' f 1.1.

 

.U.L 11‘ 'A' L I LJL' 1 1 . u 1'
. '3 3 ‘ "'3‘ WNW/"L '41' 1' ". L '.'I ..' . 11':' “i .11" '1'L"' "M'i‘
' .' _,"(1 a 'L' ':""1 in NW 'L'I ' 1'.': '1 1"1'\ "'"fi'n"
' ""' " ”'"' """'." "'L " "'1 L1" ‘0'" W'.I'.L'LLL"'LL' ."L‘ 'I " .."LL‘.L|~ " ' “Pill" L "L “'11.. ("'" "TALL";
'% ' L" 1 'LL'L‘LL L11; '3'3'. .11 .HI L
.g

  
     
 
  
 
  
      
 

  

. '1"
‘1’ $01.... I';IL}.'.3'1 a. .10:

1.311'1'1311‘ «1'4»

 
   

‘1':.:l

. -- “W" "77L"? '5
121' 1111133 1

11
.; 9'" I’LL;

  
  

2.27“ 1 'oIL'I'i',‘
'1' ‘, .

 
 

. .43
." f. x;

1- {b u

, '13-'1' fi'k 'EK , $1! 0‘...
1 3.3%. $.11 a...“ "‘3‘": A
1 . .. " " 1‘- .v «r

F 'P‘vkpb.. V6" ‘ . :

d. -

   

    
 
 
  
 
 
  

"'""L11"L‘ '1 L:

 
  
     
     

"1' V"
1111 1.; L
1 1"1
' {'1 '
3.
uéfimnfl
’flhr, Tl: '-1'" ' a. II... ‘ I. '1 b \I v',‘ -" \ 1 1 1' '.-_.-._' “"p1l'x'7" "'1‘” 5'0“. "'I:“F' "‘ ', will
“‘21 11ml. 1 ~ . I 1.1 "\AN1

“1' WW "'1'."
‘I:I’|& '0

. _ 1‘1 - -.e 1- '11 . . .
~-.1'1__ ' ' .‘.-.. '- ' ' 1 .11 ‘ -:1 11L" r" K""‘I‘I"““‘ "2'1 ‘Y.
. 1 n f.‘L"\'.'

Vb} ‘L'L'Jj. N‘L‘ J-LL LWH "

  
 
  
      
    

   
  
   

      
   
 

         
 

      

    
     

    

 

 

 

1' I a ‘ 1; “1“? "'l3 “\l “\L'r ' ' veg-W
. 1' .,;.|‘ ' Lu. | l‘.':;-_ i I 1 1 1‘
'1}! L; “'"'r'- '1 '--- "1. \";‘:o}"' LI "11‘ L'-‘ R“; L '
.. "\ ' ‘1".:1‘ ‘I'f‘fi.’ . III 1 1' 1.‘ -
ti}: .1 [$5M ""fip LLLLL". "'L1'L'\'"""1";' . '\;".' q...“ ‘\. \11:¢""I' o1.'3511'1'?1"3"m"" '.:""' :3
1“" ~1.1'1';'11‘.11‘j,l'l" "'1""'L":3‘"1'." """:L"" .:1; W'”{
' ' Wu ”' 1Q' ' 1 \‘1"*""r"1'1'";" 1 wk.
,‘i‘ l I‘ 5:?" 1.1"“: .‘1. 'fi}'fihh1\'l 1% “5" 3‘IJ1‘LR‘3'H' 'ngg; '6'” "Mg '1'"¢%;$: ”\‘fifi‘iww‘vutl ‘Lé'LL', UL"?
. .‘ 1 . 1 : ":I r '51.? a v . ‘ H V
U“ ‘ I . 4 ,v. ‘ ' I' "<"'11\‘- 11""."".}. L'" "1' "‘2 "" 1,1; ""1 NL'L' 1L" ' ' '
1'31" | I" ‘ LVN“ "\'""' "'IL "ht 4|'N;1‘11{'1.'::" 1""" -"" "m 'I'S" """1 "\ '
. 111-... .1 1 1 ‘ . - 1.1.1... 1.
nw'ik', r‘k " " """"" " "‘:" ":1" l '1' '1 I'l '1' '.‘ 5"" . ‘L:\'V' '9‘ - ' "
o .. .. 1:. 1:;:!- ;..:.;J _‘.. 1w ~ . :~-.-“-1-
'j 1...’ ' ' '1 1"..1‘~.:.r.'....:1-.".'~1';."?:;':'..'i' ;.1"1.:'r1~ 17‘ ‘ 1 . a? . .. ..
. .11 1‘.“ fiv' .."1:'u~.".“.'.1"f 311.3,”1'" -‘ '4 " .‘....1.1.. .
"'1" r' L!" "-'" ' M 1' . "L'L'I"~"'}CI"L 5-" "1 .19 "" . ,' 1‘ . _ $457":- -1 "91.11.? '
L"‘1"'"f.‘:[l "'"*L1';~M 111'. gawk...” '"LL L V." L 11 ‘. 4.. 1'1' ‘ , ‘-"~'1":j'1-11\ l“: '1,-!\' ‘ .1': 1.1,. ".'
‘ "'""1‘L" . .1 .. L'LU'LL'L'IM'L .L, t. 435‘ _ . LC' 1‘ ‘1; ‘ . 4;“, '4'1:;;..L" “111.1“."1‘1 .- .:L1'1 H 1
.' 1:1 1 ‘v '1". 1. 11‘. '. "" "“ ""
111-.” . "fL.,'.!13:-.. .111 1.111 1..- .1 .11, ~ . :1'-‘.1'1"-".L'. 1. 1L ' ‘ '
J§L\ " "' L?" "'- ' {@319}: “'N '7'" :2" ‘ ‘.'$ H 'h ‘ ' ’ 11". '. "WK“ '1 I '.'. .‘u'l'l' '."'11 ‘ 1 '
1'c' - 1.‘ 12'1'1‘12" .1'~:15L;.L. .- ~1-f+1.;:1.4‘41L114
.1 ,w .1 . ' "11'3" ‘ 1.‘
. '1

 

12-» 1L"
I‘Ifi-L "'11

o": '
""'" 5 '1 "Lfi'rflh' "..L'LIL-" .LW'.‘ L? : I11 L “11‘2" ”1 “l' L.’ ‘
'L'L" ML" Q‘LLLLJ' . '13”. L'1"I','i""1 L. .

’ ".'.L
.. 1W1 1W1

 

 

       
   
       

 

 

2..qu .11.

' '1éLkgr"-ik';LL"'gl

$111.11.; 115,33. #1.
1......1'LL ““11

.IL ""\"' :L'LLE'L 1"" L?"
‘A" l" .|:-'1"1'~' p'lé'
11.:-..1';'{:}§! 12.1? .':L;'

  
     
 
    
 

 
 

 

¢I'L\~I‘

J' v “"‘Hu.

  

1:1» 11‘"

'5'11‘ .L' I: I'
' @I'é'g'fl LP

THESIS

 

 

 

This is to certify that the

thesis entitled

MEDIA EXPOSURE AND COMMUNITY INTEGRATION
AS PREDICTORS OF POLITICAL ACTIVITY

presented by

Joey Reagan

has been accepted towards fulfillment
of the requirements for

Ph. D. Mass Media

degree in

 

 

Major professor

Datewwfl

0-7639

 

'" llllllllllllllllll"'

OVERDUE FINES:
25¢ per day per item

RETURNING LIBRARY MATERIALS:

Place in book return to remove
charge from circulation records

 

Wt}.
“£3.30! I. .4
, N H187
1.5 ,.. “if“:
l : W3?
r’flj
flJCW‘D-s
flj1‘3gfg‘
I00 A18?

PJNOZr

 

 

MEDIA EXPOSURE AND COMMUNITY INTEGRATION
AS PREDICTORS OF POLITICAL ACTIVITY

By

Joey Reagan

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Haas Media Ph.D. Program
College of Communication Arts and Sciences

1981

ABSTRACT

MEDIA EXPOSURE AND COMMUNITY INTEGRATION
AS PREDICTORS OF POLITICAL ACTIVITY

By

Joey Reagan

Mass media use is a good predictor of voting and major party
political participation. Previous research has ignored "quasi-mass"
media as well as political activity outside major party national
campaigns. The present research explores the impact of quasi-mass
media in a causal model predicting two types of political behavior:
voting and political participation. The model includes the simultaneous
effects of media use, community integration, education and length of
residence.

The hypotheses predict that media use and community integration are.
causes of political behavior. Mass media use is predicted to be a stronger
cause of voting while quasi-mass is expected to be the stronger predictor
of political participation and community integration.

Data collected in personal interviews in 17 United States cities are
subjected to a LISREL maximum likelihood analysis. The analysis explains
92 percent of the variance in voting and 59 percent of the variance in
political participation.

Mass media use is the stronger predictor of voting, although print
mass use is a positive predictor while electronic use is negative. Quasi-
mass use is the better predictor of political participation and community

integration.

The following conclusions are drawn: 1) mass media use continues
to be an important predictor of political behavior; however, 2) it is
'important to include quasi-mass media use in a model of communication
effects; 3) because the measurement of media use was restricted.t0‘
exposure other factors should be included in a multidimensional approach
to measuring media use; and 4) research on the effects of media use
should keep in mind the nature of communication and its effects as a
process-~other variables like community integration can mediate communica-

tion effects.

Copyright 1981

Joey Reagan

All Rights Reserved

ACKNOWLEDGEMENTS

The one person who made this finally come about, who provided
the security and love and stability necessary for my motivation to
proceed unencumbered, is my wife, Janay Collins. She begins her
doctorate now that I have finished mine. And I expect that when she
finishes she will be that many years beyond me.

Of course, the people who gave me the encouragement when I was
young to pursue whatever my intellectural interest desired were my
parents. My father is not alive, and while I don't believe in heaven
I do believe that a part of him is inside me to see all this. I am
glad my mother can see this, I hope I've made her proud.

James is always older and he taught me everything when I was a
kid. He read to me and taught me to play baseball and even challenged
me when I said something stupid about math or physics. That made growing
up a lot easier, but I think it made school too easy. I've never learned
how to work hard enough.

Julius accepted me, even though I was a long-haired, scraggly,
unemployed, misdirected, etc. That helped me think there was some good
left.

And my new parents have, I hope, taken a shine to me. I have to

them.

iii

There are my colleagues. My committee, Tom and John and John
and Stan who encouraged me, made me feel like I might have latched onto
something unique. John and Stan, special thanks for help with the
drafts, and for not letting me get away with any shit. And fellow
students I've laughed and worked with, Brian and Gail and Judi (whereever
you are) and Rick and Kim. And Jayne, who always did everything before
me because I'm such a chicken—shit. But we cried that second year for
no good reason. Thanks.

The people at the (other) University of Michiganinsthe Department
of Communication kept the heat off me this year so I could finish. I
like it-—keep it up. And Richard helped me conceptualize some of the
model and analysis.

But all work and no play . . . So for the lighter side of life
and contributions to sanity: Ed and Janice and Michael and Rick and
Sara and Hal and Jane and David .

And special thanks and a round of applause to one of the best
friends a fella could have (ta-dah): the vegetable garden. Seriously,
folks, I could just eat it right up. (Sounds like something from

Mickey One . )

 

Oh, and don't forget the motorcycle and the Carcinoma Angels. I
won't. And the house. OK. And the washing machine. Sure. And ...
So quit already!

I'm sure I forgot somone, something; I wonder .

iv

TABLE OF CONTENTS

LIST OF TABLES ................. . ................ ........... 1x

LIST OF FIGURES ...... ..... ........... . ..................... x

CHAPTER Page
I. INTRODUCTION AND DEFINITIONS ........................ 1

II.

IntrOduction .00....O.....OOI.......OOOOOOOOOOOOOOOOO 1

The Changing View of Media...................... 2
The Changing View of Political Participation.... 3

Definitions ................................... ...... 5
Definition of Communication ............ ........ 5
Definition of Political Participation .......... 13
Definition of Community Integration............. 16
Definition of Cause ............................ 17

The Need to Study Political Participation ........... l7

RELATION OF COMMUNICATION USE TO COMMUNITY INTEGRATION

AND POLITICAL PARTICIPATION.......................... 21

Predictors of Political Activity .................... 21

Relation of Communication to Political Participation. 21

Political Participation ....................... . 21
Exposure vs. Use ..................... ...... .... 29

Relation of Community Integration to Political
Act1v1tYOCOOOOOOOOOOOOOO......OOOOOOOIOOOOOOO00...... 35

Personal Uses of the Community ................. 35
Politics ......COOCCCOOC0.00.00.00.00. ..... .0... 36

III.

IV.

Page

Demographic Predictors of Political Activity ........ 37
Causes of Media Exposure ....................... ..... 38
Predicting Community Integration ........ ............ 40

Causal Model with Types of Media and Types of

Political Activity ............. . ........ . ...... ..... 43
Summary of Hypotheses ............................... 44
Predicting Political Activity ........ . ..... .... 44
Other Predictors ... ............................ 45
mmOD . . . O . OOOOOOOOOOOOOOOOOO . ....... . OOOOOOOOOOOOOO 47
Development of the Questionnaire ........ ............ 47
Measurement of Variables .................. .. ........ 48
Demographics ..... .............................. 49
Media use . . . ...... . ............................ 49
comnity Integration . . . O . O . . . O . . . . . O . O . 0 O . . O . . 51
Voting .... .................................... 51
Political Participation ........................ Sl
sampling ............ ................................ 52
InterViewer Training . . . . . O O . O . . O . O . O O . O O . O . O . O . 5 3
Field Work ........... ......................... . 54
Interview Verification ......................... 54
Sample Data . ........................................ 54
Model Analysis ............ ..... ...... ......... . ..... 55
Model Criteria ...................................... 63
LISREL Program Estimates ....................... 67
RESULTS . . . . ......................................... 69
Results of the First Model ........... ........ ....... 73
The First Model and the Six Criteria ........... 73

Problem Areas in the First Model ............... 76

vi

Results of the Second Model .. ....... ... ............ 79

The Second Model and the Six Criteria ......... 79

Tests of Hypotheses ................................ 87
Comparing Coefficients ........................ 89
Predictors of Political Activity ................ ... 89
Hla to Hld ... ............ . .................... 89
H2a and H2b ................................... 91
H3a and H3b ................ ..... .............. 91
Predictors of Other Endogenous Variables ........... 92
H4a to H4c .................................... 92
HSa to H5c .................................... 92
Other Results ...... ........ ........................ 93
Indirect Effects .............................. 95
Summary of Results ................................. 95
DISCUSSION . . . . . .. ........................... g ...... 97
Causes of Political Activity ...................... . 97
Mass Media ........................... ...... ... 97
Quasi-mass Media ........ . .................... . 101
Mass vs. Quasi-mass Effects ...... . ........... . 101
Community Integration ..... ......... . ....... ... 103
Education .... ..... .......... ..... . ............ 103

Causes of Community Integration .................... 105

Measuring Media Effects . ..... . ........ . ........... . 106
The Role of Media .......... . ................. . 106
Measuring Media Use ...... ..... ................ 106

Future Research ........................ . ........... 108

Conclusions .... ............................... ..... 110

vii

The Model as Process......
The Importance of Quasi-mass Media ............

APPENDIX A: STUDY QUESTIONNAIRE ........................

APPENDIX B: LISREL MATRIX SPECIFICATIONS FOR FIRST MODEL.

APPENDIX C: LISREL MATRIX SPECIFICATIONS FOR MODEL TWO...

LIST OF REFERENCES

viii

Page
110
111
113
121
123

125

LIST OF TABLES

Table
1. List of Parameters and Their Meanings for Figure 6b.
2. Parameters and Theoretical and Measurement Variables
for Figure 6b . . . . . . . . . . . . . . .
3. Means and Standard Deviations for Variables Used in
the Measurement Model . . . . . . . . . . .
4. Percentages for Categories of Variables Used in the
Measurement Model . . . . . . . . . . . . .
5. Correlation Matrix of Measurement Model Variables
6. Some of the LISREL Estimates for Model One .1; . . .
7. Parameters and Theoretical and Measurement Variables
for theiModel in.Figure 7b- . . . . . . . . . .
8. Correlation Matrix of Measurement Model Variables in

Figure 7b . . . . . . . . . . . . . . . . . . .
9a. LISREL Estimates for the Theoretical Model in Figure

9b. LISREL Estimates for the Measurement Model in Figure

ix

Figure
1a.

lb.

1c.

63.
6b.

7a.

7b.

LIST OF FIGURES

Characteristics of Three Media Along the Mass-

Interpersonal Continuum: Message Characteristics .

Characteristics of Three Media Along the Mass-
Interpersonal Continuum: Institutional
Characteristics . . . . . . . . . . . . .

Characteristics of Three Media Along the Mass-

Interpersonal Continuum: Audience Characteristics

Causal Model Predicting Political Activity from
Education, Media Use and Community Integration

Causal Model with Education Added as a-Predictor
Of media use . . O . . . . . 0 . . . . . . .

Causal Model with Length of Residence and Media Use

Added as Predictors of Community Integration

Causal Model with the Three Types of Media Use and
Two Types of Political Activity Added . .

Basic Theoretical and Measurement Model . .

Theoretical and Measurement Model with Notation .

Basic Theoretical and Measurement Model for Model Two .

Theoretical and Measurement Model with Notation .

Model Two with Standardized LISREL Estimates

30

31

32

33
56
S7
80
81

88

CHAPTER I

INTRODUCTION AND DEFINITIONS

Introduction

 

While there has been a considerable amount of research devoted to
the relation of mass media use, interpersonal media use and political
participation, a gap in communication research exists in two areas:
1) media other than mass or interpersonal (defined later as "quasi-
mass") have received scant attention, and 2) political participation
other than voting and major party identification (such as working for
minor parties, social protest and community participation) has not been
explored as an effect of media use. A

This dissertation will try to fill part of this gap by focusing not
only on mass and interpersonal media use, and voting, but also on "quasi-
mass" media use and other forms of political activity. The research
presented here will build a theoretical causal model (discussed in Chapter
II) that relates media use--mass, interpersonal and quasi-mass--to
political participation. Because social behavior arises in an environment
that includes other effects, the model will also include demographic and
community integration variables.

Before developing the model we will examine the need to incorporate

new ideas into our own views of media use and political activity, and

then present definitions of the constructs used in the research. In
addition, a discussion of the importance of predicting political activity

is presented.

The Changing View of Media

Traditional definitions of media have categorized media into two
groups: mass and interpersonal. But mass media are playing a relatively
less important role, and new communication technologies are opening new
media to public exploitation (Parker, 1973). Parker notes that, histor-
ically, writing was once reserved for the elite but became near-universally
used; so too audio and video may also become the domain of the many.
Parker further noted that access to new technologies means access to the
conduits of information, and that this access, if on a mass scale, can
help reduce the inequalities in the distribution of wealth and resources.

Maisel (1973) sees the United States engaged in the third stage of
industrial development (for more information on the third stage of develop-
ment see Bell, 1968; Clark, 1957). This third stage is characterized by
a shift from manufacturing to service industries. Along with this shift,
according to Maisel, comes a need for specialized forms of communication
to meet the growing needs of specialized services. These new media
would direct themselves at smaller, more homogeneous audiences. In an
historical analysis Maisel found that while mass media use continues to
grow, its growth rate has slowed considerably, and there is now a shift
from rapid increases in the use of mass media to more rapid increases in
the use of more specialized media.

It is possible to expand the "marketplace of ideas" with the

diSSemination of information through expanded use of new communication

technologies (Emery, 1978). The "marketplace" here does not mean the
traditional town meeting. It refers to the means by which people exchange
ideas. Using a network of micro computers, citizens could address

various sources of information: electronic mail, electronic publishing,
libraries, electronic town meetings, etc. This network would offer a
variety of uses, from interpersonal through mass communication, that could
be selected by the receiver. Emery posits that the potential values of
such a network would range from expanded access to instructional materials
to greater political participation; however, the system may portend limits
on such uses through its tight control by a few individuals or organiza—
tions. Emery states that, regardless of who controls it, the new tech-
nologies should offer the best chance to deal with the "staggering volume
and variety of information necessary for modern life." (p. 80) Access

to communication media, the development of new forms of communication and
less reliance on mass media suggest a need to assess the impact of the use

of such media on a person's relation to political and social processes.

The Changing View of Political Participation

 

In addition to the traditional view of communication media there is
a traditional view of political participation. In communication research
the focus has been on large group functions within the existing political
structure, i.e., a focus on electoral politics or major party (Democrat
or Republican) participation, with a major emphasis on presidential
politics.

