmpm
woymmwygwfifi mvoo -:;a--:ll.r::l-a EoyH mynmypa> mnomaymo
as; mcflvoo maamflaw>
mmcswgong

 

C 0384.8;

mou-unm magma

 

 

 

ages yon mmov u y «x90: 50% ow yamemoymsm
mm myamyym>m cmaz u m cmyyo 30: wow 50% ow x90: yo wayx ymaz man mvymyso
zyywmmym mayo: u m .mmm my maOm mEyyuyamQ m m>ma 30% on m.y:mvcommmm«
.mco you mm mammm may now ywmgmycy
aose mm mmEyy mmm oy m>ma cane:
mam .mammh 03y yo vcm may yyyab
ooy m>oam mm mEmm omm may xmma mmm yon mysoo mama My Am
mm m>oam mm mewm . my mwmmao mommny may Aw
mm .m>oam mm mEmm w my mymg ymmamycy may Am
a my
ooy m>oam mm mamm mymyQnym may .m>oam COyymmsw may CH Aa
9
m . my 0mm mo
w: .zmcoE may mo mms may now 9mm»
9mm w: myoa: Ama mmm omym mam .pmmm
mco mo cum may ym 0mm may %mmmn
pmzmcw ymeLOUCy u o oy wmmywm mam .myons pma 509m 0mm
mm Amncam ymmymom u y zopgoa oy Uma mama .myomoya m man 09 Am
Axmazyv
i - 3.30m 5mg . 23mm m5: gyms
amya oy :oyv m u o m>ym mzy mom mmxoom ozyzzsm wm amxmazH "m onmxm>
mamEmyy acmya may cynyyyy zmyv ycmmmpmmn
mcmuoym mcyzoyyoy may ow myoswopm mcmwoym cm
was yomgyoo oy I oo yma: wygmmxm maymyygm>wm cm 50% ma< ”m ZOHmmm>
gmumflwmmemmwsttvst mwoo Ii!:t:i2s- ,, Emyy myamypm> myommywo
1: v3.2.4.5 3%..22,
vw;:am3nrtaH

 

vmscyyrooz:>m myamb

mnymyyxmmaov wycmeyymymcy %amE ooy
.mmcoe amss ooy mmymm3.nmayoe .hmcoe
amas ooy mmymm: hmaymm «mmayoym mom
hmcoe awsonm yo: .awya ooy m.ycm& may
.myyya poymow same ooy .ymymcmw Cy
myyya zsme ooy .vooy mom mmcos awsocm
yo: "mmyapoz mmcoE ymmmwya nymay mam

 

 

xcyay 30> om ymaz memyaonm hmcoe myamamm
3m mmxmmao mmyagoz m n o ysoam mane: myampmm 950% anyay so» on "y onmmm>
mmyaaoz
hmcoE coyymycmyno
“yo pmaezz amasmcoo
figwfi ..u o
xmm: m mEyy y sway mmmy u y mnmaayy
nu xmm: m mEyy y u m :y mmycsym
mu xmms m mmfiyy mymsom u m ycmwcommmfi
mm mmmmum>m u : «mummayy may :y zosym pm>m so» on mmamswmam
m\a oy a: u o
y my a: u y
N\H a op a: u m
N oy as u m
N\H m op a: u a
m my a: u m
m\a m op as u m mmymsym
9 my as n b mmvsym yamwaommmn
a>a .mga : away mmoE u m 90% om %mw m mmsoa mums 30a ysoa< magma mo nmasnz
amyyyyyamyy mmoo Emyy myamypm> muowmymm
um; mcycoo myamypw>
muwzmmm.:H

mossyyammtamm myazb

meyy mo coynma

 

cma u y m pm>o mycmemmm %yaycos yo mmyamm M
5m mommamo u w my wwcyay pom maymmm :1 myomEyymym yyompm
ooow u m .:y :y mwcyay msa oy mmoy cma wamsoy
m no mmmy ooow m m.yy xcyay 90% on mosyyyy<
nm>mc u o
om>mc ywoeym u y mmEyy mo ooymmm m
mm mmEyymEow u m 9m>o wycmsmmm hyaycoe :y mmo cmay mam yywmhm
%yymSm: u m mcyayoyo man 9m>m zyyemy 990% mmon yo man hyyemm -
:m m>oam mm memm myy ayy: ow oy yam: ymos so» wysoz ymaz ycmwcommma
.30: yawyy omm 50> m>mm maOmEom My "m onmmm>
mm Ham mamam u a wyy gpyz om m.>mny
, mm mpom 0cm macaw n m xcyay no» om ymaz .oooym mxyy .hmaoe mycmpma
yam m>mm u m mo yoy m mycmpmm 990m m>mw maomeom My "y onmmm>
mcoyymyamyao
m>mmuoommmm
mfiymyyxmmamv mmyaaoa
amopomma .mxmmcm ammyymmzm yooamm
.yzo mayom ammamazy .wcywcmmm
.mmayoyo .myaogm oom mmnofi amwocm yo: ycmvcommmfi
mwm wmxmmao mmyypo: m I o ”mEmyaoym %mCOE ywmp g:0% mam ymaz "N onmmm>
mayyyyaimyy mwoo EmyH myawygm> anowmymo

