- r

 

 

3

J”;

.~-1-:y\
' Ind Lian-'-

w r ”-a

a!

__ fl-.. - ma.) ,7

  

' {B "' -"- ’1'“;
IV; :E’liffrrat: State
(J .
Umvermty

   

TH chc

 
   

This is to certifg that the

thesis entitled

Longitudinal Analysis of Senate Behavior:
A Methodological Inquiry

presented by

David Jerome Peterson

has been accepted towards fulfillment
of the requirements for

Ph.D. degree inPolitical Science

/////Qf

Major pro ssor

Date 14 May 1970 ‘

O~169

ABSTRACT

LONGITUDINAL ANALYSIS OF SENATE BEHAVIOR:
A METHOLOGICAL INQUIRY

By

David Jerome Peterson

In this dissertation the longitudinal (non—cross—sectional)
perspective has been defined (within the context of studies of
Congressional behavior) to include not only dynamic studies such
as before-and—after designs or panel designs, but also "static"
studies which make systematic use of data from more than one
Congress.

The first three Chapters discuss problems associated with the
predominant cross—sectional approach and the nature and hypotheti-
cal advantages of longitudinal analysis. The longitudinal analyses
in Chapters IV-VI were designed to broadly check explicit metho-
dological hunches about the utility oflongitudinal analysis and to
further examine substantive assumptions about consistency and
change in Congressional behavior.

It was found that the sharp limits of the cross—sectional per—
spective reduce opportunities for measurement, explanation, and
prediction. The longitudinal perspective, on the other hand, in—
creases the opportunities to reliably and validly measure otherwise
elusive behavior such as the practice of, support for, or opposition
to McCarthyism (Chapter IV). The longitudinal per3pective helps us

to identify predictors of recurring behavior, identify pivotal or

David Jerome Peterson

marginal decision-makers and thus helps clarify the strategic
alternatives open to political leaders (Chapter V). And the
longitudinal perspective helps to explain subtle behavior (for
example, voting on foreign policy) that is likely to be misinter-
preted in a cross-sectional perspective.

Probably the clearest advantage of longitudinal over cross-
sectional analysis is in the study of change in individual be—

havior. If we accept consistency—change in rank order and consis-

 

tency—change in magnitude as two basic dimensions for developing

a typology of longitudinal analyses of consistency-change in leg—
islative behavior —- each of the three types exemplified in
Chapters IV, V, and VI fits in a different one of the three cells
representing logically possible analyses. Chapter IV, a static
study, fits in the cell representing consistency in both rank order
and magnitude. Chapter V fits in the cell representing change in
magnitude but consistency in rank order. And Chapter VI, a dynamic
study, clearly fits in the cell representing change in both
magnitude and rank order.

In addition to the methodological insights regarding the
analysis of consistency-change there are substantive hypotheses
about consistency change in Congressional behavior. The subsequent
research (reported in IV, V, and VI) did not negate these assumptions.

Most Senators have rather well formulated policy tendencies
by their first term in the Senate. These policy tendencies are

supported by numerous interrelated and relative enduring values,

David Jerome Peterson

sympathies, identifications, beliefs, and attitudes, which guide
cognitive processes such as perception, thinking, and memory.

Given the excessive cost of frequent and systematic surveys,
many politicians hesitate to change much from "winning positions"
taken in previous elections. Knowing that they won when they took
positions x, y, and z hardly constitutes precise scientific evidence
of why they won, but it may often be salient information for those
who lack precise information. Furthermore, the rank order of con-
stituencies is not likely to change dramatically in the short run
on many of the constituency variables relevant to the overall
policy positions of Senators.

There is an ideological bias in the social exposure and
communication patterns of Senators that results in social reinforce—
ment of their policy tendencies. That is Senators, once in office,
establish working relations with like-minded people and tend to "hear"
disproportionately from those who agree. This differential contact
is due both to the predispositions of representatives and those of
lobbyists, constituents and others who disproportionately contact
politicians who lean their way.

These and the other informed hunches about consistency-change
especially receive support from the research reported in Chapters

IV and V, which like the "library research" cited in Chapter II,

reveals impressive consistency over time.

LONGITUDINAL ANALYSIS OF SENATE BEHAVIOR:
A METHODOLOGICAL INQUIRY

BY

David Jerome Peterson

A THESIS

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

DOCTOR OF PHILOSOPHY
Department of Political Science

1970

ACKNOWLEDGEMENTS

I greatly acknowledge the generous assistance of Dissertation
Committee members, Dr. Harold Spaeth, Dr. Joseph Schlesinger and Dr.
LeRoy Ferguson. Special appreciation is extended to Professor
Spaeth whose interest and guidance contributed greatly to the comple-

tion of this dissertation.