These are not the only political processes open to citizens, and
these are not necessarily the ones that have the most impact on the daily

lives of citizens. Other political and social participation can have a

more direct impact on the psychological and social well being of
citizens. For example, when one feels disassociated from traditional
politics, will the response be to work foran.independent candidate or
minor party? Or will the response be to organize social protest such

as a revolution? Or are there local organizations through which citizens
can exert influence on a community level to achieve day—to—day better
lives (such as prodding city hall to fix the pot hole in the street)?

The function of local services may mean more than the election of a
president.

This research examines the traditional views of the relations of
communication to political participation, but a main focus will be on
quasi-mass communication and non-traditional forms of political particpa-
tion.

Of particular interest will be the development of a model that helps
explain the causal relations between media use variables (including quasi-
mass),identificationwdth a community, and political participation
(measured with items that include non—traditional political participation).
In addition, two demographic variables which have been shown to be the
only strong, consistent predictors of media use and community integration--
length of residence in a community and education-~will be included in
the model.

Before discussing the literature that relates these variables the
variables are defined and then a discussion of the importance of studying

the relations between them is presented.

Definitions

 

Definition of Communication

 

In defining communication we need to examine earlier definitions
of mass and interpersonal communication.

Mass communication has been distinguished by the following charac-
teristics by Wright (1959): 1) directed toward large, heterogeneous, and
anonymous audiences; 2) transmitted publicly, often to reach receivers
simultaneously; 3) transient in character; and 4) operated within a
complex organization that may involve great expense.

Additional characteristics provided by Menzel (1971) include:

1) standardized messages uniformly broadcast to all who may be concerned;
2) contacts too fleeting for messages to be tailored to the recipients;
3) severely limited feedback; 4) special expertise required in operation
of the medium; and 5) full control by the originating source.

Gumpert (1970) adds other characteristics: 1) 'the code of the
message is known to all, i.e., there is little use of jargon; 2) direct
cost to the receiver is minimal; 3) the communication is rapid; and
4) it is consumed on a short term basis.

Most definitions of interpersonal communication, especially in
relation to interpersonal media, use the opposite characteristics of mass
cummunication presented above.

Bienvenu and Stewart (1976) evaluated the characteristics that related
to the development of interpersonal communication. Several factors
related to characteristics within the communicator, self-disclosure and
self-awareness, while others related to the external nature of interper-

sonal communication, acceptance of feedback and clarity of code.

Barnlund (1968) identified five characteristics of interpersonal
communication: 1) physical proximity; 2) single focus of attention;
3) exchange of messages; 4) use of many senses at once; and
5) unstructured setting.

Mass media and interpersonal media would, of course, be the devices
or environment through which the communication takes place. These may
include everything from television and its adjuncts (the TV set, airwaves,
etc.) to face-to-face encounters with eyes, ears, sound waves in the
air, etc.

Imagine a continuum running from ideal mass communication to ideal
interpersonal communication. Using dichotomous pairs, taken from charac-
teristics described above, limits can be set up within which characteristics
of media and communication type would fall. These characteristics can be
grouped into three areas: those concerned with messages, those concerned
with the institution within which the communication takes place, and
those concerned with the audience.

Messages would be described between the following pairs: 1) public/
private; 2) standardized vocabulary/individualized vocabulary; 3) rapid
transmission/leisurely transmission; 4) transient/persistent; and
5) control by sources/considerable control by receiver.

The institution would be described between the following pairs;

 

6) complex,expertise required/simple, little expertise required;
7) limited access/ongoing opportunities for access; 8) high cost/low
cost; and 9) source physically far away/source proximal.

The audience would be described as between the following pairs;

10) large/small; 11) heterogeneous/homogeneous; 12) limited feedback/

instantaneous feedback; and 13) anonymous/known.

The continua of characteristics and the'mass-interpersonal continuum
are presented in Figures 1a - 1c. Media can be represented as having
various "amounts" of each characteristic. Taking all characteristics
together, media can be ranked as more or less "mass" or more or less
"interpersonal" on the main continuum. For example, the metro daily
newSpaper is portrayed as having characteristics close to the ideal type
mass medium, i.e., it ranks high on all characteristics; therefore, it
also ranks high on the mass-interpersonal continuum. Likewise, face-to-
face communication ranks low on all 13 characteristics; thus, it is ranked
lowest on the main continuum.

Of special interest are the media that fall between mass and inter-
personal. As noted above, these media are becoming more important because
of the increase in technological innovations that leads to greater access
to these media. Menzel (1971) has called this "quasi-mass communication."

The newsletter of a community association serves as a good example
of a quasidmass communication medium. The newsletter would relate to the
13 pairs described above in the following ways:

1) It is not public since it is intended for less than mass distri-
bution, yet it is not private since copies are often distributed in public
settings such as at a city hall.

2) The message may include vocabulary specific to the interests
of the association, like acronyms, but it must be written so that members
with various levels of knowledge may read it, e.g., new members.

3) It may be transmitted (distributed) at varying rates, at meetings,

through the mail, or left on counters in public buildings.

moaumaumuumumsu owmmmoz

 

 

 

 

 

 

 

 

 

 

uo>wooom scamm«5mswuy huwaaemuo>
up Houuaoo uooumfimuom haousmowg oonwaooua>fivaH oum>wum
[III
reveodeooooeoodve oeeoo
mousom woodmawue :owamwfimcoua huoasnmoo> owaaom
an Houuaoo enema emuaeuaeamum

A<20mmmmmMHzH

”Ssssfiuaoo Hmcomuooumuofilmmos one macaw memos manna mo moaumwuouumuononl.w~ ouowam

 

soaumu«saaaou
mummlouloumm

wouuoamsoz
cowuoaoomm<
hawGDEEoo

noemomsoz hawoa
omuwaooouuoz

mm<z

 

mm<z

uHmaao

 

muaumauouooumso souuouauoca
”Eascaucou Hmoomuoououcalmmms meu macaw memos money no moauoauouomumnoli.n~ ouowwm

Hoafixoum omoou< moufiooom omwuuooxu
mouaom umoo 304 uoovssn< oHuuHA .oamafim

 

 

 

_:o«umofisaaeoo
4 oommuouloomm

 

v00eecoeeeereotdveeeeoeo'eoe

... nouuonSoz
coauowoomm<
zuficaaaoo

Hoooamaoz Magma
smuuaooouuoz

 

 

 

 

 

 

 

 

mm3< use uwoolmmwm moaned ooswousm
maawuammem ouuaom mousse; omwuuooxm .xoaoaoo

10

mowumwuouumumso mooofims< “Esoafiuaoo Hocomumououcwlmmma osu wcoao mamas mousu mo mafiumuuouumuo;UIl.o~ ouswfim

 

 

 

 

 

 

 

 

 

 

 

 

someooom

esoox mooocmucmumaH msomcowoao: IHHoEm
: l , L L coaumuwooa600
oommlouloomm
nouuoamsmz
i............... ................l.......OOOOOO........... GOHUMHuomg
hufiooaeoo
_uoomomsmz magma
d souHHooouuoz

mooshaoc< xuonmoom msooawwououom owned

e323

ll

4) The messages may be related to continuing issues, such as a
school bond referendum. It will then have meaning beyond the time of
distribution.

5) Contributions to the newsletter may be open to all members of
the association. The actual composition of a newsletter, however, may
be limited to a committee.

6) A newsletter does not require the expertise of the journalist or
typesetter. It does, however, require some skill if it intends to commun-
icate with a variety of receivers.

7) Membership on the newsletter committee may rotate among members
or be open to volunteers. A designated staff is usually needed to
consistently turn out a newsletter.

8) Cost would be lower since volunteers would be used and printing
costs would be minimal. There would be more cost than with face-to-face
communication (excluding opportunity cost).

9) The source would be the association, close to the receiver in the
neighborhood, but not next door to everyone.

10) The number of members (audience) and prospective members would
probably not include all members of the community, but the association
would need more than a few members just to exist.

11) Members would be homogeneous to the extent that they share a
common interest (perhaps a position on a school bond referendum). They
would likely not all be personal friends with the same interests.

12) Feedback mechanisms would be built into the association:
election of leaders, speaking at meetings, letters to the editor of the

newsletter, and talking among each other. There would be limits on

12

feedback, e.g., limits on speaking at meetings to preserve order.

13) Members would know each other by face and through a membership
list, but not all would be personal friends, knowing each other perhaps
fleetingly.

Having defined communication, our attention turns to classifying
the media of communication, since it is these that will be of major
interest later in trying to relate the use of certain types of media to
political and community behavior. It would take too long to describe in
detail the reasons for defining the following media as mass, quasi-mass
or interpersonal. The reader might select a few and try to imagine the
characteristics of the structure of each one (cost of entry, for example)
as well as the messages usually presented by each and the audiences
usually attentive to each.

Examples of mass media are:

broadcast television

broadcast radio -

daily, weekly and Sunday metropolitan newspapers

certain cable television uses (broadcast, pay movie, sports,

entertainment)

general use magazines (including news magazines)

movies a '

books
Examples of quasi-mass media are:

certain cable TV uses (public access, data exchange)

trade magazines (to which professionals have access)

rofessional journals

newsletters

memos

church bulletins

non-highway citizens band radio use
interactive computer use, among small groups

13

Examples of interpersonal media are:

face-to-face

telephone

personal letters

interactive computer use, on a one-to-one basis

Definition of Political Participation

 

It seems that the definition of; political participation is as
varied as the studies about the subject (and this study will not break
that tradition). Researchers have used various terms to describe
political actions: "political activity," "political behavior," and
"political participation." There is no consistency across studies in
either the use of these terms or in the operationalization of them. In
a major review of literature on the effects of mass communication use on
political behavior Kraus and Davis (1976) never specifically defined
political behavior. They did say that most studies had focused on electoral
politics, but that there are also other forms of political behavior. But
even their discussion of the varying theories of social relation and
social movements--pluralism vs. elitism, for example-~focused on electoral
politics and campaign effects.

Kline and Tichenor (1972) discussed the effects of mass communication
on information diffusion, socialization and voting behavior. Even Stamm's
(in Kline and Tichenor, 1972, pp. 265-294) discussion of "environment"
focused on knowledge and attitude rather than behavior, and did not
address the development of the environmental movement as a political force
for specific societal change (See: Schoenfeld, 1979; and Schoenfeld,
et al., 1979). The emphasis of Schoenfeld was in changing attitudes

toward environmental issues rather than behavior.

14

Welch (1977) in a study of MexicanrAmericans focused on registration
and voting behavior as a measure of political behavior. In addition,
she considered only attitude toward the protest activities of Mexican-
American political organizations as a measure of such behavior.

Jackson-Beeck (1979) used five measures of political activity in
her analysis of mass communication effects across the U.S. All five
measures revolved around electoral politics: voting, attempting to
influence voting, opinion expression, working for a party or candidate
and wearing campaign buttons or displaying a bumper sticker.

Acock and Scott (1980) in developing a model for predicting political
participation used two measures of political participation: low-visibility--
passive behavior such as watching political programs on TV; and high-
visibility-—including the five measures used by Jackson-Beeck, above.

These measures of political behavior all revolve around electoral
politics and bear a striking resemblence to Matthews and Prothro's
"Political Participation Scale" (in Robinson, g£_§l., 1973, pp. 427-430).
This scale not only focuses on electoral politics, but also emphasized
party politics-and this among the major parties-dwith items such as:

"Do you ever talk about public problems with government to people in
politics, I mean Democratic or Republican leaders?" and their list of
clubs specifically emphasizes "Young Democrats or Young Republicans" as
political clubs or groups.

The emphasis on major parties and electoral politics ignores other
significant forms of political participation. These include: electoral
participation for non-major parties, independent candidates, minor
parties and the like; other forms of working within the political system

that do not involve elections, such as lobbying a political official and

15

major attacks on the status quo outside the legitimate political area--
everything from sit—ins to revolutions. Welch (1977), for example,

found that, among Mexican-Americans, identification with the ethnic
community, though low, was more related to engaging in protest than in
voting for Democrats or Republicans, and was a considerably higher pre-
dictor of protest than working in the major party campaign or registering
to vote. (Further relations between media use, community identification
and political participation will be discussed in the next chapter).

Thus, added to a definition of political participation should be
activity revolving around support of minor parties, independent candidates,
and ballot issues.

As will be noted later in the discussion of the relations between
media use, community integration and political activity, there is a need
to distinguish between voting as an indicator of political activity and
other political activities. This distinction stems from: first, the
literature shows clearer relations when using voting as the criterion
variable than other activity; and second, there is a logical difference
between voting and other activity. Voting involves a right-~often con-
sidered a duty--that can be executed secretly, and can express with
relatively little effort support for a political candidate or issue.
Other political participation involves tangible public action. Obviously
petitioning and canvassing involve the committment of time as well as
exposing one's political beliefs to public scrutiny, albeit some of
these publics may be very small in number. But even the simple act of
‘writing one's name on a petition involves making a public commitmentznuii

placing one's beliefs on public record.

16

Thus, for purposes of this dissertation political participation
will encompass voting and political participation (which will include

activity for major and minor parties and issues).

Definition of Community Integration

 

Because new communication technologies may engender greater feelings
of attachment to a community, and because community integration can
lead to greater political participation within the community (discussed
in Chapter II), the concept of community integration needs to be defined.

Many talk about community integration but none has defined it. From
social movement theory to communication models, the phrase "sense of
community" has been used but not defined. McLaughlin (1969), in
defining a social movement, included as essential to the movement organ-
ization a sense of community. Granosvetter (1973) proposed that organ-
izations are created more easily where there is a free flow of information
and ideas within a group, noting that this free flow takes on a "sense of
community." (p. 1373) In discussing the uses of CB radio, Gatseos and
West (1979) and Dannefer and Poushinsky (1977) each include creating a
greater sense of community as a beneficial consequence.

In the social movement and protest area, the community variables
discussed usually revolve around those that relate to the feeling of
integration within a community: primary group (family, friend or neigh-
borhood) integration (Isaac, et al., 1980), or community attachments
(talking with friends and neighbors) (Useem, 1980).

In the creation of a scale to measure sense of community, Abel

et a1. (1980) found that the items generated in their in-depth interviews

17

related primarily to feelings of belonging in the community, (The
"Sense of Community Scale" is used in the present study, and will
be discussed more fully, below.)
So this study defines community integration as: feelings of belonging

in and feeling of attachment to the community.

Definition of Cause

 

Since the model of interest will involve looking at not only inter—
relations between media use, community, demographic, and political activity
variables, but at the process in which these variables operate, it will
be important to develop a model that defines causal relations among the
variables. And it is important now to define "cause."

Causality has been distinguished by three factors by Asher (1976,

p. 11): l) covariation between two variables; 2) time ordering; and

3) elimination of other possible causes. The covariation and time ordering
will be posited in the theoretical development of this dissertation.
Covariation will further be tested in the analysis of the data to fit

the model. And other possible causes will be considered as part of the
analysis: if unexplained variance is large and systematic then one would
conclude that other possible causes have not been included in the model.

Thus, cause will be a mixture of theoretical development of expected

causal relations with confirmation or rejection based on the analysis.

The Need to Study Political Participation

 

We come now to the "who cares?" part of the dissertation. What makes

the study of political participation and of its causes important? Aside

18

from the "pure" nature of research--research gratia research--political
participation is an essential element to the maintenance of a democracy.
In addition, relatively new (new compared to 1776) media may have an
impact on political participation and thus an effect on the maintenance
or demise of democracy in the United States.

Mencken (1926) summarized the dogma of democracy with two axioms:
1) that people have an inalienable right to govern themselves; and
2) that they are competent to do so. In order to achieve self-government
people are to sustain political parties that support the aims of democracy
(Daveis, 1947). The people are also supposed to remove their support when
the party strays from support of the democratic ideal. This ideal is the
happiness of the community through the regulation of the actions of
people (Leggett, 1947).

The historic view of democracy is that political participation should
be through the development and support of appropriate political parties.
But the test of a true democracy is how well it extends representation
to the diverse elements of the nation--both through extending suffrage to
all and through representation of all classes (Mill, 1862). Lipset (1960)
noted that there are restricted choices in ideology among American parties.
Thus, those seeking an alternative ideology (such as intellectuals) are
forced to support extreme actions—~inc1uding revolution--to bring about
that choice.

So political participation can take a variety of forms--from trad-
itional party support and voting to revolution. Where there is dissatis-
faction with existing parties, the public's support should turn to other

parties or candidates if the tradition of democracy is to be upheld.

19

If the selection of alternative parties is blocked, more radical
approaches like revolution are necessary.

The ability to make political choices comes from the ability to
communicate. In his discussion of the functions of free expression,
Emerson (1966) found an identifiable function for free expression that
allows participation in community decisionemaking; the best judgments
for the common good are reached through judgments by all rather than
by an elite. Since all persons are not of one mind on any issue there is
a need to allow open communication among citizens to allow all alternatives
to be considered. Thus, there is also a need to allow access to channels
of discussion. Historically, these have been assembly and the printed
press.

Modern channels, such as television, have added a new dimension to
expression. With large amounts of time devoted to electronic media (for
example, the average American viewed over 25 hours of television per week
in 1980; almost 90 percent of households watched the major party conven-
tions in 1976; A.C. Nielsen, 1980) questions have been raised about the
impact of the mass media--media over which citizens have little control,
certainly less control of the content than a face-to-face encounter.

Posited effects of the use of mass communication media have ranged
from increasing voter turnout (Kraus and Davis, 1976) to the massification
of society which leads to individuals unconcerned with societal good
(summarized in DeFleur and Ball-Rokeach, 1975).

For the democratic ideal to be upheld-for citizens to participate
in community decision-making--channels of communication must be open to

citizen use. Where the system strays from the democratic ideal--where the

20

need for a new party arises, for example—~there is a need for the
dissatisfied to communicate among themselves in the creation of a new
party. Hence the need to assess the relation between communication uses

and political participation.

CHAPTER II'

RELATION OF COMMUNICATION USE TO

COMMUNITY INTEGRATION AND POLITICAL PARTICIPATION

Predictors of Political Activity

There are three areas of interest from which predictors of political
activity wild be selected: media use, community integration and demo-
graphics. Media use is important--as discussed in the previous chapter—-
because of its ability to open avenues of expression and thus enhance
political activity. Community integration's importance is discussed below;
its importantce lies in the positive relation between feeling a part of
the community and active participation in the community, both social and
political. Demographics are considered for two reasons: first, one must
at least consider the effect of characteristics that one must carry
through one's life, such as race; and second, other acquired characteris-
tics have been shown to relate to political activity. Actually, there is
no consistent demographic predictor-~except one--which will be discussed

below.
Relation of Communication to Political Participation

Communication might be used in two ways to affect political behavior;

first, it can make available to the citizen political information through

21

22

which political decisions would be made; and second, it can be used as
a persuasion tool to influence political behavior.

It is possible that through the control of information the mass
media can "mold" community sentiment, and thus control the perceived
need for political change? Can the mass media keep current political
parties in power by limiting the information available to citizens and
controlling their voting behavior? This would be true if those in office
controlled the sources of information. But it is editorial limits that
are imposed on the media by the editors rather than control of information
flow by politicians (Donohue, et_a;,, 1972). The mass media may still
determine who remains in power, through the biases of the media. Donohue,
et_al. (1972) summarized the literature on the "gatekeeping" function of
the mass media. They found that the media industry and the psychology
of the editor tended to limit the information a community can receive
noting that a local news story does not result from the needs of the
audience, but through the limits imposed by the bureaucracy of the
industry and the perceptions of what the editor thinks is true. In
addition to shaping messages, they found that certain information is not
reported; stories tending to expose institutional faults are "buried"
while those tending to unify the community are used.

The "agenda-setting" function of the media could limit the issues
around which citizens develop their political decisions. Kraus and Davis
(1976) noted that the mass media are selected as the primary source of
political information in several studies of political campaigns. Does
the selection of mass media as sources of information affect political

decisions? At least in voting this does not seem to be the case.

23

In the area of campaign effects, while the mass media have been
seen as effective in bringing sweeping campaign victories, research
evidence indicates that the mass media are relatively useless in achieving
political conversion (McCombs, 1972). Rather than the mass media leading
to changes in campaign decisions,McCombs found that those who are more
identified with a major party are heavier users of mass media than are
those with low identification or who are neutral, i.e., people who have
made a campaign decision then turn to the mass media. Basically, the
mass media serve as a useful tool in mustering the loyal partisans to the
polls and in reinforcing their party identification.