‘.w hzanvou

‘.;ynomay

QHQUHnHC>

silt
lvl‘Ii'I-‘OO-

«mamyomo mcymaoam
:y no .mooaaoaawyma 930% my
.czoyczow mcyayoyo mzsa hyymzm:

 

 

mm 96%. mm memm mnymny 990m mamas mmgoym may 92 Aa
mmpmycmo
mm mmpm y vmyym u y wcymmoam :y yo wmpoyw czoycsoo mya
1* mm mmmpm m vmyyo u a may :y no mmpoym wooapoaawymc :y
mmmpm m omyyo u m 00am 0y ammmom hyyemm Loom mmoa Am Axmmzyv
coyymycmyao
2033520 2&3 $6.5m mzmS mcynaonm
.. 3 20220 a I m 2; 5:. mom mmmomm mzHEsm E nmxmmzy >35.“
mfiymyyxmmaOV
pmayo .wyyyw .wymomm .mmyaaoa .myygmam
.mmyyomzw yooamw «msy>mm ayamECymymmycm
nmxmmcm ammyymanom ,moaomm .mmCmewma mayoammm
Ahymygm> amya 0y zoyv .mmypca nmmayoyo nmmaocsy .mymymamo yamwcommmh
wom vmxmmao m:myy my I oo "Loy >3c0£ pom» woman 30% 0o yma: mo hymyam>
cmyyyyyamyy mcoo amyy myamymm>
IQawiflmoo

nmxzmTHoych

Ill'l\
III!" -II

ameCwync

j

%Homoywo

CHQQMQQ>

“InuilINIm UNIDNVH.

.wyymymv mom Hyw ymymmao mmm .mmyymmym ymcoyymmsmmo mayvom 0y cm>y0>ay yamEmaSm
m>yyommaaw yo mmymmo pmymmyw may yo mmnmoma mmqma ycyom I y m ayayyz yamEmmymm 0y mmumymy mm: myymyynm
may .Emyy wyay co .mwoo mmymmym m 0y yamEmmywm mm: mao myay ymmmxm mEmyy yym co myymyypo myyyyamyymma

Amway yamoyoyyv .mymmm may :0 mym>my m30ypm> may ym ma0m may 0y a0m m.pmcpmm mwmz :ymE may yo mwyymmym may m0
aoyymmmyma yymay mayamyme .yamsmwSm m>yy0mwasm m mme mpmvoo .mymmm may wayuyyyy: cH .ymamym mumma cam

amma away mwyywmym mwmy u o mpoyomyyom mwmaymw .nmxmym magma 0cm :mma n y mnoymymao wqmywzma ammmnyymz n a
mgaympmao 00yymym m0y>nmm .mnoyomy myxmym 90y poymymmo ymyyoa n m maoyymyompom wmmyay 90m amsmmymm .mqua0m=
oysm u : mamammywoyoam ymmmmo .hammeom acyymyyommcmyy 90y xymyo mymy n m whammeom mmmyw 90m yoymEyymm ymoo
am0y>ymm maymyyam>om 90y nmwmame yyvmpo u o mymwmama yamyasooom ywoo m>yyoeoysm .pmagoz coyym>pmmcom yyom u h
mzcmmfioo myyyomym pom ycmmm maywmaogom .mcmmeoo yawn moyaoyymmym pom ymmcywam mmymm n m mamammmy yooamm

mamym .ymmaywam ymoyEmao wayyHSmcoo u m .moyoy gym 90m poymspymcy yymyoyym .yomaamyayymasm yooamm myyasa

no ycmymymmm I oy myommmmoym mwoyoymom .ymmzmy m yo mwyywmym may a yy mmmzzmy m away mwyymmna mace u my "mam
.m mmmom mymmm may .yamoyogy .o wcyyym> ha ammoym>mo moo my ammo mwyymmym ymaoyymmsooo m0 mpsmmme maym

 

 

 

Ammyymmym amoc :omymm mwyymmym

yma0yymasmmo awya myay mmoo yma: «mayo; comamm myay ymcoyymasmoo

n8 3 :03 fl .. 8 $8 $2: ~38: byemm as» my 23 3 205mm?