ii

TABLE OF CONTENTS

Page
Chapter
I. THE GENERAL PROBLEM . . . . . . . . . . . . . . . . 1
Introduction
The Theoretical Structure of a Highly-Developed
Science: A Typical Structure in Political Science 1
II. THE NATURE OF LONGITUDINAL ANALYSIS . . . . . . . . 11
Definition of Longitudinal Analysis 11
Attributes, Facets or Properties that may be
used for Constructing and Distinguishing Types
of Longitudinal Analysis 13
Four General Types of Longitudinal Analysis 15
Reasons for Expecting to Make Longitudinal
Analysis Fruitful 17
III. THE ADVANTAGES OF LONGITUDINAL ANALYSIS . . . . . . 26
A General Summation of the Advantages 26
Longitudinal Analysis Promotes the
Specification of Concepts 27
The Advantage of Longitudinal Analysis for
Replacing Plausible but Scientifically
Deficient Ex Post Facto ”Explanations" with
Scientific Explanations 31
The Advantage of Longitudinal Analysis for
Suggesting the Relative Importance of Variables 35
The Advantage of Longitudinal Analysis for
the Study of Change 36
The Subject Matter of Chapters IV, V, VI 42
IV. THE MEASUREMENT OF McCARTHYISM: A LONGITUDINAL
ANALYSIS 0 O O O O I O I O O 0 O 0 O O O I O O O O 43
Specification of the Concept 45
The Theoretical Import of the Concept 50
The Ten Indicators Used in the Test of the
Scale Hypothesis 60
The Test of the Sclae Hypothesis 65
Implications 65

iii

V. URBAN AFFAIRS DECISION MAKING IN THE SENATE:
A PREDICTIVE STUDY . . . . . . . . . . . . . . . . . . 71

The Department of Urban Affairs Decision 73
Constituency Variables and Policy Position 78
Party Affiliation and Policy Position 79
Some Additional Implications and Extensions 86
Advice to (the Urban) Prince(s) 92

VI. CONTINUITY AND CHANGE IN FOREIGN POLICY POSITION:
LONGITUDINAL ANALYSIS AS AN AID TO CAUSAL INFERENCE. . 94

Bipartisan Delegation Model 98

Nominal Party Affiliation 101

Party Identification Medel 106

The Before-and—After Analysis: Purpose, Span

Indices and Results 109

Relation Between Party and Direction of

Change 1960—61 113
VII. CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . 118

iv

I. THE GENERAL PROBLEM

Introduction

 

Measurement lies at the juncture of theory and
experience--the first contact of reason and nature.

Political scientists have frequently expressed dissatisfaction
with the progress of the discipline. Often these expressions have been
stated in terms of conflict between two alternative approaches or metho-
dologies, for example, hyperfactualism versus general theory. Such
explanations of the state of the discipline in terms of conflict between
a fruitless and a fruitful approach may provide interesting, at times
even dramatic reading, but such single-cause explanations provide insuf-
ficient clarification and guidance. One response to the plaint that we
have "too much description and not enough theory" is the truism that
"descriptive research and theory must develop together." But both the
initial plaint that "we have too much description and not enough theory"
and the response that "description and theory must develop together"
suffer from oversimplification. It may be that we have had too much of

some kinds of description--but what kinds of description have we had too

 

 

much of? And it also may be that theory and description must develop
together--but what kinds of description contribute to theory? Clearly
there is need to differentiate between types of description that "we have

too much of" and the types of description essential to theory-building.

 

lHenry Margenau, "Philosophical Problems Concerning the Meaning of
Measurement in Physics," in Measurement: Definitions and Theories (New
York: John Wiley, 1959).

2
Before attempting to identify a few aspects of description that
contribute to theory, it is well to ask: What is the role of descrip-
tion in science? And how does the interplay of empirical evidence and
theory in political science compare with that in the scientific ideal?

The Theoretical Structure of a
HighlyéDevelgped Science

 

 

In the scientific ideal there are tight formal connections among
the concepts and epistemic connections (empirical indices) to link some
of these concepts directly to the data. In Carl Hempel's terms the
formal connections endow concepts with theoretical import, and the episte-
mic connections endow concepts with empirical import.2 If the formal
links are sufficiently tight it may be possible to get by with few em-
pirical indices. All concepts not directly linked to data may neverthe-
less gain empirical import through the indirect connections provided by
the formal net. In such a theoretical structure it is legitimate to
include statements about unobservable entities.3

Margenau's description of the nature of a well-developed science

may be diagramed as follows:4

 

2Carl G. Hempel, Fundamentals of Concept Formation in Empirical
Science, International Encyclopedia of Unified Science, Vol. II. No. 7
(Chicago: University of Chicago Press, 1952), pp. 39-50.

3H. Feigl, "Philosophical Embarrassments of Psychology," The
American Psychologist, XIV (1959), 127.

4Henry Margenau, The Nature of Physical Reality (New York: McGraw-
Hill, 1950), chap. 5.

 

 

 

 

>H>U

Figure l. Circles represent constructs.
Single lines represent formal connections.
Double lines represent epistemic connections
(rules of correpondence).

mmbw><wmmwo

 

 

mZOHH><WmmWO

 

A Typical Structure in Political Science
Political scientists have innumerable theoretical notions and
innumerable observables but a poverty of scientific theory. A major
reason for this lack of theory is the weakness of the connections between
the theoretical notions (on the left, below) and the observables (on the

far right, below).