Kraus and Davis (1976), reviewing the "classic" voting studies—-

1940 through 1973, from Erie Country, Ohio to Michigan and Texas studies--
support the notion that the mass media do not provide a conversion effect,
but are primarily reinforcers of those currently supportive of the
political status quo. Those outside the traditional political arena,
eSpecially shifting voters, are the least exposed to mass media.

However, there is considerable evidence that heavier mass media use
increases voter turnout. Studies of political advertising in mass media
also seem to show the same effect.. While recall of information can be
enhanced by advertising, the less partisan voter is not converted by the
ads.

The notion that the mass media are effective in reinforcing existing
political partisanship is supported in recent works by Roberts (1979)
and Jackson-Beeck (1979).

Roberts found that persons who primarily used mass media for political

information about the 1976 presidential campaign found it easy to remain

24

partisan, supporting the reinforcement effect. However, those who
had gathered information through interpersonal association found it
more difficult to maintain their original decision on who to vote for,
suggesting that to change loyalty or insert a new loyalty (perhaps
to a new party or candidate) interpersonal media would be more useful.

In an analysis of national election data from 1976 collected by
The University of Michigan Center for Political Studies, Jackson-Beeck
(1979) summarized what seemed to be the general effects of mass media use
as well as demographic factors on voting and political participation:
heavier use of both newspapers and television is related to heavier rates
of political participation, 223 mass media use does not erase basic
differences in political participation among population subgroups, e.g.,
men are more active than women, the middle-aged are more active than
the elderly, etc. Jackson—Beeck's conclusion was that massimedia exposure
facilitates political participation among those already predisposed to
participate. A

While the use of mass media causes increases in political knowledge
or the use of specific political topics, it cannot be linked to political
conversion.. The effects of mass media use seem to be limited to:
1) increasing general levels of voting; and 2) reinforcing and mustering
those already identifying support for a candidate, party or issue.

For those who find themselves outside the mainstream of American
politics, how does their dissatisfaction.translate into a new political
party, support for an independent candidate or for independent "legislative"

action (such as petitioning for a ballot issue)?

25

One should find that where political parties and the current
political system have been legitimized the mode of communication should
be from the political arena to the citizens. Basically, one would
expect that one-sway, heterogeneous communication would take place--
mass communication. However, where the development of a competitive
party or system takes place there is a need first to develop the party.
This requires communication that involves easier access to feedback
systems, smaller, more homogeneous audiences, i.e., quasi-mass and inter-
personal communication. The need for such communication is clearer when
the system is to be overturned or there is a direct attempt to change
existing ideology (for a description of communication uses for social
movements, see Reagan, 1981). The need for such communication among
potential new party members is not as clear and has been given little
attention in the literature.

Blacks (who are also a large part of another dissatisfied group:
the poor) have found that the two major parties have ignored their economic
and political problems, especially at the community level ("New Party,"
1980). This has spurred the development of a new party (unnamed by the
reporter) whose goals include promoting candidates and becoming involved
in community organizational activities (including protest marches).

What type of communication was most effective in developing this
new party and its attendant protest activities? The use of press
releases to the mass media came 2:55: the core of two thousand members
had already defined the party's goals and ideology. Then the need was to
recruit the "mass" support necessary for state and national activities.

For the development of the party structure and goals and for its

26

community activity, quasi~mass and interpersonal media were more
useful.

There is a dearth of research on political activity designed to
challenge the major parties. Further, most studies have concentrated
on national politics and have ignored the need for alternative political
activity for regional or local problems. However, some relations have
been found between media use and feelings of being outside the existing
political arena.

O'Keefe (1980) found that "political powerlessnesS'was negatively
related to television and newspaper consumption while "political
alienation" was not. Thus, lighter users of these media were less
likely to consider themselves adequately represented by politicians
(O'Keefe referred to the then current government and major party can-
didates in his study--Nixon and McGovern) from the status quo. However,
this lower use of mass media does not mean lower political participation
since lower use did not relate to alienation.

Given the relative uselessness of mass media in assisting those
outside the existing power structure leads us to infer that other types
of communication are necessary.

Fainstein and Fainstein (1974), for example, observed the problems
encountered by a West-Side Manhattan school governing board madéiip 'of ‘
the neighborhood citizens. Deteriorating buildings and minority staffing
problems had been ignored by the school district's board and the city.
Finding themselves outside the current power structure, they found a
need to organize the community to lobby or compete with elected

officials. Even area newspapers ignored the efforts of the group since

27

the problems did not affect the entire area, and since there was no
readydmade channel available and since the group did not create one,
communication with the community was not achieved and the needs went
unfulfilled. Here then is an example of the uselessness of the mass
media and the failure of a cause through the lack of communication.

It seems clear that when one begins to compete with those in power,
one is less likely to receive mass media coverage (see Gitlin, 1980).
Further, the mass media are relatively useless until an issue becomes
"noteworthy." The case of a Chicano union strike in Texas received no
mass media coverage until two years of marches and a national boycott
forced the issue to the fore (Rada, 1977).

When trying to find the media that are useful for developing new
political ideas, we can take a cue from the literature on the dissem-
ination of innovation. After all, the purpose of developing a competing
party, candidate or issue is to introduce a new idea to the public-~one
that will be likely to be accepted.

Katz (1962) showed the importance of the use of non-mass media in
the dissemination of new agricultural practices to farmers and new drugs
among doctors. While mass media conveyed information, there was a
need for both groups to use other forms of information in acquiring
acceptance of the new ideas: friends, neighbors, colleagues and
specialized journals (uses of interpersonal and quasi-mass media).

While interpersonal information is useful in dissemination through
the "two-step" flow, Rogers (1962) suggests that interpersonal communica-
tion contributed more to the explanation of the variance in acceptance

of innovation than did mass communication.

28

For the acceptance of an idea in an engineering academic community,
Dahling (1962) noted especially the importance of "centers" of infor-
mation exchange.

To the extent that non—traditional political participation parallels
innovation acceptance, interpersonal and quasi-mass use is more important
than mass media use.

The foregoing discussion suggests that heavier mass media use is
related to increased voting behavior, voting being the easiest form of
political behavior and the one to which more people are culturally pre—
disposed. But it also suggests that other types of participation are
related to relatively less use of mass media and more use of quasidmass
and interpersonal media. Keep in mind that among both groups, those who
support major parties and those who do not, activity besides voting is
related to heavier quasi-mass and interpersonal media. This is true of
the former group because they find the mass media useless in fostering
change. It does not mean that mass media are not important or that they
will have a negative effect on political participation. It just means that
mass use will be relatively less important than it is for voting. Indeed,
one would still expect a positive relation between mass use and political
participating because the mass media are still useful for engaing mass
support.

Since all three types of media use are expected to cause both types
of political activity the following hypotheses are suggested:

Hla: Media use (all three types) is a direct cause of
voting.

Hlb: Media use is a direct cause of political participation.

29

But the use of each medium is expected to have a relatively greater
or lesser effect depending on the type of political activity. Thus:

ch: Mass use is a stronger predictor of voting than
is either quasi-mass or interpersonal.

Hld: Quasi-mass and interpersonal use are stronger
predictors of political participation than is
mass use.

The hypothesized causal relations are diagrammed in Figure 2 and Figure 5.

Exposure vs. Use

 

A question arises about the appropriate operationalization of media
use--should media use be measured as exposure to a particular type of
medium (e.g., hours of television use) or should one determine if the
medium was used for a specific purpose (e.g., using television for
political information rather than entertainment)? Both methods have been
used. Tan and Vaughn (1976) used hours of television exposure as a pre-
dictor of public affairs knowledge and Black political militant behavior.
But other studies (O'Keefe and Liu, 1980, e.g.) first asked which sources
were relied on for political information and then tied those responses
to political behavior, principally voting, finding significant relations
here.

Which measure to use might relate to what one is trying to prove:
select exposure if one is expecting a relation between exposure and
political activity, or select use if one is expecting the relation to be
with use. One might be tempted to think that it is logical that unless
a medium were used for political information it cannot be expected to
have an effect on political behavior. This is the logic behind the oper—
ationalization of media use as use of a medium for "How much the respon--

dent counts on television for making up their mind about who to vote for,"

30

souuwuwmuofi huacaasoo mam om: mauve .cowumosoo Eoum huw>fiuom Hmoauwaoe wowuowmoua demos HomemUII.N snowflm

 

ZOHH<MUMHZH
WBHZDZZOU

 

 

 

VHH>HHU<
A¢OHHHAom

 

 

 

 

 

 

mmD
<Hnm=

 

 

 

 

onH<UDnm

 

 

31

 

 

>9H>HHU<
A<UHHHAom

 

 

om: waves mo wouowvouo m on women :ofiumosvo nufia demos Homomoln.m ouswfim

 

onH<mumezH
VHHZDZZOU

 

 

 

 

 

mm:
<Hnmz

 

 

 

 

ZOHH<ODQN

 

 

32

ceauouwoucw kuwcosaoo mo mucuofioouo no women om: names mom oucmofimou mo nuwcoa sues Homes HomomUII.¢ ousmfim

 

 

ZOHH<MUMHZH
wHHZDZZOU

 

 

 

 

 

>HH>HHU<
A<UHHHAom

 

 

 

 

muzmnHmmm
mo mewzma

 

 

 

mm:
<Hnmz

 

 

 

 

ZOHH<UDGM

 

 

 

33

venom mua>wuom Hmoauaaoa mo momma osu mom waves «0 mmoau mouse onu Sues Homes Homaooll.m ouowwm

_\

 

 

onH<&UMHzH
MHHZDZZCU

 

 

 

 

ZOHH<mHoHHm<m
A<UHHHAom

 

 

 

R
w

e

UZHHO>

 

 

 

 

 

\

4/

V

 

A<zommmmmmHzH

 

 

 

/

mwSle m<DO 4/

 

 

 

 

mm<z

 

 

\

 

 

mozmnHmmm
mo mhwzma

 

 

 

onH<UDam

 

 

 

34

etc. (O'Keefe and Liu, 1980, p. 125)

When one considers causal relations, however, what little research
has been done on causal effects of media use on political attitudes and
activity suggests that media ggposure precedes political activity while
‘g§g_may, in fact, be caused by preceding political attitudes.

Tan and Vaughn (1976) explored relations between media exposure,
public affairs knowledge and political behavior (in their study political
activity focused on militancy). If use (vs. exposure) is the cause of
political behavior one would expect that exposure precedes public.affairs
knowledge which precedes militancy. While Tan and Vaughn did not do a
causal analysis, their results contradict the notion that use is the
causal element. If, on one hand, one finds that there is a relation
between exposure and public affairs knowledge, as well as a relation
between public affairs knowledge and militance, at least the basic
relational aspects of a causal relation between use and political activity
are SUpported. If, on the other hand, one finds that there is a relation
between exposure and public affairs knowledge, between exposure and

militance, but no relation between public affairs knowledge and militancy,

 

then one would tend to reject use as the causal element in favor of
exposure. In fact, Tan and Vaughn found the latter to be the case.

One causal analysis of political attitudes and media use suggests
that use is the result of previously held attitudes. Kimsey and Atwood
(1979) found that earlier attitudes were better predictors of later exposure
to mass media political messages about Democrats than mass media use was
of attitudes toward Democrats; and that earlier attitudes toward Democrats

was a better predictor of later attitudes than was media use. This suggests

35

that use of media for specific political information is a result of
previously held political attitudes,

The above results tend to support a model that uses exposure to media
as the relevant element to predict later political activity. From this
point both "exposure" and "use" will be employed as "exposure" to a

medium.

Relation of Communigy Integration to Political Activity

Personal Uses of the Community

A community is more than a social organization. It is also a
collection of individuals who interact to greater or lesser extents with
other individuals and organizations in the community.

Without a sense of community or feelings of integration, i.e., social
alienation from one's community, the individual will withdraw from the
community. The result may range from simply not using community services
to personal destruction.

Young and Wilmot (1962) found that residents of East London who felt
more a part of their community were also more likely to use the community
within which they lived for day-to—day services, shopping and going to
pubs.

Community integration is also related to common cultural understanding.
helping promote community understanding on a larger scale. Posner (1974)
believes that community integration is related to cultural and artistic
understanding which can lead to a more responsive community in distribut-
ing its economic services. This relation between community integration,

and community understanding in major metropolitan areas was the basis

36

for a cable TV system design in Detroit to help promote community
integration (Cable TV Study Committee, 1972).

The major concern among the elderly is crime (Louis Harris and
Associates, 1975). In searching for the social determinants of fear of
crime, Yin (1980) concluded that four major factors are important in
reducing fear, especially unreasoned fear, Three of these relate to
one's integration in the community: 1) interactions with friends and
neighbors; 2) the existence of a local support network; and 3) familiar-
ity with the structure of one's neighborhood. Yin notes especially that
reduced fear is related to feeling a part of the community and feeling
free to communicate with one's neighbors.

The most devastating effect of a lack of community integration is
self—destruction. Durkheim (1951) long ago proposed that there was a

relation between suicide and an individual's sense of community.

Politics

While community integration relates to several areas of social
behavior, one thing is clear: those more integrated into the community
are also more likely to participate in its political processes.

Feeling a part of and taking an interest in one's community relates
to political participation as well as the holding of power in a community.
Lazarsfeld, Berelson and Gaudet (1968) pointed out that those who were
more interested in community affairs were more likely to vote than those
whose interest was low. In a study of public affairs in Swindon, England,
Croll and Husband (1975) also found that those who felt more a part of

the community and had more interest in the community in general also

37

participated more politically. Hunter (1953), in his study of power
in the urban community, found that those who felt a part of the
community and who participated in community affairs also were more
likely to be those in power in the community.

Isaac et a1. (1980), in the only quantitative study of causal relations
between community integration and political activity known to the author,
found that their measures of community integration (family, friend and
neighborhood integration) related positively to political protest activity.
Integration was found to be a stronger predictor than demographics.

With only a few studies of community integration and political activity
available, there is a consistent suggestion that higher community integra-
tion leads to more political activity--ranging through all types of
activity, from voting to political protest. This suggests adding community
integration as a relevant element in the model predicting political
activity (see Figure 2). And it also suggests the following hypotheses:

H2a: Community integration is a direct cause of voting.

H2b: Community integration is a direct cause of political
participation.

These relations are diagrammed in Figures 2 and 5.

Demographic Predictors of Political Activity

 

In addition to their discussion of media use and political activity,
Kraus and Davis (1976) also summarized demographic relations to political
activity. They found a single consistent predictor: formal education
While other variables may have been shown to be related (race, socio-

economic status, for example) to political activity as well as media use

38

education is the single consistent predictor. "Education" is used here
as years of formal education. ~This is probably due to the fact that
education is either related to or is a major component of the other
variables: income is related to education, income is related to race,
education is a component of status, etc. A

In the Isaac, et al. (1980) analysis of causal elements predicting

 

protest activity, the same variable (education) was the strongest predictor
(of the set of four demographic variables which also included age, income
androccupation).
It seems reasonable to add education to the model (see Figure 2) as
a predictor of political activity. And adds the final hypotheses predicting
political activity:
H3a: Education is a direct cause of voting.
H3b: Education is a direct cause of political participation.
These relations are diagrammed in Figures 2 and 5.
.There are three predictors of political activity: »media use,
community integration and education (Figure 2).
We now turn to other areas of the model. Since we have identified
the predictors of political activity we can now examine causes of these

predictors.

Causes of Media Exposure.

 

There is a single area of consideration for which there is a
consistent cause of media use: demographics. One might wonder if media
use might also be caused by community integration, but the discussion of

the causes of community integration, below, suggests that it is media

39

exposure that causes integration rather than the other way around.

The relation of various demographic characteristics to media
exposure has not been consistently supported. Low socio—economic status
has been associated with heavier use of television and lower use of
newspapers (Martin, g£_§l,, 1976). Differences based on race have been
noted by Greenberg, g£_§i, (1970), and Comstock and Cobbey (1978). Even
position in the life cycle (Dimmick, E£_§l3' 1978) and geographic location
(Shaw and Riff, 1979) have been shown to be related to differences in
media use.

Although these studies purport to demostrate relations between
various demographic characteristics and media exposure, generally such
relations have not been consistent across all studies, or they can be
explained as representing relations based on other characteristics. For
example, while race was noted as a significant cause of media exposure,
Allen and Bielby (1979a) question whether this is truly because of race
differences, or--as their study showed-dwas really based on differences
in socio-economic status.

Kraus and Davis (1976) in trying to overcome the lack of consistency
in demographic indicators proposed that education was the only consistent
predictor of media use. This notion is supported by recent research by
Allen (1981) who found that among his predictors of television exposure,
education was the only significant predictor (the other demographics
included were: age, occupational status and income).

These studies have focused on the mass media, ignoring quasi-mass
and interpersonal exposure. However, lacking other evidence, it is

proposed that education is the only demographic that can be used as a

4O

reliable causal element in preducting media use, and this relation
shall be added to the model (see Figure 3), and the appropriate
hypotheses are:
H4a: Education is a direct cause of mass media use.
H4b: Education is a direct cause of quasi-mass use.
H4c: Education is a direct cause of interpersonal use.

These relations are diagrammed in Figures 3 and 5.

Predicting Community Integration

Having disposed of the causes of media use, we now turn to causes of
community integration. One can predict sense of community or community
integration from several perspectives. Two will be considered here:
media use and demographics.

The research done on the relation of types of media used to one's
integration in a community suggests that heavier use of mass media is
related to a lower sense of community, while heavier use of interpersonal
and quasidmass media is related to a higher sense of community.

Mass media are relatively useless for community integration because
they communicate principally "newsworthy" events. The local issue, the
personal conversation or small group interaction is not recorded in the
mass media, i.e., the mass media are not useful for local organization.
Fainstein and Fainstein (1974) noted that in West—Side‘Manhattan there
was only a mass medium (newspaper) available to the citizens. With no
quasidmass media available and with little personal contact, citizens
were not integrated with each other, leading to inaction on their part

when faced with threats to the operation of the schools in the community.

41

If mass media cannot help the individual communicate with the
community, what can? We need to look at interpersonal and quasi~mass
media use. Ganovetter (1973) proposed that interpersonal associations,
eSpecially associations among those with whom one has weak ties, those
extending beyond close friends and family, are important in contributing
to the flow of information and ideas in a community.

Conrath and Thompson (1972) believe that new technologies are
creating new forms of communication which are neither mass nor interpersonal
and which can help integrate an individual into the community.

Not only did Posner (1974) and the Detroit Cable TV Study Committee
(1972) recognize the value of community integration in contributing to
community understanding and shared values, as discussed above, but they
also recognized that non-mass media were necessary in helping to develop
a sense of community. Posner saw that the media of new technologies,
such as cable TV, could be used on a more community oriented basis.

The Cable TV Study Committee specifically suggested community access cable
TV channels, in neighborhoods and across the city, as a means for integra-
ting individuals into neighborhoods within the city

Even in the small town the availability of mass media can thwart the
continuation of community integration (Vidich and Bensmen, 1977). Local
persons, especially local politicans, because of the importation of
outside information no longer trust themselves or local experts for
community decisions. This helps erode confidence in the community. Thus,
more reliance on mass media would relate to a lower feeling of community
integration.

In a study of the relation of community involvement to radio use

Surlin (1977) found that involved citizens were less exposed to radio.

42

Those more involved would be more likely to use media that allow some form
of receiver control (such as quasidmass or interpersonal) than the mass
media.

In the study of political protest orientation by Isaac,,g§_al.

(1980) the model included demographic factors which were used to predict
family, friend and neighborhood integration. A single variable significantly
predicted all three and that was age. The other three demographic variables
considered were education, income and occupational status. Older persons
were more likely to be integrated in the community.

Age is also related because of changes in media use that are related
to changes in the life cycle. Dimmick, g£_§i. (1978) proposed that media
uses varied through several stages in life. For example, early adults
may use television news for relaxation while older persons may seek more
serious news.

Warren (1978) reported that the most important factors in predicting
community integration were homogeneity and mobility. Mobility is related
to age; very young persons are restricted to the local community by their
parents and lack of access to autos; older persons are restricted because
of physical and economic limitations.

Thus, based on the above discussion, one would expect the following
to relate to higher community integration: older, heavier use of inter-
personal and quasi-mass media, and lighter use of mass media.