. msymym
myeoaommIOymom %ammpmoemm
amyyyyyamyy maoo Emyy myamyym> anommpmw
In; mayfioo myawdym

o szymyCH . >

I\r|‘
" ‘I-III III"

U zayyncoIIbm myam,

174

 

pmayo u H
mm mycmymm
ayoa ayy3 om>yy ma
amymoyway yamaaommmy u N wm>yy 50% 0a an2 ayyz mamy myyemm
x903.y0c mmoo pmayoe u N msymym
mm mEnyyqu u m wayoz %may 00 cmyyo ycmamonEm
mEnyyysy wxy03.pmayos u a 30m «mayo: myyemm 990m cy 0a: m.ymayoz
mm mfimgSma yayemm my I mo maymyaxompov o>yy so» om sag: gpyz mNym yyyemm
ymmyo no Nmmy u o Nomo ymmzm:
so ymymy p0 mmmy u y w.%yyemm 900% m0 ymmh may my ymaz A0
r2mm y 90 o u o wmyyemm
om mymo : 0y N u y 950% my m>ma :0» 0a mymm mama 30m A0
mmmoam N 0y y u o magma 30a .mmm mH
om mmcoaa m 0y m n y choammymy m m>ma myyemm maom mmoa AA
980.» m 0y m u o Nm>ma mmsoa no ycmEypmmm
msm meooy by 0y oy u y m.>yyemy ysoz mmoa meooy Name 30: Am
Axmazyv
Amsymym yogum mzmyy mach maoymmmmmoa
II amya 0y Boyv : I o mzb xom mumoom 02ynzam wm meszH “N onmmm>
cmyyyyyamyy 0000 I Emyy myamyam> muowmymm
Ima maywoo myamypm>

mmvsmymyay

WMMMWyCCUIINN myflmfi

175

 

prm u a
coy 50a u N Nyyym m 90 moa m 50% my< xmm
ayyao
%yco my yamaaommmy u y
ayyam ymmvyo u N
Uyyam mymaye u m
oyyam ymmmazoz wmmmm yymay mum yma: 0cm m>ma 30% 0o
wmm my yammcommmy u : mymymym mam mamayoya home 30: ymoyo ayyym
mmyyyyyamyy mooo EmyH myamyym> mhowmymo
Ima maycou myamyym>

mums mymyay

l‘IiII‘I

ammayy200II>N myamy

APPENDIX B

INDEXING PROCEDURES

...

'4‘

 

 

 

IXDSXING PROCEDURES

 

From the original Wisconsin questionnaire of 212 items, 90 were
originally tapped for the purposes of the present study. These 90
items were coded into 145 different initial variables potentially use-
ful to test the hypotheses outlined in chapter II. An a priori
decision was made to group items into indexes where such a procedure
was warranted. Of the 62 final variables in this study, 13 were
indexes reated by summing one or more of the original items. The
indexing procedure generally involved two steps: 1) first a logical
decision was made on what items seemed to be tapping the same behaviors;
and 2) the logical decision was tested by computing Pearson product

9

men at correlation matrixes for each set of items that might form an
index. Zechs, standard deviations and code ranges of the original
items were also examined so that constructed indexes would be based
on items having roughly the same variance and contributing roughly
the same weight to the final index.

The correlation matrixes for this index construction Operation
were computed by deleting all respondents with non-responses on one
or both of the items involved in each individual correlation. To
establish a criterion for significance a conservative baseline n of
200 was selected to compensate for the varying n's for items. .Actual
n's for the items used in indexes varied between 229 and 252. At an
n of 200, the critical value of r at p1(.05, using the r to z trans-
f rmation test of significance, is .14 (McNemar 1962). All tables of
correlation matrixes which follow use this criterion for significance.

Aipendix C reports the code ranges, means, standard deviations, and non-

ressonses counts for all final indexes.

177

 

the five items are listed in Appendix A. Table 28 below reports the
original code ranges, means, standard deviations, and non-response
counts for the five items. Table 29 reports the Pearson product moment

correlations between the five items.

 

Table 28: Code Ranges, Means, Standard Deviations, and Non-response
Counts For The Five Items Used To Create Two Indexes Of
Permissiveness

 

Item Code range Meana s.d. Non-responses

 

 

Hsa-s on reetdays 0 - 8 “.18 2.09 23
hours on wdehends . o - 8 n.94 ‘ 2.28 22
Tell where are after

school 1 - 2 1.16 .37 0
Tell Iherc are in

evening 1 - 2 1.06 .23 1
Tell where are on

recknds 1 - 2 1.13 .33 l
a For all five items, the means, medians, and modes were equal.

 

Table 29: Matrix Of Pearson Product Moment Correlations Between
The Five Items Used To Create Two Indexes 0f Permissiveness

 

(1) (2) (3) (u) (5)

(1) Hours on weekdays -

(2) Hours on weekends .53 —

(3) Tell where are after school .11 .1u -

(4) Tell where are in evening .09 .06 .27 -

(5) Tell where are on weekends .11‘ .09 .MS .32 -

 

Since the correlations for the five items clustered into two major

groups, two separate indexes of parental permissiveness were computed.