 

 

|
|
\
\
\
l
l
l
l
I
I
l
>+—1:>U
mZOHH><WwaO
mmrw><wmmwo

 

Figure 2.

Broken single lines represent weak formal connections.

Double lines represent epistemic connections.

Absence of lines represents insular concept (no

connection).

Weak epistemic connections also account in part for the weak for-

mal links (represented by the broken lines between the concepts in
Figure 2). Little descriptive differentiation and flexibility (as with
dichotomization) often means insufficient flexibility to formulate laws:
That is, a low level of measurement or failure to measure not only may
mean a weak epistemic connection but also may seriously hinder the expres-
sion of formal relations among concepts. Thus, partly because of the
failure of much of the discipline to move beyond classification to
ordinal and higher levels of measurement, we usually must resort to loose
verbal connections between concepts.

Furthermore, because such loose verbal connections predominate in

contemporary political "theory" and because most of the empirical indices

5

are also deficient, it is advisable for political scientists to develop
many more indices than would otherwise be necessary. As Royce puts it:

Lack of a sufficient number of convincing formal

multiple connections, . . . combined with weakness

of epistemic correlation . . . suggests that less

well-developed sciences . . . need to stay closer

to the perceptual plane until they can provide

tight formal nets in the conceptual plane.

Exclusive attention to either side of the above figure is scien—
tifically fruitless. Concepts must have both theoretical and empirical
import. Because political scientists have mountains of low-level descrip-
tions with little theoretic import but few effective epistemic connections,
it is only prudent to devote more attention to the final stage of descrip-
tion--measurement. If this claim errs, it is in the direction of under-

statement. Perhaps Torgerson comes closer to correctly stating the

importance of focusing more on the connections between data and theoretic

 

concepts. After demonstrating how the low investment in measurement has
made many theories immune to adverse results (by allowing the theorist
to attribute negative results to weak measures), Torgerson states that:

The concepts of theoretical interest tend to lack
empirical meaning, whereas the corresponding
concepts with precise empirical meaning often

lack theoretical import. One of the great problems
in the development of science is the discovery or
invention of constructs that have . . . both.
(Italics mine.)b

 

 

 

It is therefore crucial that the reaction to hyperfactualism

(excessive attention to the far right of the diagram) not be the equally

 

5J. R. Royce, "Factors as Theoretical Constructs," American Psychol-
ogist, XVIII (1963), 524. If staying "closer to the perceptual plane" were
interpreted as meaning indiscriminate gathering of poorly connected facts,
the consequence would be countinued hyperfactualism. It is interpreted
here as meaning multiple epistemic connections.

 

6Warren S. Torgerson, Theory and Methods of Scaling (New York:
John Wiley and Sons, 1958), p. 8.

 

6
sterile error of confusing speculation for scientific theory. We cannot
develop theoretical science first and then worry about mundane matters
of measurement. "The development of a theoretical science . . . would
seem to be virtually impossible unless its variables can be measured
adequately."7

Recoggizing the sterility of the twin excesses of hyperfactualism

 

and spggulation devoid of empirical import is one step toward scientific

 

progress. Deciding to respond to the shortage of concepts having both

 

theoretical and empirical import by devoting more attention to the con-

 

nections between observables and theoretical concepts is a second step.

Identifying_§pecific difficulties that impede the attainment of the final

 

stage of description-measurement--is a third. But scientific payoff is

 

attained only after a fourth step: the identification, development, and

 

utilization of tactics and strategies to cope with these obstacles to

 

measurement. The preceding pages have dealt with steps one and two. The

 

next few pages will briefly sketch some of the difficulties that impede
measurement, offer examples of research tactics that can overcome some
of these difficulties when the data are available, and suggest that the
lack of data (to meet the criteria of scientific measurement) is frequently
a function of the researcher's limited time perspective rather than of
natural impossibilities or excessive cost. This will be followed by an
exploration of longitudinal analysis.

A general requisite of measurement is isomorphism. There must be
correspondence between some formal aspect of the numbers assigned (for
example, the rank of ordinal numbers) and some aspect of reality (for

example, the order of the subjects on a dimension). Alternatively, a

 

7Ibid., p. 2.

7

major difficulty with measurement has been the frequent lack of iso-

morphism or of any evidence of isomorphism. Numbers are easily assigned.

But assigning them so that at least one formal aspect of the number

system (for example, rank) will probably correspond to some aspect of

the order or relations among the subjects is another matter. It is

safest to assume that scores never solely reflect actual behavior. To

some extent they reflect:

l.