A recent study by Reagan,_g£_gl. (1980) supports the notion that use
of quasi-mass and interpersonal media is positively related to feelings
of belonging in the community. In their study of predictors of sense of

community quasi—mass and interpersonal media use were stronger predictors

43

of a sense of community than were the mass media measures. Their results
also suggest that while age is a significant predictor, another variable—-
highly related to age-sis a stronger predictor: length of residence in
the community. ,Other studies that found age to be a strong predictor
may have been tapping length of residency.

Thus, the following predictive relations of community integration
are added to the model: media use and length of residence (see Figure 4).
The appropriate hypotheses are:

H53: Length of residence is a direct cause of community
integration.

HSb: Media use (all three types) is a direct cause of
community integration.

And since the use of non—mass media is related to higher community integra-
tion:

HSc: Quasi-mass and interpersonal use are stronger predictors
of community integration than is mass use.

These relations are diagrammed in Figures 4 and 5.

To some extent the causal relations are imposed upon the model. It
makes little sense to look at political participation causing media use
unless our interest is in increasing audience size as a program producer
or advertiser. The interest in this study is to look at factors that cause
political behavior. Determining factors that cause change in the political
structure can be used in a strategy by groups attempting to change the

political structure—-as in the replacement of a useless political party.

Causal Model With Types of Media and Types of Political Activigy

Figure 4 specifies the causal relations between general areas of

consideration. The model includes specific relations between media type

44

(interpersonal, quasi-mass, and mass), type of political activity
(voting and political participation), and other variables in the model.
In Figure 5 these specifications have been made.

Based on the foregoing discussion and hypotheses development we
expect the following: 1) education is a positive cause of the three
types of media use and the two types of political activity; 2) mass,
quasi—mass and interpersonal media use are positive causes of voting and
political participation, although mass is a stronger cause of voting, and
quasi-mass and interpersonal are stronger causes of political participation;
3) length of residence is a positive cause of community integration;

4) community integration is a positive cause of both political activity
variables;and 5) the three media use variables are causes of community
integration, with mass a negative cause, and interpersonal and quasi-mass

are positive.

Summary of Hypotheses

 

In Chapter I a definition of cause was given. That definition
will be expanded here so that the terms "direct cause" and "stronger"
are more fully defined. "Direct cause" means that a path coefficient in
the model is "significant," i.e., its chance occurrance is <.05. A
"stronger" predictor is presumed if the standardized path coefficient

is greater than the coefficient of another path.

Predicting Political Activity
The primary interest in this research is to find causes of voting

and political participation. Therefore, most of the hypotheses deal

45

with the predictors of such activity:

Hla:

Hlb:

ch:

Hld:

H2a:

H2b:

H3a:

H3b:

Media use (all three types) is a direct cause cf
voting.

Media use is a direct cause of political participation.

Mass use is a stronger predictor of voting than is
either quasi-mass or interpersonal.

Quasi-mass and interpersonal use are stronger
predictors of political participation than is
mass use.

Community integration is a direct cause of voting.

Community integration is a direct cause of political
participation.

Education is a direct cause of voting.

Education is a direct cause of political participation.

Other Predictors

 

Within the model there are other expected causal relations beside

just predictors of political activity. These are the predictors of

media use and community integration:

H4a:

H4b:

H4c:

H5a:

H5b:

H5c:

Education is a direct cause of mass media use.
Education is a direct cause of quasi-mass use.
Education is a direct cause of interpersonal use.

Length of residence is a direct cause of community
integration.

Media use (all three types) is a direct cause of
community integration.

Quasi-mass and interpersonal use are stronger
predictors of community integration than is mass use.

Having developed the set of hypotheses, our task is now to take those

hypotheses and the theoretical model and submit them to an empirical

examination.

The measurement of the constructs in the model, collection

46

of the data, and the specific analytic technique used to test the model

with the data are described in the next chapter.

CHAPTER III

METHOD

Develgpment of the Qpestionnaire

The questionnaire was developed as a personal interview instrument
for the "Media Environment Study" at Michigan State University. [Funded
by National Science Foundation Grant #DAR-79106l4, principal investigators:
Dr. Thomas F. Baldwin and Dr. John D. Abel.] The general instrument
was developed in relation to major research questions arising from the
"Media Environment Study." These dealt with comparisons across different
media environments (cities with many choices of media, like Detroit, and
those with fewer choices, like McAlester, Oklahoma of perceived uses for
the major mass media (TV, Radio and Newpapers). The additional questions
relating to political participation, community integration and other
media use were introduced by this author. However, all questions then
were compiled into a draft instrument and subjected to criticism by the
project team. The questionnaire was pretested with a convenience sample
of about 20 persons who were selected to represent a variety of ages
and socio-economic backgrounds. These included, for example, an 84 year
old widow, a poor middle-aged female, and a graduate student. In addition,
other comments from the agency conducting the interviewing (Market

Opinion Research of Detroit) were incorporated into the final draft. The

47

48

process of developing the questionaire took six months (October 1979
through March 1980) with the pretesting and rewriting taking another
two months. The questionnaire underwent six revisions.

The final questionnaire was printed on one side of 8-1/2" x 14"
paper. It was 43 pages long, and required approximately one hour to
complete the interview in the home.

Because of the possibility that interviewee fatique would affect
reports of media use, the three sections that dealt with TV, Radio and
Newspaper use were systematically rotated so that a third of the inter-
viewees were asked TV questions first; one-third got radio first; one-third
got newspapers first.

In assessing weekend media use half the respondents were asked about
the previous Saturday and half the previous Sunday.

In addition, several areas of the country selected for the sample
did not have daily newspapers. So an additional two versions were
prepared: one asking about daily newspapers and one for weeklies. Thus,
there were twelve versions of the questionnaire: for example, TV first-
daily newspaper-Saturday media use; TV first-weekly-Saturday; TV first-

daily-Sunday.

Measurement of Variables

 

A portion of the questionnaire usedflrlthe "Media Environment Study"
is reproduced in Appendix A. Only those questions relevant to measures
of the variables in the present study are contained in Appendix A.
These questions are used to operationalize the measurement of the variables

discribed below.

49

Demographics

The two demographics used in the model were: length of residence
in the community (years); and education (highest level obtained: less
than 8th grade, 8th, some high school, high school diploma, some college,
college degree, some graduate school, graduate degree) (associate or
trade degrees beyond high school were coded as: some college).

In addition, the following were also aSsessed as descriptive
statistics of the sample: age (years); whether the residence was owned
or rented; respondent's marital status (married or not); income (in
increments of $5,000 from "$0 - $4,999" through "$50,000 and over");

gender; and race.

Media Use

Media use was operationalized as exposure or time using the media.
The following were considered mass media: television, radio, weekly and
daily newspapers, movies (number seen in previous month), magazines
(number read regularly) and books (number read previous month).

The following were quasi-mass: trade and professional journals
(minutes read, previous week), newsletters (read/do not), church bulletins
(read/do not), citizens band (CB) radio use (hours, previous week).

And the following an index of interpersonal: face-to—face (persons
talked with the previous day) plus telephone (personal phone calls,
previous day). These two were combined because they shared over 20 percent
variance.

Some indicators of mass media exposure were indexed: TV exposure--
respondents were asked by day-parts the number of minutes they watched

TV on the previous day and the previous Saturday or Sunday. Average

50

weekly TV use was computed by: 5 x (previous day total minutes) + 2 x
(previous Saturday or Sunday total minutes) = total weekly minutes

(TV use). Media exposure generally focused on the previous day since

it was reasoned to be a more reliable estimate of exposure. Allen
(1980) found that previous day estimates of media exposure were the most
reliable, a specific day next most reliable, and average day the least
reliable. Previous day and specific day had reliability coefficients of
.85 and .83 for exposure estimates over two points in time, and average
day was .71 in Allen's study.

Radio exposure was assessed the same as was TV exposure.

Newspaper use was measured in two ways: daily and Sunday exposure
was indexed similar to TV and radio: 6 x (previous day's daily exposure)
+ (Sunday exposure) 8 Total minutes newspaper exposure. Weekly paper
consumption was total minutes read in the previous week. All estimates
were for local papers.

The reliability of using previous day estimates for media exposure
may be confounded if systematic biases enter the data gathering. For
example, asking previous day use only on Thursday in one city and on
Monday in another could introduce artifical variance between cities;
asking only on Thursday in all cities could i‘nf‘i'ate TV viewing estimates.
The problem was handled by randomly assigning interviews across all days,
Tuesday through Saturday--eliminating Sunday and Monday means that the
"previous day" was not a weekend day. Although there was some variation
from an even distribution because of callbacks occurring on a later day,

the overall pattern was to distribute interviews throughout the week.

51

Community Integration

Community integration was assessed using the Sense of Community
Scale (Abel, L31 ., 1980). Seven items, measuring feelings of belong-
ingness and neighborliness, were asked. Each item has five responses:
strongly agree through strongly disagree. This is a summated scale
comprising a single item to assess integration with a range of 7-35
(higher scores indicating a higher sense of community). Abel, g£_§l.
report reliability for their entire scale (which includes a total of
19 items that also measure community activity and tolerance) as an alpha
of .82. The present study found that the scale had an alpha of .77 with
all item-total corrected correlations in excess of .43 (over 18 percent

variance shared with the scale).

Voting

Voting behavior was assessed with the following three variables:
1) did respondent votezhnthe 1976 presidential election? 2) in the
1978 congressional election? 3) in the last local election? These three
indicators of voting behavior were dichotomized as did (1) or did not
(0) vote,auuisummed to form an index of voting behavior with a range of
0-3. The scale had an alpha of .73 with each variable having item-total

correlations in excess of .56.

Political Participation

Political participation was also indexed, this time with items
similar to those used by Matthews and Prothro (in Robinson, 1973).
However, additional questions to assess participation with minor parties

and independents was included. Respondents were asked: 1) whether they

52

had tried to get a candidate on the ballot; 2) whether they had

given money, attended rallies or helped canvas for Democrats or
Republicans; 3) whether they had tried to get an issue on the ballot;
and 4) whether they had given money, attended rallies or canvassed for
independents or minor parties. These four items were summed (range 0-4)
to form a single scale of political participation. This scale had an
alpha of .57 and item-total correlations in excess of .312. This lower
alpha is not unusual for a scale with so few items (Nunnally, 1978).
Note that all items contribute in excess of ten percent of the variance

in the scale.

Sampling

The sample was drawn from a national selection of 17 cities.

Sampling within each city was done by Market Opinion Research (MOR),
a survey research organization with offices in Detroit. The firm has
been responsible for field work on studies for several major educational
institutions, including Michigan State University, Northwestern University,
Yale University and The University of Michigan.

MOR used a cluster sampling method-~called the "probability-proportion-
ate-to-area" cluster. Within each city the total number of interviews
was divided by five to determine the number of clusters for each City.
Numbers in the remainder were assigned randomly, one each, to the initial
clusters. Clusters were based on the most recent counts of occupied
dwelling units available. In each cluster five or six households were

selected, based on the following plan:

53

Each interviewer was provided with a map of the cluster to be
sampled and the total number of interviews needed within the cluster.
When six households were in the cluster, the interviewers were instructed
to interview three males and three females. When five households were
in the cluster, the interviewers were assigned three males-two females
and three females—two males alternately. A randomly chosen dwelling
unit was designated the starting point. The remaining households were
identified by a skip interval and of four occupied dwelling units. The
interviewers were provided a walk pattern and directions for callbacks.

Callbacks were conducted twice at each household--three times in the
two largest cities, Dallas and Detroit, because of the greater difficulty
of reaching people in large cities. Interviewers were instructed to
make callbacks at different time periods during the day, in order to
increase the probability of finding an appropriate respondent at home.

If no one was home at the selected household, the interviewers were to
contact the first house on the left, following the same callback procedure.
If this failed to produce an interview, the first house on the right

was selected.

Interviewer Training

Training in Detroit was conducted by the MOR director assigned to the
project. In the other communities, Where MOR had subcontracted the
interviewing, the project director trained the on-site supervisor by
telephone. The supervisor then conducted face-to-face training sessions
with the local interviewers. During the local sessions the MOR project

director in Detroit was available to answer questions by telephone.

54

Field Work

 

Personal interviews were conducted in the selected cities from
June 17 through July 26. The following communities were used: Albany,
GA; Augusta, AK; Buffalo, SD; Cedar Rapids, IA; Clovis, CA; Dallas, TX;
Detroit, MI: Eureka, NV; Liberal, KS; Manchester, NH; McAlester, OK;
Mesa, AZ; Missoula, MT; Portland, OR; Quincy, IL; Randallstown, MD;

Tell City, IN.

Interview Verification

MOR verified ten percent of each interviewer's work or at least one
interview for each interviewer. The verification.was by telephone: the
respondent was called and asked selected questions from the interview
including length of residence and age. In addition, the respondent was
asked if the show cards required for the questionnaire were used and the
approximate time the interview took. If any of an interviewer’s work
failed to verify, MOR then verified all of that interviewer's work.

Michigan State University project personnel telephoned two respondents
from each city in the sample. These respondents were selected from those
who had not beencontacted by MOR. The MSU verification caller asked if
the respondent had been interviewed and thanked the respondent for

participating in the project.

Sample Data

 

Interviews were completed with 1828 respondents. The lowest number of
interviews in a city was 75, the highest was 121. Almost half (48.8%)
were male; 53.5 percent had annual household incomes at or above $15,000;

84.1 percent were White, 13.0 percent Black; 39.1 percent had some

55

college education or more; 68.0 percent were married; 79.2 percent
lived in a house, 5.6 percent in a mobile home and 14.3 percent in an
apartment; 71.9 percent owned their residence; 19.2 percent lived in

rural areas.

Model Analysis

 

The model developed in Chapters I and II lends itself logically to
an analysis applying the maximum-likelihood approach. This approach
is apprOpriate since it does several things: 1) it takes into account
sizeable measurement error often encountered in social science,
2) several equations can be analyzed simultaneously (for example, in the
present model, the several equations relating variables to both voting
and political participation as well as the equations relating measurement
variables to theoretical variables); and 3) where there are problems
with the model the analysis can indicate which parts of the model are
causing the poor fit and suggest changes for a better model. (See,
variously: Joreskog and Sorbom, 1977; Joeskog and Sorbom, 1978; Long,

1976; Kluegel, et al., 1977; Maruyama and McGarvey, 1980).

 

A computer program developed by Joreskog and Sorbom (1978) called
LISREL (Linear Structural Relations) is designed specifically for
analyzing a model through the maximum-likelihood approach. It is this
program that will be used to analyze the theoretical model presented above.

The program requires a set of parameter specifications for the
theoretical model as well as sets of indicators of each of the theoretical
variables (the measurement model). The theoretical and measurement

models are specified in Figures 6a and 6b. Figure 6a presents just

56

Hovoe unwaousmmma can HnoauouooSu owmmmll.no musmfim
macaw
huficaaaoo
mo macaw

 

 

 

 

 

 

 

    
   

  
  

 

 

 

 

  
  
  

     
   

xmvaH coaume munmpammm ouawvammm
:ofiumaauauumm ¢I Ifiofiuumm $IIIIIIIIIIIIIIHI. mo sumswq mummy
Hmuauaaom Hmuauaaom
ET 'I
€32, ‘ Iv 3
mm: mfivmz
xwvaH

Hmcomuwa
luwucH

HwaomumnuwuaH

wHSmONNm
mm: \_ a.

mava
mmmz

llllllllllllJv wusmomxm
Omem _
musmomxm nmmmm
ampssm\%afima
musmo xm
mm:

 

 

 
  

 

om:
OHUMM no

 

umamm thwoB

 

 

wusmo xm

_ mm: ousmo xm wusmo xm
Hacksaw omega zoom mcwummmz mfi>oz

 

mm:
aflmiam £326

 

 

kuumeBUZ

57

o.o away unnuo umxaw uouum mo.~ cmsu umnuo vmxfim Heumofivcfi

o.o um noxaw uouum mo.~ um vmxam noumowvcam

AmSOfiuwawmmw you N a g mmHnwu womv coauMuo: sags Hopes unwamusnmma was HQUfiumuowseul.no whamam

 

 

 

 

: )

 

 

 

 

 

.mu
..N

7»:

 

IJv . -

 

 

 

 

 

 

 

.: m“ I "mm!
:1 i. .. f ’ V

u t
/ _ \

m ha
a N «A final. : m...

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Arllo

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

58

Table 1.--List of parameters and their meanings for Figure 6b

 

 

 

Parameter Meaning

Y Measure of dependent variable

X Measure of independent variable

6 Residual of dependent measure

5 Residual of independent measure

n Unobserved dependent variable (endogenous)

E Unobserved independent variable (exogenous)

y Coefficient of interrelation of endogenous with
exogenous variables

-8 Coefficient of interrelation of two endogenous

variables '

; Residual of endogenous variable

¢ Covariance of two exogenous variables

w Covariance of residuals of two endogenous variables

A Coefficient of measure of unobserved variable

 

59

Table 2.-Parameters and theoretical and measurement variables for

 

 

 

Figure 6b
Theoretical Model Measurement Model
Parameter Variable Parameter . Indicator
EXOGENOUS:
El Education X1 Years of formal education
52 Length of residence X2 Years living in community
ENDOGENOUS:
n1 Mass media use Y1 TV exposure
Y2 Radio exposure
Y3 Daily/Sunday newspaper
exposure
Y4 Weekly newspaper exposure
Y5 Movie exposure
Y6 Magazine exposure
Y7 Book use
n Quasi-mass media use Y Trade/professional
2 8 journal exposure
Y9 Newsletter use
Y10 Church bulletin use
Y11 CB radio use
n Interpersonal media Y Index of persons talked
3 use 12 with and phone use
n“ Community integration Y13 Sense of Community Scale
n5 Voting behavior Y14 3-item voting index
"6 Political participation Y15 4-item political activity

index

 

60

the variables used in the model. The circles contain the theoretical
variables which are connected by paths indicating the causal relations
developed in Chapter II. The rectangles contain the measurement variables,
those used to operationalize the theoretical variables, with paths
indicating which measurement variable is used as an indicator of which
theoretical variable. The measurement variables are those discussed
earlier in this chapter. Figure 6b contains the model in complete
notational form, with coefficients and indicators of measurement error
entered into the model. Tables 1 and 2 contain complete definitions of
the parameter specifications in Figure 6b.

Note that the model in Figure 6b contains not only specification
of the variables of interest, but also the error associated with
measurenent (e.g., 51, 62) as well as error associated with each set of
equations (e.g., cl, :2). Each path--for both indicators of theoretical
variables and paths within the theoretical model—-has an associated
coefficient that is the wright used in the estimating equation for that
part of the model. For example, the relation between mass use and daily/

Sunday paper exposure is: Y = 1 (n1) + 5 Likewise, the theoretical

3 5 3'
model has an estimated set of weights, for example: n

4 ‘ 841‘“? + 842
(n2) + 843 (n3) + Y4 (52) + :4 (The minus sign on the BS in Figure l is
due to the fact that the matrix appears on the left side of the
estimating equation.)

In addition to the model accounting for measurement error, it can
also account for correlated error terms. This can account for underlying
systematic variance that is not specified in relations in the model.

The decision to allow error terms to vary or not vary together may come

from two perspectives. First, there may be compelling theoretical

61

reasons to allow covariance among error terms, For example, Allen
(1981) allowed the measurement error for his time-one measures of

media exposure to vary with the time-two errors since there was reason
to suspect related errors through the use of the same measure over time.
(In fact, his results indicate the errors were unrelated.) Second,

one may allow covariance among error in order to allow the model to fit
the data more precisely and thus obtain an overall model that provides

a better general fit. This has been done by Isaac, et a1. (1980) who

 

first kept covariances among errors fixed at zero and then allowed errors
to covary, one-by-one, until an acceptable fit of the model to the data
was obtained.

There are instances, however, when one just assumes that measurement
errors are randomly distributed and proceeds with fitting the model as
best as possible on this assumption (Acock and Scott, 1980; Maruyama and
McGarvey, 1980).