A

179

respondent hours -- was created by summing

original scores for items 1 and 2 with any respondent with a non-

rl'
.’ J"
(D

reasonse on one or both items being recorded as a non-response on the
new index. The second index -- parent knowledge of respondent whereabouts --

1'53.

U)

created by summing the original scores for items.3, 4,.and 5 with
any respondent with a non-response on two or more of the items being
ecorded as a non-response on the new index. A respondent with a non—

response on only one item had the original mean for that item plugged

in for the computation of his final index score.

'ee items.in the original questionnaire seemed to tap various

of child-parent communication. The original wording and codes
‘ .

1;
(J
(A
0)

f1}
()
’(1

car the three items are listed in Appendix A. Table 30 below reports
133 original code ranges, means, standard deviations, and non-response

counts for the three items. Table 31 reports the Pearson product moment

correlations between the three items.

 

Table 30: Code Ranges, Means, Standard Deviations, And Non-response
Counts For The Three Items Used To Create An Index Of

Child-Parent Communication

 

 

Item Code Range Meana s.d. ' Non-responses
Talk to parents on problems 1 - 3 2.57 .60 0
Parents ask R.opinion 1 - 3 2.54 .67 0
Parent reaction if R comes

in late 1 - 3 2.10 .82 6

 

2? all three items, the means and medians were equal but the
n Igginal distributions were skewed with a discrepant mode.

 

’4
C’)
O

 

. son Product Moment Correlations Between The
ems Used To Create .n Index of Child-Parent

 

(1) (2) (3)
(1) Talk to parents on problems -
(2) Parents ask R opinion .29 -
(3) Parent reaction if R comes in late .22 .13 -

 

To create the child-parent interaction index, scores on the three
items were summed. Any respondent with a non-response on only one of
the th.ee items had the rounded off mean value for that item plugged
into his final index scores. Respondents with non-responses on two

or more of the original items received a non-response on the final index.

9

Ischerness index

 

no itdms in the original questionnaire seemed to tap the degree

of family group activity or family togetherness. The original wording
and codes for the items are listed in Appendix A. Table 32 below reports
the original code ranges, means, standard deviations, and non-response
counts for the items. The Pearson product moment correlation between

the two items was.43.

 

Table 32: Code Ranges, Means, Standard Deviations, and Non-response
Counts For The Two Items Used To Create An Index Of Family
Togetherness

 

Item Code ranges Meana s.d. Non-responses

 

9r]

-anily participation in
0 activities 01 - 10 6.1% 1.83
ily works on group projects 1 - 3 1.92 .68 5

(D

1

a.
'1'"
4....1

”'1

 

a For both items, the means, medians and modes were equal.

can which activities his family did together. The second item came
.etimes-often” response to whether his family worked
.er. In order not to give the work on projects

on undue eight in the final index, the item was converted to a
”yes-n0" basis by recoding an original response of "sometimes" or
”often" as "yes" and (code 1) and an original response of "never"

as "r 0" (code 0). The original scores on item 1 and the recoded
scores on item 2 were then summed to create the.family togetherness
index. Any respondents with a non-response on one or both items was

recorded as a non-response on the new index.

3 hone respon nsibilities index

Fourteen items coded from the original questionnaire asked the
raspondent the frequency with which he handled various home"
responsibilities such as housekeeping and yard care.. The original
wording and codes for the items are listed in Appendix A. Table 33
below reports the original code ranges, means, standard deviations,
*nd non-response counts for the 18 items. Table 3a reports the.‘

correlation matrix for the set of items.

 

 

 

chile 33~ Code Ranges. Means, Standard Deviations, And Non-Response
Counts For Tne 14 Items Used To Create And Index Of
Tespoident tone Responsibilities.
Itena Code range Meana s.d. Non-responses
(1) hop for Food 1 - 3 1.06 .55 0
(2) Take care younger children " .96 .88 l
(3) Clean own room " 1.72 .u8 0
(a) Help clean whole house ” 1.19 .69 u
(5) Wash family's clothes " .36 .56 u
(6) Wash own clothes " .50 .65 3
(7) Iron family's clothes " .su .64 u
(8) Iron own clothes " .88 .78 3
(a) Do yard work " Lush .70 1
(10) Cook meals " .88 .59 3
(11) Wash the car " 1.07 .67 l
(12) Wash the windows " .89 .68 2
{13) Take out garbage " 1.38b .75 1
'5+) Shovel snow " 1.60 ‘ .64 1

A.

DI

All but three of these items had equal means, medians, and modes.. The
rest narktd with a subscript "b") had roughly equal means and medians
but disc epant modes.