3.

the peculiarities of the particular imperfect data selected
for analysis (type of data or time-span of data);

the nature and limits of the general measurement model

 

selected (summated index, Guttman scale, Lingoes' multiple
scalogram analysis, factor analysis, Kruskal's multidimen-
sional scaling, Guttman's smallest Euclidean space analysis,

and so forth);

the nature and limits of the specific theoretical version (for

 

example, principal-factor or centroid solution) of the more

general model;

 

the nature and limits of the comppter program (for example,

 

SHARE or BMD) used to apply the previous model to data;

the specific pptions selected from the computer program (for

 

example, choice of criterion of rotation, choice of criterion
for determining the number of factors to rotate, or choice of
R or Q technique);

the interpretation given to the scores (that is, what property

 

of the subjects is said to correspond in what way to what
formal aspect[s] of the numbers assigned);

other considerations that will be touched on in other chapters.

8
If at each point the researcher repeatedly selects the same options
he is not likely to discover how or to what extent the scores that he as-
signs are products of his particular methods. He is likely, at best, to
have a very narrow perspective.

When we restate the general difficulty of attaining isomorphism

 

in terms of many_specific difficulties that are partly consequences of

 

specific decisions, tactics for cppinggwith the problem begin to suggest

themselves; and the_prob1em becomes more manageable.

 

If, for example, we consider the question, "Can indices formed from
voting data possibly tap the same variable reflected in the content of
speeches?" or "Does the rank of the subjects on the hypothesized dimen-
sion vary greatly with the type of data used?"--the tactic of complemen-
tary use of different types of data or the tactic of using a criterion
of validity is suggested.8

For another example imagine that a researcher studying policy
behavior on labor-management issues finds that the initial scores he
obtains divide senators into two opposed groups, one that steadfastly
sides with management and another that steadfastly sides with labor. He
may question whether this lowest degree of differentiation (dichotomiza-
tion) corresponds to reality: i.e., is isomorphic. Although labor-
management conflict is clearly one of the most bipolarized dimensions of
Congressional behavior, the researcher may suspect that the senators'
over-all positions on labor-management issues may be more complex and
differentiated than is indicated by his measure (based on a small subset
of responses to labor-management issues). When the question of isomorphism

is directed to the specific problem of how the degree of differentiation

 

8These will be discussed below.

9
in numbers assigned corresponds to the actual degree of differentiation
in policy positions, one thinks of the tactic of using longitudinal data
to obtain a more comprehensive view of the degree of differentiation.

It is vital that we reformulate the4general Question of isomorphism

 

in terms of mapy problems specifically enough stated to suggest tactics

 

for making these_problems more manageable. The frquent and effective

 

use of these tactics seem unlikely, however, if we habitually conceive of

 

research in cross-sectional terms. No single session or Congress is

 

likely, for example, to offer data for a measure of "right-wing indis-
criminate hostility" or "support for McCarthyism" that offers realistic
differentiation, a test of relations among indicants, an independent
test of validity, and evidence that the behavior is not ephemeral. To

"naive,"

even ask for such is to risk being labeled "impractical" or
(i.e., unacquainted with the difference between measurement in psychology
and education based on intimate studies of students, patients, and other
compliant persons; and measurement of characteristics of extremely busy
politicians playing roles that deter revealing exposure to panel inter-
views, experiments, narcosis, polygraphs, projective tests, and so forth).
And to ask those who claim to understand the dimensionality of legislative
behavior to occasionally predict or to show how they have at times con-
trolled relevant variables might bring an equally strong reaction.

While it is unfair to expect those who study important elective
officials to fully meet all conceivable criteria of measurement, it is
fair to insist that they not fail to meet standards because a narrow time

perspective leads to the false conclusion that "we just don't have the

data."

10
Because many of the research tactics essential to improved measure-
ment-—and thus essential to improved "juncture of theory and experience"—-
depend on a longitudinal perspective, it seems appropriate to:
1. define longitudinal analysis;
2. list properties or attributes that may be used to generate
types of longitudinal analysis;
3. illustrate how types of longitudinal analysis may be
generated;
4. briefly discuss four general types of longitudinal analysis;
5. offer reasons for expecting sufficient consistency to make
longitudinal analysis fruitful;
6. hypothesize methodological advantages of longitudinal

analysis.

II. THE NATURE OF LONGITUDINAL ANALYSIS

Definition of Longitudinal Analysis

 

The term longitudinal analysis is used below in its broadest form
that encompasses many Specific types of analyses extending over time.

It is used as the counter-term for cross-sectional analysis, thus avoid-
ing the more cumbersome term "non-cross-sectional analysis."

The term cross-sectional analysis of congressional behavior will
be used below to label static studies using data from a single Congress.
Longitudinal analysis of congressional behavior will be used to refer
to: (1) static studies using data from two or more Congresses, and (2)
dynamic studies (for example, before-and-after designs).

Although the examples of longitudinal tactics appearing in
Chapters IV, V, and VI are based on the measurement of recurring behav-
ior, I do not wish to continue the sterile argument between behavioralists
and traditionalists as to whether humans and events are unique and im-
measurable. I wish to concede at the outset that all real events and
individuals are unique, non-recurring, and immeasurable. But I hasten

to add that scientists can abstract, measure, and relate properties of

 

ll

12
events, and individuals.1 That scientists do so imperfectly does not
damn the scientific enterprise if these studies are incrementally valid
or lead to studies which are.