» The perspective that this research takes is that errors will be
allowed to covary if there is a compelling reason to do so. Otherwise,
the errors will be assumed to be randomly distributed and uncorrelated.
This follows from the first perspective described above. However, it
does not allow error to correlate in order to provide a better fit of the
model--the second perspective. There are several reasons for this:
First, to merely allow various combinations of correlated errors and
selecting the one that provides the best fit of the model merely
capitalizes on chance. With 105 bivariate comparisons for the dependent
variable measures alone, at least 20 of these ought to improve the model

by chance, if our criterion for model fit is .05. To allow all error to

62

covary would certainly overfit the model. Thus, one would be compelled
to try various permutations of errors until a better fit was obtained.
This would result in an even greater opportunity to capitalize on
chance. Second, Maruyama and McGarvey (1980) point out that such
manipulations run the risk of overfitting the data (p. 508) and violate
one of the criteria for judging the fit of the model (See below, "Model
Criteria"). Third, Maruyama and McGarvey further point out that manipu-
lations of the error covariance parameters uses the LISREL analysis for
exploration when it is designed for confirmatory analysis. While there
may be some changes suggested in the results of the proposed model our
interest is in keeping changes to a minimum in order to remain as close to
the confirmatory and theoretical processes as possible. In the model
in Figure 6b the errors in the theoretical variables for media use are
allowed to correlate as are the two for voting behavior (W ). This is
because underlying components are expected. The measures of media use
are taken from a single perSpective, exposure. Because there are other
factors that may compose media use, such as the purposes for which media
are used, these other factors may cut across the distinction made in
this model (mass-interpersonal). Therefore, correlated error terms will
give us an indication about the importance of such other underlying
factors. Likewise, political activity may involve other components
beyond simply an active or passive political activity.

As with error terms for the theoretical model, error terms for
the measurement model can be allowed to vary together. That has not
been done for this model. There is no theoretical reason as compelling

as there was for the theoretical variables discussed above.

63

Generally, specification of error terms has been done to test the
feasibility of separating theoretical variables. Acock and Scott
(1980), as in the present study, were interested in separating two
types of political behavior, and, therefore, specified correlated error
to both account for such error in the model and as an indicator of other
underlying characteristics.

The results of the LISREL analysis will allow us to determine:

1) the overall goodness-of-fit of the model; and 2) the relative
usefulness of the indicators and the theoretical path coefficients

through a significance test (t—ratio) and comparisons of their standardized
coefficients. The hypotheses, of course, can be tested using t-ratios

and comparison of the standardized coefficients.

Model Criteria

 

There are several criteria by which one determined whether or not
there is a good fit of the model to the data. These include:
1) Is the model correctly specified? 2) Is the X2 test nonsignificant?
3) Are the first order derivatives of the fixed parameters in the
model zero? 4) Are the residuals of the input minus the predicted
matrix (S ~23) as small as possible? 5) Is the explained variance in
the theoretical model as high as possible? and 6) Is the standard error
for the coefficients within the model low enough to allow discrimination
between coefficients and zero, i.e., are the coefficients significant?
The six criteria are explained more fully in the following paragraphs.
For a model to be correctly specified it must provide unique

path coefficients. Overspecification--identifying too many free para-

meters--will generate unidentifiable coefficients. The LISREL program

64

will tell the researcher if overidentification occurs with the following
statment: "THE NTH FREE PARAMETER MAY NOT BE IDENTIFIED." If this
statement is absent one assumes the model is correctly identified.
Usually, the goodness-of-fit is tested with a chi-square looking
for a value small enough to produce a probability greater than .05.
Unfortunately, with large sample sizes it is unlikely that one will obtain
a nonsignificant chi-square (Joreskog and Sorbom, 1977, p. 318; Long,
1976, p. 171). This is not necessarily bad. As sample sizes approach
infinity they are unbiased with regards to violations of normality
assumptions (Long, 1976, p. 166). In addition, the chi-square is merely
an indicator of relative fit of the model. Joreskog and Sorbom (1977,
1978) state that a chi-square with a probability less than .05 is accept-
able with large sample sizes, that one merely uses the chi-square as an
indicator of how a change in the model affects the fit. Joreskog and

2

Sorbom suggest that X can be used to test the relative improvement in

model respecifications. This can be done by comparing the reduction

in X2 with the change in degrees of freedom. Changes that are about the
same as the change in degrees of freedom are probably just capitalizations
on chance, whereas changes that are larger than the change in degrees

of freedom indicate genuine improvement in the fit of the model.

Allen (1981, p. 246) provides a sample calculation. A test of significance
is used based on the change in X2 with degrees of freedom equal to the
change in degrees of freedom from model one to model two. In this case

one wants a significant x2 because one is testing for a change in model
results that are not due to chance.

The first derivatives of the fixed parameters should be zero

(Joreskog and Sorbom, 1978, p. 15). If they are not then it indicates

65

that some fixed parameters should be allowed to vary, starting with
the fixed parameter having the largest first derivative.

The residual matrix (input minus predicted matrix) should contain
relatively small values. No specific level is given as being too
large. Joreskog and Sorbom (1978, p. 15) and others (Maruyama and
McGarvey, 1980; Acock and Scott, 1980; Isaac, g£_§l., 1980) use the
residuals as a subjective guide to the overall ability of the model to
predict the original input matrix. Several large residuals, relative to
the overall matrix, indicate a need to restructure the model. The large
residuals also give a clue about which parts of the model to change.
As a test of the magnitude of all the residuals, Maruyama and McGarvey
computed the mean correlation and the mean residual, excluding diagonal
elements. The lower the ratio of the mean residual to the mean correlation

the better, since this indicates relatively lower residuals. Maruyama

 

and McGarvey had a ratio of .33. This will be used as a guide in testing
the results in the present study.

Acock and Scott (1980) use explained variance in their endogenous
variables as an indicator of the fit of the model. This follows logically
from the fact LISREL accounts for measurement error. Thus, explained
variance in the theoretical model should be relatively high. The explained
variance (R2) is computed as one minus the residual.(l.-C). Acock and
Scott found explained variances of 25 percent and 40 percent in their
political participation variables. For purpose of the present study,
the explained variances of the two political activity variables--the
primary variables we are trying to predict--relative to the Acock and

Scott results will be used as an indication of the fit of the model.

66

Finally, examination of the path coefficients will tell us how
useful the model is with respect to causal relations. A large number
of nonsignificant paths may indicate relatively large standard errors.

A restructured model that acquires significant paths indicates a
reduction of the relative standard error and a better fit of the model.
(Of course, if a restructured model does not produce significant
coefficients it may mean there are simply no relations.)

The literature on what can be done to improve the fit of the model
by restructuring the model is rather scanty. Joreskog and Sorbom (1978,
p, 15) only suggest freeing some fixed parameters. No one has addressed
the issue of a major restructuring of the theoretical and measurement
variables. For example, an option available beside freeing parameters
would be to eliminate one or more variables. Another would be to restruc-
ture the measurement model with respect to the theoretical model. Of
course this takes us into a more exploratory realm. But even the freeing
of parameters has eliminated a strict confirmatory analysis, and if the
results indicate restructuring that does not conflict strongly with the
theory then--as they say--why not?

There is, however, a solid statistical reason for altering the
model beyond merely freeing some parameters. This is embodied in the
concept of "construct validity." To the extent that the results fit the
theory the measurement can be said to be valid. Where the measurement
fails to fit the theory significantly (in a statistical sense) one can
consider the model "invalid." See Woelfel and Fink (1980, pp. 85-86)

for a discussion of theory and measurement as determinants of validity.

67

LISREL Program Estimates

In order to begin the 'iterations of the LISREL program, estimates
need to be made of most of the coefficients in the model. This will
do two things: first, it will help reduce the time involved in computa-
tion, and second, it will increase the precision of the program's solution
for the model by providing error estimates for some of the fixed para-
meters, thus, eliminate this source of confounding variance from the
model solution. The start values and specification of fixed and free
parameters are contained in Appendix B.

Estimates for the endogenous variables with multiple indicators will
be made using factor analysis with a single factor solution. The lambdas
will be estimated with the factor scores and the epsilons will be
estimated with the residuals for each variable (l-hz).

Several of the endogenous variables and the two exogenous variables
have single indicators. Normally these would be estimated as "1.0" with
error assumed to be zero. However, since several have been created as
indexes or scales scale reliability estimates will be used to estimate
these coefficients. Their residuals will be used to estimate errors
(Winer, 1971, p. 285; Acock and Scott, 1980). These values will remain
fixed. Technically, one need only'indicate some start value other than
zero in order to have a free parameter. Indicating start values merely
saves time in running the program. However, for fixed parameters, such
as error estimates for fixed parameters, indicating start values will
give the program more information and allow a solution that gives
greater explained variance in the theoretical model.

The matrix to be analyzed will be the correlation matrix. This is

done for the reasons stated by Maruyama and McGarvey (1980, p. 509);

68

the data are from a single population, cross-sectionally gathered,
and-~most importantly--standardiéed coefficients are far easier to
interpret than are nonstandardized coefficients, especially when
comparisons of coefficients are to be made. Use of the correlation
matrix also fits the theoretical relations proposed in the hypotheses.

The results will search only for significant predictors ("causes") and
relatively larger coefficients. For the latter tests, standardized
coefficients are required. Keep in mind, however, that there are limita-
tions on the results. Having standardized our units we can no longer go
back to the original data, i.e., we cannot then say that a one unit
increase in education would result in an "X" number of units increase in
voting. Of course, as discussed in Chapter I in the section defining
political activity, different researchers use different measures of
political behavior. So even with unstandardized units it is difficult to
compare across studies. In addition to losing the ability to use the,
original data, the use of standardized units is dependent on sample results,
namely the standard deviations, and are not appropriate for comparisons
across samples because the path coefficients may change as standard devia-
tions change (Blalock, 1979, p. 482). These limitations apply to the next

chapter on results of the analysis.

CHAPTER IV

RESULTS

Although the primary interest is in the analysis of the causal
model there is some interest in descriptive results. These results
give a basis for comparing the present study with results of another
sample. An additional reason to present the descriptive results is
so that a reader can have the Complete data necessary to replicate or
extend the present LISREL analysis. The means and standard deviations
are presented in Table 3, percentages for categorical variables are
presented in Table 4, and the correlation matrix is presented in Table
5. It is only necessary to have the correlation matrix and the spec-
ifications of free and fixed model parameters and start values (see
Appendix B) for one to replicate this study. The means and standard
deviations, however, are necessary if one wishes to perform other
LISREL analyses such as those employing the covariance or moment matrixes.
This results chapter will focus on the results of the LISREL analysis.
Because problems were encountered in fitting the first model (the model
developed in Chapter II) a second model was restructured from the first.
This chapter will first review the criteria for acceptance of a good
fit of a model to the data. These criteria will be applied to the

first mode1--noting the fitting problems. The model will be

69

70

Table 3.--Means and standard deviations for variables used in the
measurement model

 

 

 

Variable Mean Standard Deviation
Years living in community 19.82 ' 18.41
TV exposure (minutes per week) 1272.62 1111.39
Radio exposure (minutes per 984.89 1329.89

week)
Daily/Sunday newspaper use 220.53 239.70
(minutes per week)
Weekly newspaper use (minutes 13.05 29.13
per week)
Movie use (number per month) .69 1.47
Book use (number read per month) 2.59 6.80
Trade/Professional journal use 29.90 99.35
(minutes per week)

CB radio use (hours per week) 1.34 21.84
Interpersonal index (persons 18.79 84.16
talked with per day) '
Sense of Community Scale 27.06 4.28
Voting index 1.78 1.28
Political participation index .63 1.01
Education 4.20 1.55
Newsletter use* .40 .50
Church bulletin use* .57 .50

 

*
lsdo use; O=do not

71

Table 4.-Percentages for categories of variables used in the
measurement model

 

 

 

Variable Z (N)
Education: (1813)
Less than eighth grade 5.1
Thru eighth 7.9
Some high school 15.5
High school diploma 32.4
Some college 22.7
College degree 9.5
Postgraduate work 2.9
Graduate degree 4.0
Newsletter use: (1826)
Use 40.0
Do not use 60.0
Church bulletin use: (1827)
Use 57.0

Do not use 43.0

 

72

 

.md .oH .m~
ooo~ oHN oq~
ooo~ ofim
oooH

.efi
ofio
m_ol
moo
oooH

.m~ .N~
oNo who
mNoI new
mgo me
flool ofiol
ooo~ HNoI
ooofl

.-
oNN
mofi
Mao
woo
woo:
oqd
ooo~

.o~ .o
and omo
mmfi goo
omo moo
coon mmo
ooo dfio
cad goon
owu N¢~
ooofi wmo
ooo~

.o
~m~
no“
omo
oool
ooo
«fig
omfi
«ed
coo
ooo~

.n
«no
omfil
who!
ooo
moon
moon
mmo
Nmo
Nmo
m-
ooou

.o
ooH
onH
ca“
omol
ofiol
ooo
Nmo
moo
NNo
m-
nmol
oood

.m
«cg
NmN
Nod
omOI
-ol
omd
ooo
oco
ooou
on“
«no:
woo
ooo~

.c
mmo
moan
anon
mfiol
woo
mmol
mNo
omo
coo
ono
«ea
fioo
Nool
ooo~

.m
cgol
anon
omol
moon
ado
QMOI
moo
mnol
N~o
moo
ngI
ooo
moo
moo
ooo~

.N
moon
mom
~m~
mmo
moon
meg
ofiol
coon
who:
mmOI
oHNI
«no
cos
m-
hdo
oooa

.H
com xooow coaumawoauumm
mNN amass waauo>
mmo moansaaoo mo mmewm

«Non noose HmaomumououaH

wfio on: no
moo on: cfiuoaasn souseo
owu mm: umuuwamamz
oom om: amou30n moons
mmd mm: xoom
mnm mm: onaummmz
mud on: oa>oz
NNo mm: sesame
mfifi om: amoasm\mafimo
mmo mnemoexo oaomm
neon musmoexo >8
ooNI mucooumou mums»
ooo~ coauoosom

.n~
.o~
.md
.qd
.m~
.NH

0 o
O H
H I-Ii

ammomov‘aoax

 

 

Aomuuwao muewoe Hmaauoov mmaomqum> Hoooa uaoEmHSmmmE Ho

xauuma defiumamuuooll.m wHAmH

73

restructured and these criteria will be applied to the second model,'
noting how it fits the criteria more successfully than the first model.

Having established an acceptable fit of the second model the
hypotheses will be tested with the path coefficients of the second model.
Finally, other results not expected in the hypotheses will be discussed.

Remember, the criteria for evaluating the fit of a model discussed
in the previous chapter are:

1) Is the model correctly specified?

2) Is the X2 nonsignificant?

3) Are the first order derivatives of fixed parameters zero?

4) Are the residuals as small as possible?

5) Is the explained variance as high as possible?

6) Is the standard error of coefficients low enough to
provide statistically significant estimates?

Results of the First Model

 

As mentioned above, the results of the LISREL analysis of the
original model indicate that the model requires extensive restructuring.
A portion of the results of this analysis is presented in Table 6. Only
a portion of the results is presented for two reasons: 1) the results
will not be used to test the hypotheses; and 2) the results in the
table are meant to illustrate the problems of the model described in

the text.

The First Model and the Six Criteria
The model is assumed to be correctly identified since there was no

error statement from the program.

74

Table 6.--Some of the LISREL estimates for model one (Figure 6b)

 

 

 

Standardized
Parameter Coefficient t—value Residual Variance
A .01 .29 ' c .76
1 5
A .10 .30 C .81
2 6
A .26 .30
3
1 .16 .30
L.
A .17 .30
S
A .61 .30
6
A .18 .30
7
-B .01 .17
41
-8 ' .05 .27
51
-B .22 .29
61
'8 .04 1.91
1.3
- -.02 .85
53
”B .01 .63
63

 

x2-999; df=101; p<.0001

2 B
voting

2
Rpolitical participation '

.24

19

75

The X2 is 999; df = 103; p (.0001. Degrees of freedom are
calculated as: (n(n + 1)/2) - t, where n is the number of input variables
and t is the number of free parameters in the model.

The X2 does not indicate a good fit, but this may be due to sample
size. Calculating on the basis of a sample size of 200 would yield a
x2 of 109 which is p >.05.

The first order derivatives for the fixed parameters are zero,
rounded to three decimal places.

The above three criteria are acceptable for a good fit of the data
to the model. However, it is the last three criteria that demonstrate
the problems with the model.

An examination of the residual matrix reveals several values in
excess of t .20. In addition, the ratio of the mean absolute residual
to the mean absolute correlation- is .403 (.0349/.0865). This is somewhat
higher than the criterion of .33 taken from Maruyama and McGarvey (1980),
and indicates residuals that are relatively large compared to the
correlations.

The explained variance in the two political activity variables is
relatively low, explaining 24 percent and 19 percent of the variance in
these two variables (see Table 6). Only one R2 comes close to the lowest
explained variance of Acock and Scott (l980)--24 percent compared to
their lowest 25 percent.

The major problem is that the model offers little explanatory power
because of relatively large standard error, indicated by the lack of
significant predictors and low t-values. In Table 6, coefficients and

tevalues for several selected lambda and beta coefficients illustrate

76

the problems. All of the indicators of mass use are nonsignificant.
And neither mass use nor interpersonal use have significant paths to
community integration, voting or political participation.

While the first three criteria are acceptable, the lack of accept—
able results on the final three criteria--criteria that indicate the
usefulness of the model in predicting the original data as well as
political activity-~mean that a restructuring of the model is in order.
Now one needs to find the problem areas in the model that could be

improved through restructuring.

Problem Areas in the First Model

 

An examination of the residual matrix (predicted minus actual
correlations) reveals several areas where prediction is considerably off,
several reSiduals in excess of i .20. There are confounding effects in
the mass media variables. Radio, movie and daily/Sunday use have several
high residuals, made especially important because these three have high
residuals for their relations with voting. Residence is also a problem
variable having high residuals across several variables, especially
with voting (.28). Some of the problems encountered in the model show
up clearly if one reexamines the correlation matrix (Table 5).

First, there is a general pattern of negative relations between TV,
radio exposure and media use and voting, and positive relations between
newspaper and magazine use and voting. In addition, there are very weak
relations between the electronic and print variables while those relations
are stronger among themselves. For example, TV and radio exposure

correlate with weekly paper exposure at .01, but weekly use relates to

77

daily exposure and magazine use at .068 and .115, respectively. This
suggests a need to restructure the mass media variable in the theoretical
model. A variable that assesses electronic mass media use and another
that assesses print mass media use is needed. This restructuring is not
meant to provide generic definitions for "electronic" and "print" media.
The results merely indicate separating the mass use variable into two
variables with separate indicators that are labeled "electronic" because
TV and Radio are transmitted via the radio spectrum and "print" because
newspapers and magazines are printed. In addition, book use may be a
confounding factor. Only one of its residuals is as low as .01, and it
has a relatively high standard error. Book use should be dropped as an
indicator.

Second, restructuring of the quasidmass indicators is also indicated.
Newsletter and church bulletin use each had significant coefficients.
And these two, along with trade journal use are highly related to each
other. But the other indicator may have problems. CB radio use has the
lowest t-value of any indiCator in the analysis (.08). It also shows
virutally no relation with any other variable in the analysis, the strongest
correlation coefficient being 0.025 (with voting). This indicates that
quasi—mass use could be improved as a theoretical variable if CB use were
dropped as an indicator.

Third, the exogenous variable "length of residence" as measured by
years of residence in the community might improve the fit of the model
if it were allowed to vary as a cause of the two political activity
variables. This is supported by two things: 1) there are generally

high residuals for residence with other variables in the analysis,

78

especially for voting (.28); and 2) the correlation between years
residence and voting is high (.285) relative to the other correlations.
This means adding paths from length of residence to voting and political
participation.

Finally, the model can be improved by eliminating the theoretical
variable, interpersonal use, and its indicator, the interpersonal index.
It has virtually no explanatory power. None of its betas is significant.
Indeed, its correlations show every little relation with other variables
in the analysis, none of the relations in excess of .063. While other
theoretical variables can be improved by restructuring their indicators,
interpersonal use has a single indicator; it's problems lie in that
single indicator. This does not mean that an interpersonal theoretical
variable would have no use in the model. Just this particular measure
may not be valid.

The restructuring of the model does the following:

1) separates mass media use into two endogenous variables:
electronic mass and print mass;

2) restructures the indicators of electronic and print mass use;
3) restructures the indicators of quasi-mass use;
4) eliminates interpersonal use from the model; and

5) specifies path between residence and the two types of
political activity.

The new model maintains the general causal relations developed for
Figure 4, and it only eliminates one portion--interpersonal use—-of the
model in Figure 5. Restructuring the indicators takes into account
problems encountered when trying' to specify measurement of a theoretical

COIlS tI'IJC t o

79

While the fit of the model to the data may be improved with further
tampering, one must consider that it should at least approximate the
model originally developed. This position is a compromise between two
extremes: on one hand, Maruyama and McGarvey (1980) adhere to the strict
confirmatory nature of the modeling and would allow minimal tampering
with the model (p. 508), and, on the other, Isaac,.g£_§l. (1980) began
with all error variances fixed at zero and then proceeded to alter the
model step-by-step until reaching a reasonable solution (p. 203); this
still did not provide a significant chi-square.