Table 34: Matrix Of Pearson Product Moment Correlations Between The 14
Items Used To Create An Index Of Respondent Home Responsibilities

 

(1) (2) (3) (9) (5) (6) (7) (8) (9) (10) (ll) (12) (13) (14)

3 D

\_,»z\_~lvaV\-’vv
.
N
\1

r

.07 .07

.27 -.02 .39 -

-.04 .21 .34 - '

.19 .05 .18 .28 .73 -

.13 .08 .23 .38 .H6 .H4 -

.18 .27 .33 .36 .53 .69 -

.00 —.01 —.19 -.11 -.14 -.23 -.24 -

.12 .21 .23 .30 .28 .21 .21 .02 —

.02 .15 .01 .11 .06 -.10 -.O7 .86‘ .22 -

.13 .26 .17 .20 .20 .12 .16 .23 .34 .95 -

.03 .08 .05 —.02 -.08 -.17 -.18 .31 -.O6 .15 .23 -
.03 .02 —.18 -.08 -.06 -.26 -.22 .66 .01 .38 .22 .35 -

AI—‘Ar‘ofif\ P~
.\

(... \J L! (.11 42‘ (U

.1

( I")

I
I00!“

Hcocncocowi-

C , .5 /‘\ 7' “".

3
.\ J I —~
\/ V'

’...

\'.J ;/

o c o 0
O (”J 1"" C)

I

183

(I
{a
E)
C
t
('
0
C)
'O
I“
:5
..Q

'-g-type tasks and one set of highly interrelated

'ardwork and heavier-type tasks. Approximately seven items fit into

seven into the second. SinCe it seems logical

that girls w

ould score more on the first type of task and

:oys more on the second, it was decided not to separate the 14 items
Lnto two separat

e indexes but to sum across all 14 items. Any respondent

dth non-reSponses on-three or less of the original items had the

fi‘.‘ ~-
. x. at;

ded off mean values for those items added to his final index

:-;.-. Tezpcndents with four or more non-responses were recorded as
- -reoponses on the final index.

’. o 0
efficndent knowlecce of family Operation

 

Eight items in the original questionnaire tapped the respondent's

about his family's household Operation. A typical item

Zed ”do you know how your home is heated?" and was coded simply as

5.- uJ answer" or "don't know". For the four items that were used

Ithis index, the original wording and codes are listed in Appendix A.
tie 35 below reports the original code ranges, means, standard

=fiations, and non-response counts.‘ Table 36 reports the correlation

a. 3,
'2-‘39K o

 

 

 

 

-‘:le as. God: Ran,rs, Loans, Standard Deviations, and Non-response
Cosnts For Tze Four Items Used To Create An Index Of
Respcntcnt \LOLlcdge of Family Operation

2:3; Code range Meana s.d. Non-responses

Pa-e:t no7c; worries O - l .40 .ug 5

it:il' use of credit 0 - 1 .35 , .48 0

P:r-nt debts O - 1 .33 .u7 17

Fanily health insurance 0 - 1 .35 .u7 2

e For all four lLCWS the means medians and modes were equal.

 

Table 36: Matrix Of Pearson Product Moment Correlations Between The
Four Items Used To Create An Index Of Respondent Knowledge
Of Family Operation

 

(1) (2) (3) (4)

1) Pzrert men 3 wcrries -

;~} Panil; use of cr-edit .17 -

{1) Parent d3-ts .18 .29 -

;;) Family health insurance .17 .28 .37 -

To compute the index of respondent knowledge of family operation,
;;;res on the four items in table 35 were summed. Any respondent with

non-reSponse on only one of the original items had the rounded off

{U

:ean value for that item added to his final index score. Any respondent
Vith more than one non—response was recorded as a non-response on the

final index.

185

 

-tens in the originrl questionnaire tapped aspects of

' 3
I

no original wording and codes for these

firms are lisced in r~nCnalx A. able 37 below reports the original

are ranges, means, standard deViations, and non—response counts for

flmse items. The two items were correlated.17.

 

fible 37: Code Ranges, Means, Standard Deviations, And Non-response
Counts For The Two Items Used To Create An Index Of
Parent Gregariousness

 

 

lien Code range Meana s.d. Non-responses
Participation in com nunity.

activities . 1 - 4 2.57 .87 6
1E-iti:g friend: and '

relatives 1 - 2 2.52 .55 3

r.)

or both items, means, medians, and modes were equal.

To create this index, scores on the two items were summed with any
mayor de nt having a non-response on one or both of the Original items

:ecorded as a non-reSponse on the final index.

-xsendant cr-~*riousness index

Five items in the original questionnaire tapped various aspects
he respondent's gregariousness.' One of the original items 5- a
~£sure of whether the respondent felt he has a close friend -- showed

Lnxie variance and was deleted. The original wording and codes for

tfiirenaining four items is listed in Appendix A. Table 38 below

186
e;orts the code rances, means, standard deviations, and non-response

counts for the four items. Table 39 below reports the correlation matrix.