Longitudinal analysis is not used here to refer to retrospective
studies that attempt detailed historical reconstruction of unique events,
or studies using data generated at one point in time (for example,
through a survey) but assumed to reproduce observations from widely
separated periods of time. The latter type of study might share the
advantages of longitudinal analysis if there were some way of determining
which responses represented honest reports of accurately remembered be-
havior and which responses reflected either: (1) faulty memory due to
time lapse and rationalization, or (2) inaccurate reporting due to con-
scious desire of some to appear consistent and of others to appear to
have been "right all along." The last tendency may be particularly prev—
alent among those playing political roles and suggests the advantage of
having longitudinal data (data generated over a period of time as with

records of Senate voting and verbal behavior.)2

 

1Hayward Alker, Jr. argues that "if events are unique, repetitive
observation is impossible. . . ." But operations in his Mathematics and
Politics reveal that science merely claims to abstract, measure, and re-
late properties of events or individuals (New York: Macmillan, 1965).
Torgerson states ". . . it is always the properties that are measured and
not the systems themselves. Measurement is always measurement of a prop-
erty and never measurement of a system. . . . individuals, cultures, and
chromosomes are immeasurable by their very nature. Each, however, pos-
sesses properties that perhaps can be measured." Events are not measured
but their properties, thus the fact that no two events are identical in
all their detail does not prevent the measurement of those properties which
they do have in common. Theory and Methods of Scalipg (New York: John
Wiley and Sons, Inc., 1958), p. 9.

2A further advantage of data generated over time (Senate roll calls,
data through content analysis of the Congressional Record) is the availa-

bility of many more indicants than one could gather through interview.

 

 

 

 

13

Finally, two independently designed cross-sectional studies

typically do not constitute a longitudinal analysis. But many cross-

sectional studies contain aSpects that may be extended longitudinally

(in either direction) if the second researcher consciously designs his

research to take advantage of the potential of the first study. Such

longitudinal extensions are not to be confused with those secondary

analyses or replications that may involve longitudinal extensions of the

researchers' activity but not of the measures of the subjects' behavior.

 

Attributes, Facets, or Properties that May Be

 

Used for Construction and Distinguishing
Types of Longitudinal Analysis

 

 

Choice of respondents: repeat same respondents; or use
different respondents.

Location of time periods: past-(more recent) past; or
past-future.

Division of attention or research effort among time
periods: evenly divided among periods; or unevenly
divided (for example, one period is emphasized while
another is drawn upon primarily for a criterion for

a measure developed in the period emphasized).

Number of variables traced through time: small; or large.
Aspects of behavior abstracts: rank order; magnitude,

shape of distribution; communality; percentage of

14

variance accounted for by various factors; and so
forth.3

6. Whether change is measured: dynamic; or static (for
example, a longitudinal search for enough data to
establish a measure of relatively enduring though in-
frequently expressed behavior).

7. Amenability of variables to control: controllable;
not controllable.

8. Whose behavior is (described, explained, predicted)
made more subject to control by whom: the elected

leader's behavior by the electorate, or vice versa.

 

3The communality of a variable is given by the sum of the squares
of the common—factor coefficients. In Q—analysis (analysis of a correla-
tion matrix of men, not items) of data meeting the assumptions of the
factor-analytic model, measures of communality aid in identifying deviants
or "mavericks." Moreover, this precise statistical concept contributes to
substantive political theory by promoting a distinction between types of
deviants. For example, after extracting three or four Q-factors from
United States Senate roll call data, it becomes clear that we need to dis—
tinguish those such as Goldwater who dissent a great deal but whose voting
behavior can be statistically accounted for by his loadings on a few com—
mon factors, from those who do not have substantial loadings on any of the
common factors and who will have high loadings reported for them only if
we instruct the computer to continue the rotational analyses until unique
factors are reported. In unidimensional analyses the first type has ex-
treme rank but very few nonscale responses; the second type has an unusually
large number of nonscale responses, or an unusual combination of positions
on separate unidimensional scales.

4An example of research which focuses on variables not amenable
to control is that of Thomas A. Flinn, "Party Responsibility in the
States: Some Causal Factors," The American Political Science Review, Vol.
LVIII (March, 1964). Professor Flinn concludes: "Assuming the theories
advanced in this study are nearly correct, then it seems that there is
vegy little chance of getting_or increasing party responsibility by de—
liberate action . . . whether it is desired or not." Perhaps man is
sometimes seen as a pawn in a fatalistic chess game, because of the con-
cepts chosen to guide perception rather than because this is the inevi-
table nature of reality. If Dahl, Lindblom, Brecht, Van Dyne, Myrdal,
and others are correct in conceiving of science as an aid to rational
decision, facets 7 and 8 assume considerable importance.