The reconstituted model is specified in Figures 7a and 7b. Figure
7a shows the new theoretical (circles) and measurement (rectangles)
models. Figure 7b contains the complete model with errors specified and
in notational form. The paramaters are defined in Table 7. Again, a
correlation matrix will be used as the input matrix, containing only the
measurement variables used in the second model. This matrix-a reduced

' form of the matrix contained in Table 5-—is contained in Table 8.

Results of the Second Model

 

The second model provides a much improved fit of the data to the
model. This model is an acceptable fit and, therefore, will be used for
testing the hypotheses. It is an acceptable fit based on all the criteria
set up to test the fit of the model. Complete results of the analysis are

contained in Tables 9a and 9b.

The Second Model and the Six Criteria
Like the first model, the second was acceptable on the first three

criteria. It received no error statement; so one assumes that the model

80

 

 

 

NmocH

dogwoofiowuunm All Ifiouuumm

Hauahsaoe

soaumo

 
 

 

 

 

amoeH
wcauo>

 

 

 

 

 

caumaaam chance

on:

 

 

 

on:
uwuuoamzmz

 

 

 

 

whomomxm
Hmeusoo oomuh

 

 

 

 

03» Homes you Hoooa ueosousmmoe one anewuonoonu owmmmln.mm ousuwm

 

 

madam
_huficoeaoo
mo «meow

 

 

 

on:
wswumwmz

 

    

coauouw

 

on:

own:
uoaum

 

e

 

 

 

 

ounmoewm
Momma edema:

      

 

mosmowmom oucoowmmm
mo sum—Ho..— Ilv manor

 

 

 

condos—om IIW— cowumonomJ

 

 

ousmoexm
oaonm

 

 

 

 

 

musmoowm
>8

 

 

 

 

on:
mfi>oz

 

 

 

 

 

 

 

anamomxm woman
hnoc=m\%aamo

 

 

 

81

o.o amen Macao voxfim Hound no.“ cmcu Hocuo ooxam acumuuoew

 

Shiv

 

 

 

ullw.

 

 

 

 

o.o um ooxwm uouuo mo.a um ooxfim powwowosfim

Anecdufiafimoo now u w o moanwh ammo sowuouoe nuas HmooE ucoaousmmoa one Houwuouownyll.nn muamwm

ofiulullv

A
.51!

 

 

o~

M

 

 

)VV\ .3

m
P

m: I

r
l

1.8.!!!

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

N
IIIMK.I$Y

 

 

 

 

 

 

 

 

N
Axilmo

82

Table 7.--Parameters and theoretical and measurement variables for
model in Figure 7b

 

 

 

Theoretical Model Measurement Model
Parameter Variable Parameter ‘ Variable
EXOGENOUS:
E1 Education X1 Formal education
5 Length of residence X2 Years living in
2 community
ENDOGENOUS:
n Electronic mass media Y1 Radio exposure
1 use
Y2 TV exposure
Y3 Movie use
n Print mass media use Y4 Daily/Sunday newspaper
2 exposure
Y5 Weekly newspaper
exposure
Y6 Magazine use
n Quasi-mass media use Y7 Trade/professional
3 journal exposure
Y8 Newsletter use
Y9 Church bulletin use
n Community integration Y10 Sense of Community
7 Scale
n5 Voting behavior Yll 3-item voting index
n Political participation Y 4-item index

 

83

 

.q~ .m~ .N~ .au .oH .o
ooo~ ooNI moon mow ~m~ moo
oooH com mum mmo umo

ooo~ oHN oe~ ono

ooofi odm mew

ooofi who
ooo~

.o .h
mac! cool
ohm oou
oNN qm~
mo“ mw~
m- omo
oq~ ¢-
ooo~ ow~

ooofi

.o .m
mmol «mo
mmm «No
~m~ oofl
mm~ omfi
coo ogfi
«fig ooo
owm omo
woo woo
ooo~ meg
oood

.e .m
we“ m~ml
mug Mug
cod «no
Nmm mm~1
Nod who:
on“ moot
Omo moo
coo Nmo
onfi mad
moo nmol
ooofi «no:

ooo~

.N .H

“so aNl
“so- she
«so- mmo
amen mean
who- amo-
one- mmo-
moo wmo
meow owe
mac one
coo loo
mac Noe-
one- «as
cool has
cool

unmoamou mummy

coaumusom

anon“ coaumewofiuumo Hmoauqum
xooaa wafluo>

mamom hufiesaaoo mo mnemm

om: efiumaaon nouano

om: umuuwam3wz

wusmoaxo.
Hmcu50n Hmoofimmmwoum\momus

on: oawumwmz
ouswoexo nonsense: haxmo3

ousmoqxo
uwemom3wa hmoasm\hdwmo

on: ow>oz
wuomooxo >8

muamooxo owomm

.qH

.MH

.NH

.H~

.oH

 

 

Aowuuwao mucfioo Hmawomov on shaman :« mmaomauw> Haves ucmeuamme mo kahuna coaumflmuuooll.m manna

84

Table 9a.--LISREL estimates for the theoretical model in Figure 7b
(t-values in parentheses)

 

 

Coefficient Unstandardized Standardized Residual Variance

 

* ‘ *
¢21 -.21(-9.39) -.21 :1 .92(4.73)
*
yl .12(6.09) .29 c2 .74(4.51)*
*
72 .14(7.77)* .51 c3 .71(6.04)
* 'k
73 .23(11.67) .54 g” .70(10.58)
*
y“ .19(3.29) .19 t5 .08(O.69)
*
ys .02(o.22) .02 C6 .41(4.36)
ye .31(10.60)* .31
*
Y7 .39(1o.27) .40 w .16(2.35)*
2].
ye -.08(-1.83) -.08 131 -.01(-O.16)
*
w .35(5.13)
* 32
-3 -.80(—3.78) -.34 w .09(1.34)
1+1 * 65 .
-B -1.78(-4.29) -.76
51
-B .30(1.17) .13
61
-8 .49(1.44) .14
1+2
*
—B 1.27(2.69) .37
52 *
-B 1.12(2.50) .32
62
-B .71(3.25)* .30
#3 *
-B .84(2.80) .36
53 *
-B .95(2.96) .40
63
—e -.03(-o.35) —.03
51+
*
-e .19(2.45) .19
64

 

85

Table 9b.-- LISREL estimates for measurement model in Figure 7b
(t-values in parentheses)

 

 

Coefficient Unstandardized Standardized Residual Variance

 

A1 1.003 1.00 61 0.008
A 1.003 1.00 6 0.008
2 2
it
A3 1.008 .43 61 .82(19.92)
*
A“ .O8(1.12) .04 e2 .99(30.20)
* *
A5 .72(8.11) .31 :3 .91(26.58)
*
A6 1.008 .29 e“ .92(28.30)
it
A7 .65(5.13) .19 e5 .97(29.46)*
A8 2.18(7.89)* .62 £6 .62(12.12)*
*
A9 1.008 .42 e7 .82(25.06)
* *
110 1.10(10.82) .47 e8 .78(23.31)
* *
A11 .66(8.00) .28 e9 .92(28.46)
A .778 .76 e .418
12 10
A .858 .84 8 .28a
13 11
A .588 .58 8 .668
11+ 12

 

aCoefficient fixed by program, t-values not appropriate

*
p<.05
x2-497.79; df=60; p<.05

2
Rvoting
2

Rpolitical participation.°

.92

59

86 .

is correctly specified. And the first order derivatives for the fixed
parameters are zero (rounded to three decimals). Although the x
decreased to 498, degrees of freedom also decreased to 60 and the
probability level is still (.0001. But once again this is due to sample
size. A sample size of 200 in this case would result in a :x2 of 54 and
p >.05. The test of the relative improvement in the fit of the model is
based on the change inpx2 from model one to model two. That change is
999 - 498 8 501 with change in degrees of freedom of 103 - 60 - 43. This
is a significant change in X2 (p <.05) which indicates a genuine improve-
ment in the fit of the model to the data rather than a chance variation.

It is the final three criteria, though, that conclusively demonstrate
that the new model is a major improvement and a good fit of the data.
There is considerable improvement in the residuals. Only eight residuals
are above i .10 and only one is above t .20. The mean absolute residual
is .035 and the mean absolute correlation is .097 giving a ratio of .35,
lower than that of the first model and close to that of Maruyama and
McGarvey's .33.

Perhaps the most impressive change is for the explained variance
in the political activity variables. Explained variance has increased
dramatically to 92 percent and 59 percent for voting and political
participation, respectively.

Finally, this is a model that has useful coefficients. All except
one of the free indicators and most of the path coefficients are sign-
ificantly different from zero. This arises from increased t-values which
are related to a relative reduction in standard errors for the coefficients.

These results do not mean that there is no better model to fit the

data. In fact, there are still problems with the residuals associated

87

with years residence; almost half of the large residuals are for relations
with years residence. And changes in error specifications might yield a
better fit. But these changes would involve some major changes in the
theory. So for the present the second model is deemed acceptable.

Having succeeded in developing a relatively good fit for the overall
model, we can now turn to examination of the results presented by the
LISREL estimates. These results are contained in Tables 9a and 9b, and

the standardized estimates are entered into the model in Figure 8.

Tests of Hypotheses

 

Because we have eliminated interpersonal use as a theoretical variable
in the model several hypotheses or parts of hypotheses are not tested at
this time. They are: H2c, parts of H4a, H4b, H4c, H4d, H4e, H4f.
Hypothesis 2c is dropped completely. The rest are retained with the
elimination of reference to interpersonal use. For example, H4e use to
read: "Quasi-mass and interpersonal use are stronger predictors of
political participation than is mass use;" it will now read: "Quasidmass
use is a stronger predictor than is mass use." Because we have split
mass use into two parts, hypotheses that employ mass use should be read
as including both electronic and print mass use.

With sufficient variance explained by the model we presume that two
of the three conditions for causal relations: 1) time ordering as
proposed by the theory, and 2) elimination of other factors through a
sufficiently large amount of explained variance. Now we can test the
third, and final, condition, existence of a relation, by examining the

model coefficients.

88

Amo.ve« ”anemone an ooxfiu osao>wv mmuoauumo ANMmHA wmufivuaoeoum sues 03» Hoo0211.o ouswwm

fiqulv o~
. F @
gm.
* /
.ll . N , .
mo.u\®/uo 7v x Alma o

oq
.. 3.-
i

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

mooinv ‘ no. K
8.\v.. ..2.’\1\1\®I8o4 1v ax Alhoé
/lv8. ...a.\.\\
kmm wN.
.arv ‘ \ ..

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

89

ComparingVCoefficients

Some hypotheses deal with the relative strength of the causal
relations. These are important hypotheses since they attempt to show
not only relations, but which media use is more important for which type
of social function. Therefore, it is wise to discuss how these coeffic-
ients will be compared.

Where one coefficient is significant and another is not, the comparison
is easy, the significant coefficient is the stronger. (This has been done
by Acock and Scott, 1980, p. 68—69.) Acock and Scott proceed to point
out the difficulty of stating that one coefficient is greater than another
when both are significant. Measurement error may mean the two are the
same. Keeping this thought in mind, i.e., being cautious about stating
one coefficient being stronger where the coefficients are both significant
and about the same absolute value, we will, nonetheless, treat the coeffic—
ient with the larger absolute value as stronger. Where the relative
strength of coefficients fits the theory, the relation serves to provide
further support for similar research findings. (With such appropriate
caution, this is how Maruyama and McGarvey, 1980, p. 510, treated

coefficients of similar strength.)

Predictors of Political Activity

 

Hla to Hld
These four hypotheses deal with media use as apredictor of political
activity. The first is

Hla: Media use (print mass, electronic mass and
quasi-mass) is a direct cause of voting.

90

All three are significant predictors of voting. Print mass use and
quasi-mass use are positive predictors {-8 = .37 and .36, respectively),
but electronic mass use is a negative predictor (-B = -.75). The null
hypothesis of Hla is rejected.

The second hypothesis states:

Hlb: Media use is a direct cause of political
participation.

Two of the three types of use are significant: print mass use (-8 = .32)
and quasi-mass use {-8 = .40). However, electronic mass use is non-
significant (- B- .13). The null hypothesis cannot be rejected for Hlb
as far as electronic mass use is concerned, but itis rejected for print
mass and quasi-mass.

The next two hypotheses deal with the relative strength of one type
of media use over another in predicting political activity:

ch: Mass use (print and electronic) is a stronger
predictor of voting than is quasi~mass use.

Hld: Quasi-mass use is a stronger predictor of
political participation than is mass use.

All three types of media use are significant predictors of voting. Elec-
tronic mass use has the largest absolute coefficient, but it is negative.
Print use and quasi-mass have virtually the same coefficient. We cannot
reject the null hypothesis for ch. However, it is clear that electronic
mass use is the strongest predictor, and the fact that it has a negative
sign while print use is positive suggests that this hypothesis needs
some revision.

For predictors of political participation there is a problem between
print and quasi-mass use. Both have significant predictors (-8 - .32

and .40, respectively), but quasi-mass use is only slightly higher than

91

print use. Electronic mass use has a nonsignificant coefficient

(-8 - .13). Since quasi-mass is clearly stronger than electronic

mass use andslightly stronger than print mass use, we will reject the
null hypothesis for Hld, but we will keep in mind the closeness of print

and quasi~mass.

H2a and H2b
These hypotheses predict that community integration is a cause of
political activity:
H2a: Community integration is a direct cause of voting.

H2b: Community integration is a direct cause of
political participation.

The coefficient between community integration and voting is nonsignificant

(~8" .19). The null hypothesis for H2a is retained while that for H2b

is rejected.

H3a and H3b
I These hypotheses predict that education is a cause of political
activity:
H3a: Education is a direct cause of voting.

H3b: Education is a direct cause of political
participation.

The coefficient between education and voting is significant (Y’s .19),
but the coefficient between education and political participation is
nonsignificant (Y'- .02). The null hypothesis for H3a is rejected while

the null for H3b is retained.

92

Predictors of Other Endggenous Variables

H4a to H4c
These three hypotheses predict relations between education and
the media use variables:

H4a: Education is a direct cause of mass media use
(electronic and print).

H4b: Education is a direct cause of quasi~mass use.
(NOTE: H4c was dropped because it only refers to interpersonal use.)
Education is a significant predictor of all types of media use in the
model: electronic (V'- .28), print (V'8 .51) and quasidmass (Y'= .54).

The null hypotheses are rejected for H4a and H4b.

H5a to H5c
This final set of hypotheses deals with predictors of community
integration:

H5a: Length of residence is a direct cause of
community integration.

H5b: ‘Media use (electronic, print and quasidmass) is
a direct cause of community integration.

The null hypothesis for H5a is rejected since the coefficient for length
of residence is significant (‘1‘ .31). There are mixed results for

H5b. Both electronic mass and quasi~mass are significant predictors,

but electronic is a negative predictor (-Y = -.30) while quasi—mass is an
equally strong positive predictor (-V'- .30). Print mass use is non-
significant (r”r= .14). For electronic and quasi-mass use the null

hypothesis for HSb is rejected. It is not for print use.

93

The final hypothesis predicts quasidmass to be a relatively
stronger predictor of community integration:

H5c: Quasi-mass use is a stronger predictor of
community integration than is mass use.

As predictors of community integration, quasi-mass is the strongest
positive predictor. Print has a nonsignificant coefficient. But the
absolute strength of electronic mass use is the same as quasidmass,
only its sign is negative. If we are looking for the strongest positive
predictor of community integration we would reject the null hypothesis.
Remember that mass use was expected to reduce trust in the community and

have a negative effect which it did.

Other Results

Besides tests of hypotheses there are other results that need to be
explored.

Only one of the indicators of any type of media use is nonsignificant,
and it is by far the weakest indicator: TV exposure (Y'B .04). This
indicator is almost all error (e = .99) suggesting that it is relatively
useless as an indicator of electronic mass use.

In the second model length of residence was allowed to predict the
political activity variables along with education even though there was
no hypothesized cause. Interestingly, both variables were significant
predictors of voting, but both were also nonsignificant predictors of
political participation.

Examination of the correlations of the residuals (ws) also leads to
some interesting conclusions. Correlated error between the two political

behavior variables is nonsignificant, and most of the variance in political

94

participation and virtually all in voting (given measurement error)
is explained by the model. This suggests that indeed we have tapped
independent variables. This supports the theoretical notion that
voting-~a private mass cultural phenomenon—~15 different than other
types of political activity that involve public action and more
interpersonal commitment.

The covariance of errors between electronic mass and print mass
is significant. So is that between print mass and quasi-mass, while
that between electronic mass and quasi-mass is not. This indicates
that there may be some underlying, untapped relation between print and
electronic and between print and quasi-mass. This underlying relation
is not the same across all three since the correlated error between
electronic and quasi-mass is nonsignificant. Perhaps the relation is
the "massness" between print and electronic and the "printness" between
print and quasi-mass (all three quasi-mass indicators are print oriented).
It would be interesting to see if errors would correlate had quasi-
mass included several electronic media.

One must also note the large measurement errors associated with
the indicators of media use, especially electronic. The larger measure-
ment errors may be associated with the relative precision of the measures.
Remember that the coefficients are measures of reliability (Acock and
Scott, 1980; Allen, 1981), and that measures with higher precision are
generally accorded lower reliability estimates (Woelfel and Fink, 1980,
p. 91). The electronic and print measurement indicators were measured
in minutes or in numbers seen or read while church bulletin and news-
letter use were measured as simply did or did not. Those more precise

measures, to some extent are accorded lower reliability, i.e., lower

95

coefficients in the model.

Indirect Effects

 

Indirect effects are calculated by multiplying the standardized
coefficients for the paths of interest (Acock and Scott, 1980,. p. 69).
Most of the indirect effects are negligible, providing coefficients
less than .07. However, a few of the paths draw our interest. These
are all indirect effects of education on political activity through
media use.

The indirect effect of education on voting is as strong as its
direct effect. Through print mass use it is .19 (.51 x .37 = .16)ithrough
quasi-mass it is also .19 (.54 x .36). However, the indirect effect
through electronic mass use is now negative (.28 x -.75 I -.21).

Overall, then, education is a positive cause of voting even when mediated
through print and quasi-mass use. But it can rebound to a negative
effect if mediated through electronic mass media.

While there was no direct effect of education on political partici-
pation, there was an indirect effect through print and quasidmass media
(.51 x .32 - .16 through print use; .54 x .40 = .22 through quasi-mass

use).

Summary of Results

 

The original model proposed in Chapter 11. proved to be a poor
fit of the data. A second model provided an adequate fit, requiring,
however, alteration of the model: eliminating the interpersonal use

variable and splitting mass use into electronic and print mass use.

96

Based on the LISREL analysis of the second model the following
results were obtained: 1) mass media use is generally a better
predictor of voting while wuasi-mass use is a better predictor of
political participation and community integration; 2) education is a
significant predictor of voting but not of political participation;

3) community integration is a significant predictor of political
participation but not of voting.

Generally, the media use variables were significant predictors of
community integration and political activity. However, electronic mass
use was a negative predictor of voting and community integration.

In results of the model not related to the hypotheses, there was
a large amount of measurement error. In addition, correlated error
suggested that there are underlying factors across the media use variables.
However, the lack of correlated error between residuals of the political
activity variables indicates these variables are relatively independent.

Indirect effects indicate that the media use variables are useful
as mediators of the effect of education on political activity. Indirect
effects through the media on voting were as strong as the direct effect
of education on voting. Further, though there is no significant direct
effect of education on political participation there was an indirect
effect through print and quasi-mass use. This indirect effect was
as strong as or stronger than some other significant direct effects.

The implication of these results will be discussed in the next

chapter.

CHAPTER V

DISCUSSION

Having completed the analysis and tested the hypotheses the
task now turns to an understanding of the implications of these
results. This chapter will approach the results in the same order
as did the theoretical chapter (Chapter II). First, we will explore
the causes of political activity dealing separately with mass use,
quasi-mass use, community integration and education as causal predictors.
Second, causes of these predictor variables, media use and community
integration, will be discussed. Third, other results and the general
role of the media will be discussed. Fourth, suggestions for future
research will be presented. And finally, a concluding statement will
focus on media use effects as part of a general social process, and the

importance of including quasi-mass media in that process.