 

 

 

 

 

 

T:ble 33. Code angcs, beans, Standard Dev ia ations, And Non-response
Counts For The Four Items Tapping rd Respondent Gregariousness
‘Item Code range Meana s.d. Non-responses
II ave a bunch of friends 1 - 2 1.69 .A6 3
Visit at friends ho...es 1 — 3 2.58 .53 5
Friendss stay at R home 1 - 3 2 O2 .63 3
Organizat io ons belong to 1 - 7 2.08 1.37 7
a . .
For all four items, means, medians, and modes were equal
Ta.le 30: Hatrix Of Pearson Product Moment Correlations Between The
Four Items Tapping Respondent Gregariousness
O
(1) (2) (3) (4)
(1) Have a bane h of friends -
(2) Visit at £1 iends homes .21 -
(3} Friends stay at R home .18 .42 -
(4) Organize tions belong to .28 .12 .1u -

 

ecause the item measuring the number of organizations to which
the respondent belong was not significantly related to one (and related
only barely to another) of the other three items tapping more informal
behaviors, the organization measure was excluded from this index.

However, it was left in the final analysis for this study and is

!.l
’1:

sted in Appendix A as-version 1 of the variable, respondent gregarious-
T}

.ess. Verson 2 of this variable is the index that resulted from sum-

Ding scores: on the three remaining items in Table 39. In computing

187

this index, any respondents with non-responses on only one of the

three crijinal items had the rounded off mean value for that item
added to his final index score. Any respondent with non-responses

1‘"

on two or more of the items was recorded as a non-responses on the

Outside Home orient tion index

{J

Bight items drawn from the original questionnaire tapped the
degree to which the respondent's family uses outside home resources

elaxation, enterta nment, and education. One of the original

—- frequency of attendance at taverns -- lacked variance and

|
l
d

SlutCCi. The seven remaining items are listed with their original
‘~fitga and odes in Appendix A. Table 40 below reports the.
coin ranfes, means, standard deviations, and non—response counts for

f'l ‘

the seven items. Table 41 reports the.correlation matrix.

 

Table 40: Code Ranges, Means, Standard Deviations, and Non-response
Counts For The Seven Items Used To Create The Outside Home
Orientation Index.

Ltcm Code ranges Meana ‘s.d. Non—responses

GO .ut driving 1 - 3 2.13" .63 4
Go to plays 1 - 3 1.69 .61 5
CF to concerts 1 - 3 1.64 . .64 2
&>out to dinner 1 - 3 2.24 .57 2
{b to sport events 1 - 3 2.06 .68 6
‘25 .‘CO lectures 1 " 3 1.41 .57 7
CC. 0.1". picnics 1 - 2 1.66 .47 8

 

 

{U

For all seven items, means, medians, and modes were equal.

 

 

 

V
Table, : :. :1 : natrix Cf Pea so; Product E-Zomen“ Correlations For The Seven
Items Used To Create The Outside Home Orientation Index
\
(1) (2) (3) (4) (5) (6) (7)
(1) «3:; out drivirg -
(2) (3:; 1:5 plays .05 --
(3) -3 1:0 concerts .04 .46 -
(4%) Go out to dinner .23 .27 .17 -
(5) Go to sport events .18 .12 .05 .29 -
(6) Go -0 lectures .08 .37 .51 .16 .08 -
(7) so or: picnics .31 .02 .20 .19 .14 .14 -

 

 

To create the outside home orientation index, (scores on the seven
.4’3w‘ o ‘
lLL...S teem-re summed. ny respondent with non-responses on one or two of
3 had the rounded off mean values for those items added to his
fine? ifi ~ i -
- 1-c1 x score. Any respondents with three or more non—responses
C 22;".

ed as a non-response on the final index.

(5:

~~\;‘:'~“ _ci-‘ attendance at church index

ms drawn from the original questionnaire tapped the frequency
Q ~fi, o o I Q

-L cam ily's attendance at church. The original wordings and codes

for 1*: - e
7e *‘C ems are listed in Appendix A. Table 42 below reports the

COde I“: “ _ ,
‘ ‘Nz; Q S , means, standard deviations, and non-response counts for the
3:130:13. ,.___‘ ‘-
-L me two items were correlated .53.

 

= Code Ranges, Means, Standard Deviations, And Non-response
Counts For The Two Items Used To Create The Frequency Of

\\\Church Attendance Index
Item

 

Code range Meana s.d. Non-response

 

trequm Q
Setha by attend church to-
Klbch regularly 1 - 2 1.64 .49 8

$1:
/

 

boath items, means, medians, and modes were equal.

To create this index, scores on the two items were summed with

respcndents ::ith non- esponses on one or both it recorded as non-

 

Rine items drawn from the original questionnaire required the
respondent to use his knowledge of'math. Two of these items were
deleted because the items were too difficult and variance was low. Two
additional items were deleted because they did not relate significantly
either to each other or the remaining five items. The wording and
codes for the remaining five items are listed in Appendix A. Table 43

below lists the code ranges, means, standard deviations, and non-

response counts’for these items. Table 44 reports the correlation matrix.