 

 

15

Four General Types of Longitudinal Analysis
(As Developed by Dichotomies l and 2)5

 

 

Location of Time Periods

Past-(More Past-Future
Recent) Past

R

e PREDICTION OF AGGREGATE
3 Use TREND BEHAVIOR FROM AGGREGATE
p Different DATA

0

n

d PREDICTION OF INDIVIDUAL
e Use BEHAVIOR FROM INDIVIDUAL
n Same PANEL SCORES7

t

s

llgpg,--An example of a trend study would be a study of how the
content of floor statements, perhaps as reported in the Congressional
Record, relates to severe economic changes or to important changes in
our national security. Hamilton, in his content analysis of sermons
given during the period 1929 to 1940 demonstrated how opinions can be

related to economic fluctuations.8

Another example of a trend study is
that of Cantwell in which he related congressional behavior to a series

of public opinion polls.9

 

5For a related presentation of attributes and types, see Lazarsfeld
and Rosenberg, The Language of Social Research (Glencoe: Free Press, 1955),
pp. 201-830

 

6Different samples from the same population.

7From such predictions one can, of course, also predict the be-
havior of a legislative body.

8Thomas Hamilton, "Social Optimism and Pessimism in American
Protestantism," Public Opinion Quarterly, Vol. VI, pp. 280-83.

 

9Frank V. Cantwell, "Public Opinion and the Legislative Process,"
The American Political Science Review, LV (1946), pp. 924-35.

 

16

nggl.--Pane1 studies may be illustrated by the cross-tabulation
of two or more scales taken from different sessions or Congresses which
contain a number of members in common. Such panel studies may identify
those who change rank and those who do not, while also indicating the
direction and the conditions of change. Foreign aid, civil rights, urban
affairs, and economic issues are certainly open to exploration by panel
study. Panel studies have the virtue of giving information on individu-
als as well as information on groups.

Prediction.--The two types of prediction studies shown above are

 

distinguished by whether they give information on individuals. The use
of individual scores has advantages over aggregate data for both scien-
tific explanation and practical application but does not exclude the
study of groups. Thus the type of prediction that appears in the fourth
quadrant is often preferable to that shown in the quadrant above it.
While some speak of prediction as not requiring any understanding of, or
interest in, the relation between the basis of the prediction and the
behavior predicted, it is assumed here that reliable predictions of the
policy behavior of legislators both require and contribute to the under-
standing of the conditions under which consistent policy behavior is to
be expected and the conditions under which change is likely. Panel
studies should help illuminate these conditions and provide a sounder
basis for prediction.

Some of the advantages of prediction will be suggested below. But
the interesting questions of the relationship between prediction and con-
trol, and how the choice of whose behavior is predicted for whose benefit
relates to who controls whom, will have to wait until a later work. It

may be noted at this point, however, that the earlier development of

17
techniques and financial support for the study of the electorate than of
techniques and support for the study of elective elites, combined with
the greater incentive and ability of the elective elites to bear infor-
mation costs, may make popular control less effective. The examples of
prediction that will be developed at any length below will focus more on
the behavior of elective elites than of the electorate. It is especially
important that rational democratic opinion makers (and through them
opinion leaders) receive information that will enable them to predict the
behavior of elective leaders with sufficient precision to maintain faith
in the effectiveness of popular control. Although the following work will
not accomplish this, some methodological and substantive aspects are
relevant.

It is obvious from a quick look at the number of facts (or attri-
butes) listed above that scores of specific types of longitudinal analyses
could be constructed. For this reason it is clearly more economical to
list facets that may be used to generate types than to attempt to com-
prehensively list and illustrate the much more numerous types. Although
only a few common types were discussed here, additional more specific
and less frequently discussed longitudinal tactics will be covered below.

Reasons for Expecting Sufficient Consistency

in Legislative Behavior to Make
Longitudinal Analysis Fruitful

 

 

 

While enduring or recurring behavior by Supreme Court justices has
been clearly demonstrated, we have relatively few published studies of
Congressional decision—making over time. Nevertheless, these do suggest

considerable consistency, although they do not clearly define its nature.10

 

10D. R. Brimhall and A. S. Otis, "A Study of Consistency in Congres-
sional Voting," Journal ongpplied Psychology, XXXII (1948), 1-7, 14. Re-
printed in Wahlke and Eulau, Lagislative Behavior (Glencoe: Free Press,
1959). Also Donald R. Matthews, cited below.

 

 

18
Matthews and others have noted that by the time men reach the United
States Senate they have usually developed rather stable opinions on many
policy matters. Matthews also argues that taking a public position,
especially at the time one first gains election to the Senate, tends to
encourage consistency. His argument is largely contained in the fol-
lowing quotes:

The importance of the senator's initial "mandate"--
partially self-defined, partially reflecting popular senti-
ment-~13 hard to overestimate. Senators, once in office,
establish working relationships with like-minded senators
and understandings and friendships with sympathetic lobby-
ists, administrators,and reporters that are difficult to
change. They quickly make a public record, and it is often
unwise to alter it drastically.