Causes of Political Activity

Mass Media

 

The conclusion about the effects of mass media use on political
activity in Chapter II was that mass media could be regarded as positive
predictors of voting and political participation, but that their role was

one of reinforcement, keeping the loyal loyal, reinforcing already held

97

98

political party or candidate biases. This general effect would be one
where mass media are useful for promoting already held mass cultural
beliefs in the population.

In this study print mass use was the positive predictor of both
types of political activity, while electronic mass use was a negative
predictor of voting and nonsignificant as a predictor of political
participation. Only print mass use supports the positive effect of
mass use on political actitivy. But this does not mean that the original
theoretical notion is incorrect. Remember that many of the studies of
the relation of media use to political activity used a different measure
of media use than did the present study. Jackson-Beeck (1979), for
example, asked respondents whether they had used TV and newspapers for
political information; the present study ignored the purpose for which
the media were used and focused solely on exposure. Remember also that
the question of which was a cause of political behavior, exposure or
purpose, was resolved tentatively on the basis of two studies only one
of which was a causal analysis.

How does one reconcile the difference between previous studies and
the results of the present study? How does mass media use--even TV use--
produce a positive effect on political activity in one study and a negative
effect in this study? On the surface it would seem simply that the
difference is related to the measure one uses for media use. But this is
simplistic, and does not offer a synthesis of the two results that one
might apply as a single construct in future research. In order to find
a way to explain these differences we must explore what could be
different about exposure to electronic media when compared to the print

media.

99

One obvious difference is that the electronic media used in the
present study may be considered more entertainment oriented while the
printed press and magazines are news and information oriented. This
might lead us to conclude that it is the use made of a medium that
determines its effect on political activity. TV use has been a positive
predictor of voting when it was measured as a source of campaign infor-
mation (See Jackson-Beeck, 1979). A recent study of children's social-
ization toward political activity by Conway,_g£_al., (1981) tried to
separate mass use--the measure was the news purpose to which the medium
was put--into print and electronic components. They were unsuccessful
in separating mass use, finding the two components too intercorrelated.
But they also found mass use was a positive predictor of political
socialization. Luckily, Conway,_§§_gl. did a path analysis that included
reciprocal paths between media use and political knowledge. They found
that both paths were significant predictorsrflfpolitical participation.
This study, then, supports the few other studies that show that to a great
extent political commitment leads to specific uses of media rather than
specific uses being the major cause of political behavior. Keep in mind
also that in the present study the measure for print was the same as for
electronic: exposure. The difference was discovered even though the
intent of use was not incorporated into the measure. But content can
still be an important factor if intended use is not the relevant factor,
but instead incidental learning through simply being exposed to a medium
underlies the effect.

McPhee (1963) notes the existence of "natural learning" from mere
exposure. He described a test of learning announcers' names from exposure

to a radio program. It would take nearly forty exposures for the majority

1 1“ .....I‘ll“

100

of the population to know all the names. So in terms of incidental
acquisition of attitude or information, exposure is a major component
and should not be ignored.

The results for mass electronic and print use suggest that exposure
is a useful explanatory component, but that the content difference of
various media may lead to different effects. These effects would not be
realized without exposure, and incremental exposure would have incre-
mental effects (not necessarily linear).

This leads to a theoretical position that is somewhat different
than positing either exposure as the only relevant measure or intended
use as the only relevant measure. The relevant measure would include
elements of exposure and content. But to say that print has only one
content orientation would be misleading. This study used print variables
that are generally thought of as information oriented, newspapers and
magazines. Other mass print media, comic books or mystery novels, may
not have the same orientation and may not produce the same effects on
political activity. Likewise, the electronic media considered in this
study were entertainment oriented, but the same effect may not have been
produced had the electronic medium been teletext.

Mass media then are still to be considered causes of political
activity, but the positive or negative effect is determined by the infor-
mational nature of the medium rather than the use to which the medium
is put. From a measurement perspective media use should be an amalgama-
tion of a content component and an exposure component. There are still
other perspectives on the factors underlying the effects of media use.

These will be considered along with the general roles of the different

101

types of media use after discussing other predictors of political

activity and predictors of community integration.

Quasi-mass Media

The posited effects of quasi-mass media use on political activity
were based on the more personal nature of quasi-mass media use, i.e.,
since quasi-mass allows more personal interaction it should be more
useful for the type of political activity requiring more personal
involvement. While this notion was supported for political participation,
quasi-mass also proved to be a strong positive predictor of voting,
almost as strong as print mass use.

If all types of political activity are considered social phenomena
then these results support the theoretical notion that use of quasi—
mass media enhances social interaction and thus encourages political
activity. Notice that this differs from the reason print mass use had
a positive effect. For the mass media the positive effect may be from
the information acquired due to exposure to the media whereas for quasi-
mass media the effect is from increased social interaction.

The relative usefulness of mass versus quasi-mass media use for
each of the two types of political activity helps make the distinction
between the reasons for effects from the two types of media use more

clear.

Mass vs. Quasi-mass Effects

 

Mass use was originally hypothesized to be a stronger predictor of
voting than quasi-mass while the opposite was proposed for political

participation. This was based on theoretical differences in the nature

102

of these media. Mass media were seen as more distant, less interactive,
and, therefore, more useful for the type of political activity requiring
the least social interaction: voting. Quasi—mass, however, offers more
interaction and is more useful for political participation.

While the results are mixed because of the difference in effects
by electronic mass and print mass and the closeness of the coefficients
for print mass and quasidmass, taken as a whole the results support the
poisted differences in effects of mass and quasi-mass media use. Even
though negative, electronic mass use was the strongest predictor of
voting, and print use had a slightly higher coefficient than quasi-mass.
For political participation quasi-mass had the highest coefficient,
clearly larger than electronic mass which was nonsignificant, and slightly
larger than the print mass coefficient.

These results fit the general theoretical notion that voting is
a private, mass phenomenon, requiring little personal commitment while
political participation requires some public display of one's political
beliefs. Predictors of these political activities should correspond to
the personal commitment requirements of these activities, i.e., stronger
predictors of voting should be those involving less personal interaction
while predictors of political participation should offer more oppportun-.
ities for such interaction. Such was the case in this study: the mass
use variables--those with less interaction opportunities-dwere better
predictors of voting, and quasi—mass--an indicator of more personal involve-
ment—dwas a better predictor of political participation.

The effects of the different types of media use on social phenomena
can be stated more generally. But this will have to wait until the
results of the effects of media use on community integration are

discussed, below.

103

Community Integration

 

As an indicator of how much one was involved in the community,
community integration was hypothesized to be a direct cause of both
types of political activity. It turned out to be only a significant
predictor of political participation.

If one treats political participation as requiring more social
interaction, as we have in the discussion so far, then it follows
that community integration ought to be a better predictor of political
participation than it is of voting. This is what the results support.

This supports a generalization that social phenomena requiring more
personal interaction are better predicted by variables that allow for
such social interaction. This generalization is discussed, below,

where the general role of media is explored.

Education

The generalization made by Kraus and Davis (1976) that education
was the only consistent demographic predictor of political activity was
taken as the theoretical perspective of this study. The results showed
that indeed education was a significant cause of voting. But other
results confound the strict limitation of education as the only predictor
or as a predictor of all political activity.

Education was not the only significant demographic predictor of
voting. Length of residence was a significant and stronger cause of
voting. Notice that the relation between education and length of
residence is significant although negative (4>= -.21). There is
obviously a lot of covariation between these two variables. This

relation might be explained through a third component. Remember that

104

in predicting community integration we selected residence over age which
also had been shown to be a strong predictor of community integration.
Perhaps for purposes of a process model age ought to be included. This
might help explain variance in both education 33g length of residence.
Still length of residence cannot be ruled out as a cause of political
activity. It just never appeared in previous literature because the
researchers concentrated on traditional measures of socio-economic
status: age, education, income and occupation. The least that is
suggested here is that other demographic predictors should be explored
as possible causes of political activity.

But education still has an effect on political participation
even though its direct effect is nonsignificant. It has an indirect
effect on political participation through media use. The indirect effect
through quasi—mass media was stronger than the direct effect of educa~
tion on voting. Further, the indirect effect of education on voting
was stronger than its direct effect. This underscores the importance
of including media use in a model of demographic effects on political
activity. The direct effects of education may not be as important as
the indirect effects. This leads to a proposed model of the effects

of education on political participation and voting:

————___________————-4§ voting
Media ””’)7

Education ———9 Use \
Political Participation

Although this model has been presented by Kraus and Davis (1976)
for general political activity no one has noted its importance for

political participation where education has no direct effect.

105

Causes of~Community Integration

 

Although it was hypothesized that length of residence would be a
direct cause of community integration--which it turned out to be—-the
primary interest was in how different types of media use would affect
community integration. From the idea that communities are social systems
comes the theoretical notion that those media that allow for more social
interaction would possess the greatest potential to be useful community
communication channels. Indeed, it was also discussed that the intro-
duction of mass media into a community can actually reduce feelings of
belonging. The results for this part of the model show the clearest
distinction among media types. Quasi-mass was the only significant
positive predictor. Electronic mass use was also a significant
predictor but it was negative. Print use was non-significant. This
fits with our theoretical notion that the mass media would be negative
predictors because they would engender less trust in the local commun-
ity. Vidich and Bensmen's (1977) explanationlcnf the mechanism for
causing this lack of trust is that the mass media are imported. If
the local-extralocal nature of the medium is a decisive factor it can
help explain the results. Electronic media should cause the least
trust since they are the least likely to be locally originated-movies
and network TV programs.for example, are imported; print mass should be
somewhere in the middle since some, newspapers, originate locally while
others, magazines, come from outside the community; and quasi-mass
should engender the most trust since church bulletins and many news-
letters originate locally, and since all three indicators of quasi-mass

are Open to input from members of the local community.

106

There is little to change in the theoretical notions about the
effects of media use on community integration. However, one should
note that electronic mass is a stronger negative predictor while
print mass just has no effect. This might be explained if we examine
the general role of media in predicting social phenomena and the further

perspectives one might have for measuring media use.

Measuring Media Effects

The Role of Media_

 

The roles for the media in this study can be stated generally
and simply: mass cultural phenomena that tend to maintain the anonymity
of the individual are more strongly predicted by mass media; phenomena
requiring more personal involvement and public scrutiny are more
strongly predicted by quasi-mass media. There is, however, a limit
to this generalization. Because the print and quasi-mass coefficients
were so close in a couple of cases, and because the correlated residuals
indicate another underlying component to media use, we must allow that
mass—quasi-mass distinctions are not the only important factors in

differentiating media.

Measuring Media Use

 

The fact that both print mass and quasi-mass measures contain a
print component could explain the similarity of their coefficients in
predicting the two types of political activity. One might be tempted
to believe that print versus electronic was the relevant predictor rather

than mass versus quasi-mass differences. One might then conclude that

107

print of all types (whether mass or quasi-mass) should be combined.
However, the effects of mass use on community integration argue against
this. For community integration quasi-mass is the stronger positive
predictor; print mass is nonsignificant. Keep in mind that error with
print mass is also correlated with the error of mass, and the error of
electronic mass use is not correlated with that of quasidmass.

Maybe what needs to be considered is the multidimensional nature of
media use. Perhaps it cannot be identified and measured merely on the
basis of its position on the mass-interpersonal continuum, or for that
matter on the basis of its print-electronic orientation (or purely on the
basis of its content, as discussed, above). A better measure for future
studies would combine two or more of these relevant dimensions: where
is a medium on the mass-interpersonal dimension? Where is it on the
electronic-print dimension? Where is it on an entertainment-information
dimension? Other relevant dimensions may be: local versus extra local
orientation, general versus specialized media. The grouping of indicators
of media might be far different if another dimension is added. One might
find, for example, that radio and church bulletins are closer than radio
and movies if a local-extralocal dimension is added to mass-interpersonal.

One might even venture so far as to let the population under study
determine the relative clustering of media rather than through a priori
(and somewhat arbitrary) definitions of the researcher. Woefel and
Fink (1980) have developed a method for determining the relative positions
of concepts in conceptual space. This method is called "Galileo."

One could let a sample of respondents determine the distances between
a set of media-TV, radio, newsletters, etc., and a set of other concepts--

mass. personal, local, print, etc., and let the clustering of media in

108

conceptual space determine how they will be clustered for the analysis.
The nearness of a cluster to the concepts then can help define the
relevant characteristics that make a set of media cluster together
and one set differ from another.

Whatever the ultimate relations discovered, the results of this
study argue for retaining the distinctions based on the mass-interpersonal
continuum, keeping exposure as a relevant element of measuring media use,

but looking for other components of their measures.

Future Research

 

Research on the effects of media use on political activity should
keep in mind the results of this study. First, future research should
treat communication and political activity as a social process. The
study of the relation of mass media use to voting, for example, should
not be observed in a vacuum, but in relation to other variables such as
those that have shown themselves to be useful in this study; quasi-mass
use, community integration and education. Second, the results of this
study should not be considered limiting but should suggest avenues of
exploration. Though this study used exposure as the basis of measurement,
it does not rule out other factors as bases of measurement. Inter-
personal media should be included in future studies to explore the full
range of media use. The multidimensional nature of the media use
variables should be incorporated into their measure. And, of course,
other indicators of the theoretical variables should be examined--this
study limited itself to print quasi-mass media.

Although not directly tested in the present study, effects of new

communication technologies can be inferred from the results. As discussed

109

in Chapter I, most of the new communication technologies arise from a
need for more specialized media or in-and-of-themselves are narrower

(as opposed to mass) uses of traditional mass media. For example,

cable television offers the opportunity for more specialized, more

local, more interactive channels than does traditional broadcast tele-
vision. If these media fall into the quasi-mass part of the communica-
tion continuum then one would expect their effects to be similar to those
found for the quasi-mass media in this study, i.e., they would be more
useful for promoting social or political activity that involves more
personal interaction. As these new technologies become more widespread
and as access to them increases future research should assess the effects
of these new media upon political activity along with the traditional
assessment of mass media effects.

In addition to the media use predictors, a complete model of
political activity needs to branch out beyond voting and voting-related
activities. Political activity such as protest and revolution has been
linked to differing uses of various types of media (See Reagan, 1981).

A complete model would include the full range of political activity.

Of course, a major drawback of this study, and of social science
research in general, is large measurement error. This is especially true
of the media use indicators in the present study. More precise measures
are generally considered desirable, but greater precision may alter the
relations in this model by actually increasing the estimates of error
since the model operates as if the coefficients of the indicators are
reliability estimates (Acock and Scott, 1980; Allen, 1981). The problem
encountered between reliability and precision is discussed by Woefel

and Fink’(l980, p. 91).

110

Miller (1978) has also criticized the measurement of interpersonal
communication as exposure to interpersonal incidents like face-to-face
encounters. He suggests that mere exposure is insufficient to measure
interpersonal communication. A chance encounter for a few minutes with
an acquaintance is not the same as a purposive meeting between friends.
This supports the suggestion made in the previous section that mere
exposure is not the only factor involved in media use.

A further criticism of using exposure as the measure of media use
comes from McLeod and O'Keefe (1972). They note that assessing media
exposure tells us nothing about the "why" of the exposure. But they
also note the shortcomings of other perspectives: gratifications
perspectives assume people know why they use media or will tell the truth
if they do know; categorical approaches are limited by the fact that
communication researchers cannot agree on what constitute relevant
categories. The approach best suited seems to be a combination of several
perspectives. But most research ignores this. See, for example, the
studies noted earlier which used various definitions of political part-

icipation and media use (Welch, 1977; Jackson-Beeck, 1979).

Conclusions

 

The Model as Process

 

While it is intriguing to look for simple relations between a few
variables--for example, to try to predict voting behavior on the basis
of mass media use--it makes more sense to look at a host of social
indicators that can lead to a host of behaviors. McLeod and O'Keefe
(1972) argue that controlling for social variables, as occurs in

experimental manipulations to test for communication effects on attitudes,

lll

artificially createsa situation that inflates the importance of the
observed variables relative to other possible causes. They propose
that communication studies take place in the "real" world, affected
by the presence of other intervening and coactive variables (as do
McPhee, 1963; and Chaffee, 1972).

This study has attempted, to some extent, to reflect that "real"
world by allowing the process of communication effects to take place
in a model that allows such coaction and that takes into account other
possible causes. The importance of doing this can be seen specifically
in the change in importance of the relation between community integration
and voting. The original correlation matrix (Table 5) shows that the
highest correlation is between sense of community and voting. Yet when
the LISREL analysis is performed the coefficient between community
integration and voting is nonsignificant. When other effects are taken
into consideration, as well as measurement error, what appeared to be a
clear relation has faded.

Likewise, education appears from the correlation matrix to be an
especially strong predictor of political participation with a correlation
of .266. Yet the analysis shows a nonsignificant coefficient between
education and political participation. But thus does not mean that
education has no effect on political participation since there is an
indirect effect through print mass media use and quasi-mass use.

These two examples illustrate the importance of specifying a process
model of communication effects. It helps us understand communication

within a field of interrelated social phenomena.

112

The Importance of Quasi-mass Media

 

This study demonstrates the importance for communication research
to include the study of quasi-mass media on its research agenda. Quasi-
mass is a useful predictor of social and political activity. Menzel
(1971) stated ten years ago that quasi-mass was a neglected area. It
remains so today.

As new technologies reshape the nature of our communication media,
transforming older broadcast television into specialized entertainment
channels along side local access channels and home computer networks,
and as we see expanded access to the inexpensive uses of print media-
posters, flyers, newsletters--for political party use, local neighborhood
association bulletins and political activist handouts, one cannot ignore

the possible impact this may have on our political arena.

APPENDICES

113

APPENDIX A
STUDY QUESSTIONAIRE

This is not the full questionnaire used for the Media Environment
Study. Only a portion of the 45 page questionnaire contained questions
used to measure variables for the present study. Therefore, representa-

tive questions only are provided.

Demographics
The following questions were used to assess demographic character-

istics, including sample description statistics:

 

 

 

 

 

 

 

 

First of all, how long have

you lived in the (CITY NAME)

area?
(EXACT RESPONSE OF RESPONDENT)

Do you own or rent this place Own ..................... 1
Rent .................... 2
Refused/Don't Know ...... 8

Are you married? Yes ..................... 1
No ....... ..... .......... 2
Refused ......... ..... ... 8

How much formal education have Less than 8th ..... ...... 1

you completed? Through 8th ............. 2
Some high school ........ 3
High school diploma...... 4
Some college ............ 5
College degree .......... 6
Postgraduate work........ 7
Graduate degree.......... 8
Refused/Don't Know ...... 9

 

114

 

Would you please tell me your
age?

 

(NUMBER)
Refused/Don't Know ......... 99

 

 

Here is a card with several categories of income on it. (HAND SHOW
CARD V TO RESPONDENT). Would you please tell me the letter corres-
ponding to the category which best represents the total annual
household income here?

 

 

A B C D E F G H I J K
01 02 03 04 05 06 07 08 09 10 ll

 

 

 

 

 

 

 

 

 

Refused/Don't Know ........ .. 88
CARD V
A $0 - $4,999
B $5,000 - $9,000
C $10,000 - $14,999
D $15,000 - $19,999
E $20,000 - $24,999
F $25,000 - $29,999
G $30,000 - $34,999
H $35,000 - $39,999
I $40,000 - $44,999
J $45,000 - $49,999
K $50,000 OR OVER
Sex: (BY OBSERVATION) Male ........ ............... 1
Female . ... ... .. . 2
Race: (BY OBSERVATION) White .. ................... . 1
Black ...................... 2
Oriental.. ................... 3
Hispanic. ................... 4
Other ...................... 5
(IF UNSURE OF RACE: ASK:)
May I ask your race?
Type of dwelling Apartment ................... 1
House..... .................. 2
Mobile Home ................. 3
Other 4

 

(SPECIFY)

 

 

 

115

 

(RECORD) Rural ......OOOOOO ...... 0.. 1
Within city limits ........ 2

 

 

 

Media Use

Television exposure was measured by asking respondents how much

television they watched on the previous day for several dayparts:

 

How much time did you spend watching
TV yesterday morning before 9?

 

 

(EXACT STATEMENT OF RESPONDENT)

 

 

This question was asked of the following dayparts: before 9:00 a.m.,
9:00 a.m. to Noon, Noon to 3:00 p.m., 3:00 p.m. to 6:00 p.m., 6:00 p.m.
to 8:00 p.m., 8:00 p.m. to 11:00 p.m., and after 11:00 p.m. The times
for these dayparts were summed to compute previous day's TV exposure.