Table 43: Code Ranges, Means, Standard Deviations, And Non-response
Counts For The Five Items Used To Create The Math
Knowledge Level Index.

 

 

Item Code range Meana s.d. Non-response
Percentage problem 0 - 1 .30 .46 12
Principle problem 0 - 1 .28 .45 12
Interest rate problem 0 - 1 .43 .49 12
Finance charge problem 0 - 1 .10 .31 12
Late payment problem 0 - 1 .71 .46 18

 

For all five items, means, medians, and modes were equal.

’J
LC)
0

 

‘sen Product Moment Correlations Between The
iv: Items Used To Create The Math Knowledge Level Index

 

(1) (2) (3) (4) (5)

(1) Percentage problem -

(2) Principle problem .26 -

(3) Interest rate problem .23 .61 -

(4) Finance charge problem .50 .48 .37 -

(5) Late ,ayment problem .26 .14 .17 .16 -

 

To create the math index, scores on the five items were summed.
Any respondent with a non-response on any of the five items was recorded
as a non-response on the final index. Mean values were not filled
in for any items on this index because the original coding operation
eliminated many "blank" answers which ordinarily would have been coded

-l

as no —respondes. Thus, in the original coding, coders were instructed

r1
(J
rt
"5
(D
(a
1+
{u
(.1

'i-correct" any blank answers fer respondents who had

obviously asttempted to complete the entire math section.

Family snooping orientation index

Two items drawn from the original questionnaire tapped the degree

0

to which the family utilized a greater number of shOpping areas in the

wankee area. Original wording and codes for the items are listed

in Appendix A. Table 45 below lists the code ranges, means,'

scandard deviations, and non—response counts for the items. The

correlation between the two items was .24.

191

 

 

 

Table 45 Code Ranges, leans, Scandere Deviations, And Non-response
Cour‘" For The Two Items been To Create The Family ShOpping
Cl‘ien Lu.ti0og Emile...

Item _ Code range Meana s.d. Non-response

General shopping preference 1 - 3 1.76 91 6

Preference for clothin ng stores 1 - 3 1.73 71 2

 

For both items, means, medians, and modes were equal.

To create this index, scores on the two items were summed with any
res spo ondents having non-responses on one or both. items recorded as non-

resvcns cs on the final index.

Socio-econcmic status: possession index

 

-4
o)‘
P.
H
(I
rr ‘3'

he original survey asked respondents to estimate their
family income, few of the respondents were able to do so. Thus,

this index was created in an attempt to tap a more ecologically and
economically based measure of socio-economic status than is provided
by measures of occupational status (although a measure of occupational
status is part of the final analysis in the present.study). Six items
from the original survey asked respondents for indications of the

size of their homes, age of their newest car, and so on. Two of the'
original measures (kind of dwelling and frequency of maid service)
were deleted because of the lack of variance. The wordings and

codes for the remaining feur items are listed in Appendix A. Table

46 below repOrts the code ranges, means, standard deviations, and
non-response counts for these four items. Table 47 reports the

q

correla tion matrix.

 

 

 

Table A5 Code Ranges, Means, And Standard Deviations For The Four
Items Used To Create The Socio—Economic Status Index Of
Possessions

Item Code range Meana s.d. Non—response

# of rooms in household 03 - 17 9.17 2.57 1

# of phones in household 00 - 08 2.29 1.27 n

# cars in family 00 - 04 1.50 1.65 0

age of newest car 00 - 12 8.83 2.80 8

 

 

 

a For all four items, means, medians, and modes were equal.

Table 47: Matrix Of Pearson Product Moment Correlations Between The
Four Items Used In Creating The Socio-economic Status Index
Of Possessions

(1) (2) (3) (u)

I

(1) 7 of rooms -

{2} of pho.es .47

(3) a of cars .25 .36 —

(9) age rewest car .27 .29 .46 -

 

Because of the varying code ranges of the four items involved,

each item was recoded to provide comparable ranges.

To accomplish this,

a median split was made on the four items and low values were recoded

to 0, high values to 1.

summed to compute the final index.

These recoded values of each item were then

Respondents with a non-response on

one item had the mean value for that item (recoded per the median split)

added to his final index score.

response were reCorded as non-responses on the final index.

Respondents with more than one non-

APPENDIX C

CODE RAISES, MEANS, STANDARD
DEVIATIONS, AND NON-RESPONSE CO TITS LISTED

BY VARIABLE CATEGORY AND VARIABLE NAME.