There were sizeable shifts in the ideological make-up of the
Senate during this ten-year period [1947-1956], but this was
not the result, for the most part, of incumbent senators
shifting their positions. Note that "the class of 1947"
maintains very much the same ideological complexion during
the entire ten—year period. The "class of 1949" has its
ideological center of gravity well to the left during the
entire period. Nor apparently has there been much change

in the overall stance on issues of those elected in 1950.

A senator's initial "mandate," therefore, may be a major l
influence on his voting many years after it was received.

A senator may consciously consider his initial public policy
position a promise which should be kept for moral or prudential reasons.
He may also have unconsciously associated his initial policy position
with the rewarding experience of winning, and thus conditioning as well
as conscious strategy fosters recurring policy behavior. It is also
the case that once a representative has taken a public position, he finds
it necessary to defend it by repeatedly offering some rationale. Both

the unconscious effects of repeating the rationale and the consciousness

 

l'lDonald R. Matthews, U. S. Senators and Their World (New York:
Vintage Books, 1960).

19

of having done so encourage recurring policy behavior.12

But there are other reasons to expect a large degree of consis—
tency in decision behavior. Senators must spend a large amount of time
and energy in public relations and other nondecision-making roles. Yet
they must also take a position on roughly 200-250 roll calls each session
and more than 1,000 per term of office. Usually these are not narrow
routine questions such as are handled at the lower levels of a bureau-
cracy. Typically these are policy questions with important though per-
haps uncertain consequences. Therefore, senators must consciously or
unconsciously adopt strategies to make decision-making problems more

manageable. Increased consistencyyin rank is a latent function of all

 

of the following hypothesized strategies or tactics for making decision-

 

making more manageable:

 

l. A subset of members get their initial impressions of a
stimulus or check their initial impressions of its direction
and magnitude by checking the perceptions of those who

"crystallized" earlier.

 

12One danger of elaborately defending what is felt to be a politi-
cally expedient but otherwise undesirable policy position is that the repre-
sentative gets trapped. If he isn't affected by the above—mentioned uncon-
scious processes, he may, nevertheless, feel it necessary to long continue
a socially harmful policy stance because: (1) he wishes to appear consis-
tent; (2) his position has gained the status of a promis; or (3) he, and
possibly other like him, have so reinforced his electorate's prejudices
rather than slowly and indirectly whittle away at them that he now has no
choice but to either continue reinforcing these prejudices or face greater
risks at the polls than he would have initially encountered had he adopted
a less expedient strategy. (The specious argument that it is undemocratic
for the representative to decide what is a prejudice or to place his judg-
ment against that of the majority of the moment will have to be dealt with
elsewhere. There is, however, increasing evidence that the norms of demo-
cracy are often part of our living, or behavioral, constitution precisely
because many officials, opinion-makers, and opinion leaders do adhere to
these norms when the man-in-the-street is indifferent or hostile.)

20

a) By checking both extremes on the dimension that the
senator assumes the new stimulus belongs to he may:

i) confirm his hunch about what issue is raised, or
find evidence that some other issue is raised by
the stimulus;

ii) confirm his hunch about the direction of a Yea
(or Nay) response on the new stimulus.

b) By estimating roughly where on the continuum the early
crystalllizers have divided, he obtains information with
which to estimate the magnitude (or valence) of the

stimulus. l3

2. Few if any members fully re-examine their positions on a

dimension each time another stimulus belonging to that dimen-

sion is presented for decision.

a) Members rarely spend much time examining items with esti—
mated cutting points far from their own positions on the

dimension.14

b) Many members also save time by not closely examining

items with estimated cutting points closer to their own

 

l3Ihose who decide last have the most social cues to take a position
on the continuum but the least influence on how others perceive the
stimulus.

ll‘Making decision-making more manageable by reducing in some way
the number of problems that one attends to is a necessary but not a suffi-
cient condition of rationality. But not all tactics that reduce the num-
ber of decisions to ruminate on promote rationality. Furthermore, a
tactic such as this (2a) may promote the rationality (or "education") of
those at one part of the dimension but promote irrationality on the part
of others at another part of the scale. Thus the consequences of some
tactics depend on other variables such as the position of the decision
maker.

Small
Change

21

positions but with manifest content and magnitude per-

ceived to be almost identical to that of a previously

decided question.
Incrementalism is another strategy or tactic (for making
decision more manageable) that promotes recurring behavior
and thus the fruitfulness of longitudinal analysis.15
Because the complexity of many problems is not matched or
offset by sophistication of empirical theory, adequacy of
measures of relevant variables, and a comprehensive pre-
scriptive theory to relate normative and empirical elements;
Congressmen often have a modest understanding of the conse-
quences of various alternatives. Small incremental policy
moves are made because of this modest understanding of the

complex problems and low faith in the relevance or efficacy

or revolutionary or utopian proposals requiring large change.

Types of Decision-Makinglb

 

High
Understanding

SOME ADMINISTRATIVE REVOLUTIONARY AND
AND TECHNICAL UTOPIAN DECISION-
DECISION-MAKING MAKING

Large
Change

INCREMENTAL
DECISION-MAKING WARS, REVOLUTIONS,
(often practiced by CRISES AND GRAND
Congressmen, judges, OPPORTUNITIES
party leaders, execu-
tives, and adminis
trators)
Low
Understanding

 

15

David Braybrooke and Charles Lindblom, A Strategy of Decision

 

(New York: Free Press, 1963).

l6Ibid., p. 78.