For the previous day TV exposure the following questions was asked:

 

How much time did you spend watching
TV last Saturday (Sunday) before 9?

 

(EXACT STATEMENT OF RESPONDENT)

 

 

 

This question was asked for each of the same dayparts that were used for
previous day exposure, and the times for these dayparts were also summed
to compute previous weekend day's TV exposure.

Radio exposure was assessed using the following question:

 

How much time did you spend
listening to radio yesterday
from 6 a.m. until Noon?

 

(EXACT STATEMENT OF RESPONDENT)

 

 

 

This question was asked for the following dayparts: 6:00 a.m. to Noon,

Noon to 6:00 p.m., 6:00 p.m. to midnight, and midnight to 6:00 a.m.

 

116

of the present day. These times were summed to compute previous day's
radio exposure.

For the previous weekend day the following question was asked:

 

How much time did you spend
listening to the radio last
Saturday (Sunday) from 6 in the
morning until Noon?

 

(EXACT STATEMENT OF RESPONDENT)

 

 

 

The same dayparts were used as in assessing previous day's radio exposure.

These times were summed to compute previous weekend day's radio exposure.

Time reading a daily newspaper was assessed with the following

question:

 

How much time did you spend
reading a daily newspaper?

 

(EXACT STATEMENT OF RESPONDENT)

 

 

Previous Sunday newspaper use was assessed with:

 

How much time did you spend
reading Sunday newspapers
last Sunday?

 

(EXACT STATEMENT OF RESPONDENT)

 

 

 

Weekly paper exposure was assessed with the following questions:

#1

 

Do you read any newspapers that Yes ................ 1
only come out once or twice a No ................ 2
week or every two weeks? Don't Know/Refused.. 8

 

(IF YES) How much time did you
spend during the past week

 

reading these papers? (EXACT STATEMENT OF RESPONDENT)

 

 

 

117

Magazine use was computed by counting the number of magazines-~
both news and other magazines-reported in the following series of

questions:

 

Do you read any weekly news —-Yes................... 1

magazines? No ................... 2

Don't Know/Don't
Remember............. 8

 

 

Which ones? (CIRCLE ALL MENTIONED) Time ................. l
Newsweek l
U.S. News and World
Report ..............
People ...............
Us ...... ....... ......
Other -

‘F+Fikd

 

 

(PROBE: Are there any others?)

 

Do you read any other magazines Yes ..................
regularly? No ...................
Don't Know ...........

QNH

 

[IF YES) Which ones? (ASK: Any others? UNTIL No MORE MENTIONS)
( )
( )

 

 

 

Book and movie use were assessed with the following questions:

 

How many books did you read in
the last month?

 

(EXACT STATEMENT OF RESPONDENT)

 

How many movies did you go to,
in theatres, last month?

 

(EXACT STATEMENT OF RESPONDENT)

 

 

 

118

Quasi-mass media use was assessed with the following set of

 

questions:
Do you read any professional r—aYes................. l
journals or trade magazines No ................. 2
that have information about ' Don't Know/Refused/
your job? Not Applicable ... 8

 

 

How much time did you spend
during the past week reading
professional journals or

 

 

trade magazines? (EXACT STATEMENT OF RESPONDENT)
Do you read any newsletters; by Yes................. 1
newsletters we mean anything No ................. 2
printed that carries information Don't Know/Refused.. 8

about organizations or associations,
such as labor unions, professional
associations, neighborhood groups,

 

 

etc.?

Do you read any ChurCh bulletins Yes I 0 O O O O O O O O O O O O O O O l

or religious newsletters? No ................. 2
Don't Know/Refused.. 8

Do you have a citizens band-- Yes................. 1

CB--radio? No ................. 2

Don't Know/Refused.. 8

 

How many hours did you use
your CB last week?

 

(HOURS ROUNDED)

 

 

 

Interpersonal use was assessed with the following questions:

 

How many people did you talk with
in person yesterday for more than

 

a couple of minutes? (NUMBER)

 

 

 

119

 

About how many personal, non-business
telephone calls did you make yesterday
to friends or relatives?

 

(NUMBER)

 

 

 

Sense of Communigy

 

The sense of community scale was presented to respondents in the

following manner:

 

I am going to read some things a person might say about their
community. Please indicate how you feel about the statement,
whether you "strongly agree," "agree," are "undecided,"
"disagree," or "strongly disagree" with the statement.

Strongly Strongly
Agree Agree Undecided Disagree Disagree

 

 

I feel a part of
this community 5 4 3 2 l

I do not belong in
this community 5 4 3 2 1

I feel comfortable
living in this
community 5 4 3 2

I help my neighbors 5 4 3 2

I do not know very
many people in this

 

 

 

community 5 4 3 2 1
I talk to my neigh-
bors a lot 5 4 3 2
I know my neighbors 5 4 3
Voting

The following three questions form the voting index:

 

Did you vote in the 1976 presidential Yes................. 1
election? No 00.000.000.000... 2
Don't Know/Refused.. 8

 

 

 

120

 

 

Did you vote in the last congressional Yes................. 1
election in 1978? No ................. 2
Don't Know/Refused.. 8
Did you vote in the last local Yes....... ..... ..... 1
election? . No ................. 2
Don't Know/Refused.. 8

 

 

 

Political Participation

 

The following questions from the four-item political participation

 

 

 

 

index:
Have you ever tried to get people Yes................. 1
to sign petitions to get an issue No ................. 2
on the ballot? Don't Know/Refused.. 8
Have you ever tried to get people Yes................. 1
to sign petitions to try to get a No ................. 2
party or candidate on the ballot? Don't Know/Refused.. 8
Within the last two years have you Yes................. 1
given money, helped canvas or No ................. 2
attended rallies for the Democrat Don't Know/Refused.. 8
or Republican party?
Within the last two years have you Yes................. 1
given money, canvassed or attended ~ No ................. 2
rallies for any other political Don't Know/Refused.. 8
party, independent candidate or
issue?

 

 

 

121

APPENDIX B

LISREL MATRIX SPECIFICATIONS FOR FIRST MODEL

In the following matrixes coefficients marked with a zero or a

superscript "a" are fixed values. Other values indicate free parameters.

These values are the start values for the LISREL analysis.

 

AY - Etna o 0 0 0 0
.131 0 0 0 0 0
.219 0 0 0 0 0
.159 0 0 0 0 0
.690 0 0 0 0 0
.155 0 0 0 0 0
.146 0 0 o 0 0
0 1.08 0 0 0 0
0 .689 0 0 0 0
0 .201 0 0 0 0
0 - 011 0 a 0 0 0
0 0 1.0 0 0 0
0 0 0 .77030 0

, 0 0 0 0 .85030
0 0 0 0 0 .580
1_. _

 

0 = diag. [2870 .980 .950 .970 .520 .980 .980 .960 .540 .960

.990 0 .410a .280a .6605]

 

 

B = r1.oa 0 0 v 0 0 0
0 1.08 0 o 0
0 0 1.08 o 0 0
.036 -.064 —.037 1.03 0 0
.106 -.019 .036 -.300 1.03 0 a
-.O64 -.117 —.017 -.122 0 1.0_J

k

 

 

.020 0

.020
.990

122

0000

0
O
0
0
.840 .020
.020 .840

 

123

APPENDIX C
LISREL MATRIX SPECIFICATIONS FOR.MODEL TWO

In the following matrixes coefficients marked with a zero or a

superscript "a" are fixed values. Other values indicate free parameters.

These values are the start values for the LISREL analysis.

 

 

AY . 1.0“ 0 0 0 0 0
.081 0 0 0 0 0
.811 0 0 0 0 0
0 1.0“ 0 0 0 0
0 .538 0 0 0 0
0 1.691 0 a 0 0 0
0 0 1.0 0 0 0
0 o .200 0 0 0
0 0 .666 0 0 0
0 0 0 .770“ 0 0
0 0 o 0 .850“ 0
0 0 0 0 0 580“

a
A - 1.0 0
x [:0 1.0“]

diag. [2833 .990 .890 .892 .969 .691 .700 .772 .899

(D
II

.410“ .280“ .660“

B s 1.0“ 0 o 0 0 _T
0 1.0“ 0 0 0
0 0 1.0“ 0 0
0

.494 -.212 -.500 1.0“ a
1.537-1.366-u096 -.045 1.0

L:.064 -.117 -.017 -.122 0 1.0

00000

a

 

 

l

124

 

.118 0

.191 0

.227 0

0 .219

.188 .207

.315 .200
r- '1
1.0“ -.206
-.206 1.0“

 

7‘
.153 .020 -.029 0

0 0
.020 .070 .062 0 0 0
-.029 .062 .176 0 0 0
0 0 0 .589 0 0

0 0 0 0 .198 .020

 

0 0 0 0 .020 .420

 

LIST OF REFERENCES

125

LIST OF REFERENCES

A.C. NIELSEN COMPANY. Nielsen report on television: 1980. Northbrook,
Illinois: A.C. Nielsen Co., 1980.

ABEL, J.D., REAGAN, J. & MUTH, T.A. Sense of community and media use.
Paper presented to the International Communication Association,
Acapulco, 1980.

ACOCK, A.C. & SCOTT, W.J. A model for predicting behavior: The effect
of attitude and social class on high and low visibility political
participation. Social Psychology Quarterly, 1980, 43, 59-72.

 

ALLEN, R.L. The reliability and stability of television exposure.
Communication Research, 1981, 8, 233-255.

 

ALLEN, R.L. & BIELBY, W.T. Blacks' relationship with the print media.
Journalism Quarterly, 1979a, 56, 488-496.

 

ALLEN, R.L. & BIELBY, W.T. Blacks' attitudes and behaviors towards
television. Communication Research, 1979b, 6, 437-447.

 

ASHER, H.B. Causal modeling. Beverly Hills: Sage, 1976.

 

BARNLUND, D.C. Interpersonal communication: Survey and studies. Boston:
Houghton Mifflin Co., 1968.

BELL, D. The measurement of knowledge and technology. In E.B. Sheldon
& W.E. Moore (eds.) Indicators of social chang_. New York:
Sage, 1968.

 

BIENVENU, M.J. & STEWART, D.W. Dimensions of interpersonal communication.
Journal of Psychology, 1976, 93, 105-111.

 

BLALOCK, H.M. Social statistics. New York: McGraw—Hill, 1979.

 

CABLE TV STUDY COMMITTEE. Cable television in Detroit: A study in
urban communication. Detroit: City Clerk,l972.

 

CHAFFEE, S.H. The interpersonal context of mass communication. In
F. Kline and P.J. Tichenor (eds.). Current perspectives in mass
communication research. Beverly Hills: Sage, 1972, 95-120.

 

126

CLARK, C. Conditions of economic progress. Longon: MacMillan, 1957.

 

COMSTOCK, G. & COBBEY, R.E. Research directions for the study of
television and ethnicity: Methodological considerations. Paper
presented to the Conference on Television and the Socialization
of the Minority Child, Los Angeles, 1978.

CONRATH, D.W. & THOMPSON, G.B. Communication technology: A Societal
perspective. Journal of Communication, 1973, 23, 47-63.

 

CONWAY, M.M., WYCKOFF, M.L., FELDBAUM, E. 8 AHERN, D. The news media
in children's political socialization. Public Opinion Quarterly,
1981, 45, 164-178.

 

CROEL, P. & HUSBAND, C. Communication and community: A study of the
Swindon community television experiment. Leicester, England:
University of Leicester, 1951.

 

DAHLING, R.L. Shannon's information theory: The spread of an idea.
Studies of Innovation and of Communication to the Public.
Stanford: Institute for Communication Research, 1962, 119-139.

 

DANNEFER, W.E. & POUSHINSKY, N.P. Language and community. Journal
of Communication, 1977, 27, 122-126.

 

DAVEIS, C.S. Popular government. In. J.L. Blau. Social theories of
Jacksionian democracy. New York: Hafner, 1947.

 

 

DEFLEUR, M.L. & BALL-ROKEACH, S. Theories of mass communication. New
York: Longman, 1975.

 

DIMMICK, J.W., MCCAIN, T.A. 8 BOLTON, W.T. Uses of mass media: Patterns
in the life cycle--theoretica1 perspectives and research

strategies. Paper presented to the Speech Communication Association,
'Minneapolis, 1978.

DONOHUE, G.A., TICHENOR, P.J. & OLIEN, C.N. Gatekeeping: Mass media
systems and information control. In F.G. Kline & P. J. Tichenor.

Current perspectives in mass communication research. Beverly
Hills: Sage, 1972, 41-69.

DURKHEIM, E. Suicide: A study through sociology. Glencoe, Illinois:
Free Press, 1951.

EMERSON, T.I. Toward a general theory of the first amendment. New York:
Vintage, 1966.

 

EMERY, J.D. An electronic marketplace of ideas. Journal of Communica-
tion, 1978, 28, 77-80.

 

FAINSTEIN, N.I. & FAINSTEIN, S.S. Urban political movements. Englewood
Cliffs, New Jersey: Prentice-Hall, 1974.

127

GATSEOS, P.M. 8 WEST, A.S. The citizens band personal radio service:
A socio-technological assessment. Denver: Denver Research
Institute, 1979.

 

GITTLIN, T. The whole world is watching. Berkeley: University of
California Press, 1980.

GRANOVETTER, M.S. The strength of weak ties. American Journal of
Sociology, 1973, 78, 1360-1380.

GREENBERG, 8.8. Communication and related behaviors of a sample of
Cleveland Black adults. Project CUP, Report 13, Michigan State
University, 1970.

GUMPERT, G. The rise of mini-comm. Journal of Communication, 1970,
20, 280-290.

HUNTER, F. Community ppwer structure. Chapel Hill: University of
North Carolina Press, 1953.

 

ISAAC, L., MUTRAN, E. 8 STRYKER, S. Political protest orientations
among black and white adults. American Sociological Review,
1980, 45, 191-213.

JACKSON-BEECK, M. Mass communication and political participation in
the United States. Paper presented to the International Communica-
tion Association, Philadelphia, 1979.

JORESKOG, K.G. 8 SORBOM, D. LISREL: User'sguide. Version IV, Release
2. Chicago: International Educational Services, 1978.

 

JORESKOG, K.G. 8 SORBOM, D. Statistical models and methods for analysis
of longitudinal data. In D.J. Aigner 8 3.8. Goldberger (eds.).
Latent variables in soci-economic models. Amsterdam: North
Holland, 1977.

KATZ, E. The social ininerary of technical change: Two studies in the
diffusion of information. Studies of innovation and communication
to the public. Stanford: Institute for Communication Research,
1962, 1-38.

 

KATE, E. 8 LAZARSFELD, P. Personal influence. Glencoe, Illinois:
Free Press, 1955.

KIMSEY, W.D. 8 ATWOOD, L.E. A path model of political cognitions and
attitudes, communication and voting behavior in a congressional
election. Communication Monggraphs, 1979, 46, 219-230.

 

KLINE, F.G. 8 TICHENOR, P.J. Current_perspectives in mass communication
research. Beverly Hills: Sage, 1972.

128

KLUEGEL, J.R., SINGLETON, R., JR. 8 STARNES, C.E. Subjective class
identification: A multiple indicator approach. American
Sociological Review, 1977, 42, 599-611.

 

KRAUS, S. 8 DAVIS, D. The effects of mass communication on political
behavior. University Park, Pennsylvania: The Pennsylvania State
. University Press, 1976.

LAZARSFELD, P., BERELSON, B. 8 GAUDET, H. The people's choice.
New York: Columbia University Press, 1968.

LONG, J.S. Elimination and hypothesis testing in linear models

containing measurement error. Sociological Research and
Methods, 1976, 5, 157-178.

 

LOUIS HARRIS AND ASSOCIATES. Factbook on aging: Approfile of America's
older population. Washington, D.C.: National Council on Aging,
1975.

 

MAISEL, R. The decline of mass media. Public Opinion Quarterly, 1973,
37, 159-170.

MARTIN, R.R., MCNELLY, J.T. 8 IZCARAY, F. Is media exposure unidimen-
sional? A sociological approach. Journalism Quarterly, 1976,
53, 619-625.

MARUYAMA, C. 8 MCGARVEY, B. Evaluating Causal Models: An application
of maximum-likelihood analysis of structural equations. Psychological
Bulletin, 1980, 87, 502-512.

 

MCCLEOD, J.M. 8 O'KEEFE, G.J., JR. The socialization psrspective and
communication behavior. In F. Kline and P.J. Tichenor. Current
perspectives in mass communication research. Beverly Hills:
Sage, 1972, 121-168.

 

MCCOMBS, M.E. Mass communication and political campaigns. In F.G. Kline
and P.J. Tichenor. Current perspectives in mass communication
research. Beverly Hills: Sage, 1972, 169-194.

MCLAUGHLIN, B. Studies in social movements. New York: Free Press, 1969.

MCPHEE, W.N. Formal theories of mass behavior. New York: Macmillan,
1963.

MENCKEN, H.L. Notes on democracy. New York: Knopf, 1926.

 

MENZEL, H. Quasi-mass communication: A neglected area. Public Opinion
Quarterly, 1971, 35, 406-409.

MILL, J.S. Considerations on representative democracy. New York:
Harper and Brothers, 1862.

129

MILLER, G.R. The current status of theory and research in inter-
personal communication. Human Communication Research, 1978,
4, 164-178.

 

New party seeks policy change for Blacks, poor. Michigan Daily,
December 4, 1980, 4.

 

O'KEEFE, G.J. Political malaise and reliance on media. Journalism
Quarterly, 1980, 57, 122-128.

 

O'KEEFE, G.J. 8 LIU, J. "First time voters; Do media matter? Journal
of Communication, 1980, 30, 122-129.

 

NUNNALLY, J.C. Psyghometric theory. New York: McGraw-Hill, 1978.

 

PARKER, E.B. Implications of new information technology. Public
Opinion Quarterly, 1973, 37, 590-599.

POSNER, R.A. The probable effects of pay cable television on culture
and the arts. In R. Adler 8 W.S. Baer (eds.) The electronic box
office. New York: Praeger, 1974, 544-548.

 

RADA, S.E. Manipulating the media: A case study of -a. Chicano strike
in Texas. Journalism Quarterly, 1977, 54, 109-113.

REAGAN, J. The uses of media and communication for social movements.
Paper presented to the Midwest Political Science Association,
Cincinnati, 1981.

REAGAN, J. 8 ABEL, J.D. Citizens band radio use and sense of community.
Unpublished paper, Department of Telecommunication, Michigan State
University, 1979.

ROBERTS, C.L. Media use and difficulty of decision in the 1976
presidential campaign. Journalism Qparterly, 1979, 56, 794-802.

 

ROBINSON, J., RUSK, J.C. 8 HEAD, K.B. Measures of political attitudes.
Ann Arbor: Institute for Social Research, 1973.

 

ROGERS, E.M. Characteristics of agricultural innovators and other
adoptor categories. In Studies of innovationsand of communication

to the public. Stanford: Institute for Communication Research,
1962, 63-97.

 

 

SCHOENFELD, A.C. The press and NEPA: The case of the missing agenda.
Journalism Quarterly, 1979, 56, 577-585.

 

SCHOENFELD, A.C., MEIER, R.F. 8 GRIFFIN, R.J. Constructing a social
problem: The press and the environment. Social Problems,
1979, 27’ 38-610

 

130

SHAW, E.F. 8 RIFF, D. Newspaper reading in two towns. Journalism
Quarterly, 1979, 56, 477-487.

SURLIN, S.H. Perceived need for minority ownership of radio stations.
Journalism Quarterly, 1977, 54, 587-591.

TAN, A. 8 VAUGHN, P. Mass media exposure, public affairs knowledge,
and Black militancy. Journalism Quarterly, 1976, 53, 271-279.

USEEM, B. Solidarity model, breakdown model, and the Boston anti-
bussing movement. American Sociological Review, 1980, 45, 357-369.

VIDICH, A.J. 8 BENSMAN, J. Small town in mass society. In R.L. Warren.
New perspectives on the American community. Chicago: Rand-McNally,
1977.

WARREN, R.L. The community in America. Chicago: Rand-McNally, 1978.

WINER, B.J. Statistical principals in experimental design. New York:
McGraw-Hill, 1971.

WOELFEL, J. 8 FINK, E.L. The measurement of communication process:
Galileo theory and method. New York: Academic Press, 1980.

WRIGHT, C.R. Mass communication. New York: Random House, 1959.

 

YOUNG, M.D. 8 WILMOT, P. Family and kinship in East London. Middlesex,
England: Penguin Books, 1962.