 

 

 

 

 

 

. .11 as cede IJTjCS, means, standard dev1ations, and non—response counts,
listed by variable category and variable name.
Code Number of
Variable category and name Range MeanC s.d. non-responses
Parent 7“elia use
Amount of parent television viewing 0- 3 2 0.8 0
Variety of parent te ev ision viewing 0- 8 3 1.5 2
Number of TV sets in household 0- 2 2a 0.5 3
Amount of parent radio listening 0- 3 2b 0.8 0
Variety of pa mr tradio use 0- 6 2 1.3 u
Number of ne wsprers subscribed to ---- --- --- ----
Version 1: weeklies 0- 5 l 0.7 0
Version 2: dailies 0- 4 l 0.7 O
Xu.ber of magazines parents read 0- 8 3 1.6 5
Pa rcnta l sources of news 1- 2 l .5 1
number of phonographs owned 0- 8 2 1.2 2
F"‘“*”“* “od‘a Use
.'snnt cf,t pendent radio use 0- H 2b 1.1 u
Variety of respondent radio use 0— 5 1a 0.9 6
Vaflvfy of res mp endent newspaper use 0- 8 5 1.7 l
_ “fer of magazines respondent reads 0- 8 2 1.6 8
Variety of respondent book preferences l-ll 4 1.9 l
Penjezi;:nt nreference for comics l- 2 l 0.5 l
Resp one nt library use 0- 2 l 0.6 5
Fresno: ey respondent movie going 0- u 2b 0.9 2
Variety respondent record preferences 0- 7 3 l.” l
Nadia respondent would miss most l- 2 2 0.5
R spo ondent media credibility ratings —--- --- --- ---—
Version 1: Most believed media 1- 2 2 0.5 2
Version 2: Least believed media l- 2 l 0.3 2
Family Cohesiveness
Parental permissiveness ---- --- --- ----
Index l: restriction on hours l-l6 9 3.5 33
Index 2: knowledge R whereabouts 3- 6 3 0.7 l
Child-parent communication index l- 7 6b 1.2 0
Fa: ily togetherness index l-ll 7 1.9 13
Parent orientation 0- 2 l 0.9 17
Parent domination ---- --- --- ----
Version 1: who pays bills 0- 2 l 0.5 3
Version 2: who pays allowance 0- 2 l 0.6 12
Reependent' 3 home responsibilities
index 0—26 1” u.o 3
Acapondcnt's kn nowledge of family
Operation index 0- H l 1.2 0
Iarent-child agreement on TV shows 0- 4 lb 0.9 4

 

dhile all variables are measured continuouslv. the annrenriate measure 0F

 

 

 

 

 

 

Code Number of
:ri:.1e cats o‘y and name Range Mean s.d. Non-responses
Co*men 1' '"“‘f2:cld‘
Parent gregarious.css index 2- 7 5b 1.1 9
Respondent gro~111 usncss --—- --- --— -—--
Vex sion 1: or; aniz ations 1- 7 2 1.3 7
Version 2: peer index 3- 8 6 1.2 3
Outs ideehome orientation index 7—19 13 2.3 4
Length of time in community 0-14 7b 4.6 13
Frequency of attendance at church index 2- 5 5a 1.0 13
Respondent knowledge local 8 state
figures 0- 8 4 1.9 2
Self Orientation
banner ofr espondent hobbies 0- 6 2b l.u 5
Nimber of pa arent hobbies 0- 6 lb 0.9 29
Respondent knowledge level 5 ---- --- --- ----
Version 1: TV characters 1- 2 l 0.5 2
Version 2: Ad slogans 0-10 7a 1.9 2
Veruibn 3: Math index 0- 5 2b 1.“ l8
Respondent's outside home employment 1- 3 2b 0.7 2
In so“ of hours respondent studies 0- 8 3 1.9 18
Freqtercz respondent studies in library 0- 4 l 1.1 2
3:1: “or Or_cntation
”ether or money worries ---- --- --- ----
Ver51on 1: parents 0- 5 1 0.9 7
Vcr31on 2: respondent 0— 8 l 1.5 u
Spend-save orientations ---- --- —-- ----
Version 1: parents 1- 3 2 0.9 7
Version 2: respondent 1- 3 3a 0.7 10
Family use of credit 0- 3 1b 1.1 8
Attitude toward credit 1- 3 2 0.6 u
Vari iety of respondent spending 0-15 7b 3.1 l
Pam y shopping orientation index 2- 6 3b 1.3 8
central tendency depends, of course, on the shape of the distribution of
r: pentes. Those “mean” values follo~ed by a subscript "a” actually represent
:h mole; and median value. Those "mean" values followed by a subscript

J represent the

”I
:3

mean and median value.
edians, and modes were the same.

For all other variables,

 

Code
Range

Mean

s.d.

Number of
Non-reSponses

 

 

lenders“ v

Socio-economic status
Version 1:
Version 2:

Family Size

Hother's employment status

F‘nily Type

Birth orcer

Sex

 

occupational prestige
possession index 0- u

24.2-81.4

Moon

50

9.8

Ol—‘OOOi—‘F‘
.p

HICHIGQN STQTE UNIV. LIBRQRIES

MW

102

 

Illllllill |||||l||||| lill‘llH ll
93 69

2 7574