22

Thus we have a paradox. The predictability of Congressional be-
havior stems in part from the fact that Congressmen, being unable to
predict the long-run consequences of large policy changes, resort to a
series of incremental changes. These do not resolve issues in a short
span of time, as utopians or revolutionaries would like; thus we find
recurring issues and recurring behavior, open to longitudinal analysis.

If the reader differs with these claims about the predictable
nature of much decision-making in Congress, the differences can be re-
solved by careful longitudinal analysis. Certainly one clear advantage
of longitudinal analysis is its potential for clarifying what types of
decision-making occur, on what issues, and under what conditions.17

Another reason to expect considerable consistency in the relative
policy positions of most representatives over a six-to ten-year period,
is that a representative's socio-economic background and status, and his
career-goals, values, and group memberships and identifications usually

do not change much in such a period of time.

An argument against the hypothesized fruitfulness of longitudinal

 

17It is probable that congressional decision-making is more complex
than the excellent work of Braybrooke and Lindblom suggests. They list a
series of incremental changes in congressional policy as evidence that
congressmen pursue incremental strategy. But careful longitudinal analysis
using individual scores and not just aggregate data may reveal that many
(and on some dimensions most) congressmen were not reaching their deci-
sions by incremental strategy. The corporate decision pattern may change
incrementally if some of the congressmen on one side of the median member
change their individual positions incrementally. If the distribution is
more "U-shaped" than "bell-shaped," and this is frequently the case in
the Senate, it does not require incremental changes by many individual mem-
bers to bring about incremental changes in the corporate decision pattern.
Furthermore, incremental changes in representatives' positions may some-
times be due less to their own pursuit of an incremental strategy of deci-
sion than to their response to the incremental strategy of others (interest
groups, the executive, and so forth).

The great potential of Lindblom's work may be more fully realized if
it is linked to longitudinal analyses of decision behavior.

23

analysis might be based on the assumption that consistency has its source
in doctrinal rather than pragmatic decision. If we have seen or are

seeing "the end of ideology,‘ and if pragmatic decision—making results
in lpeonsistency, why expect to discover recurring or enduring behavior
through longitudinal analysis?

TwO points should be made in response. First, doctrines are not
constructed with sufficient regard for the multiplicity of values, the
influence of changing conditions upon the efficacy and cost of means,
and the effects of changes in the latter on the ranking or weighting
of ends.l8 Thus those who originally committed themselves to consis-
tently follow a doctrine often find it too costly to be consistent in
terms of that doctrine.

Second, two of the reasons for saying that pragmatic decision
makers are generally inconsistent do not hold up under closer examina-
tion. Pragmatic congressmen are frequently said to be inconsistent be-
cause they sometimes support labor and at other times support management,
or often go along with certain civil rights groups but at other times do

not. Such behavior may in fact be quite consistent. There is nothingy

 

inconsistent about denying the relatively extreme demands of an interest

group while granting a number of its more moderate claims. The simplistic

 

lg'Ihe notion of a hierarchy of values is too simple. The ranking
or weighting of values must be related to changing conditions. Changing
conditions not only alter the efficacy and cost of instrumental values,
but may even reverse the direction of impact; much as "effectiveness" or
"fluency" in one politician may be socially beneficial, but "effective-
ness" or "fluency" in another politician may be unfortunate or even dis-
astrous: Or much as the direction of the effect of a budget deficit (or
surplus) on net goal attainment depends on variables such as the amount
of slack in the economy, price trends, and so forth. The inadequacy of
any comprehensive and detailed "hierarchy of values" does not mean however
that value decisions must be arbitrary, ad hoc, or atheoretical.

24
view of consistency as steadfast support of the demands of one interest
group or class and the equally steadfast rejection of those of other
interest groups or classes is the very antithesis of a key function of
legislatures--interest aggregation.

The sort of "consistency" that is eschewed by many congressmen--
steadfast or exclusive support of a single interest or value--is not the
consistency that behavioral scientists assume when they apply Guttman
scaling, multiple scalogram analysis, factor analysis, Kruskal's multi-
dimensional scaling, or smallest Euclidean space analysis in the study
of voting bodies. The application of these methods typically requires
that many decision makers pp£_be "consistent" in the sense of steadfast
or exclusive support of a single particularistic interest. These methods
would bear little fruit if there were but two "ranks" or if all members
were represented by but two or three points in a multidimensional space.

Another reason why it is sometimes assumed that congressional
decision-making is inconsistent and thus defies the discovery of regu-

' Editors and columnists

larities may be called the "editor's fallacy.'
frequently take a statement made by a congressman in one context and
contrast it with a statement or vote cast in another context and conclude
that his decision-making is so inconsistent as to frustrate the discovery
of regularities. Because of the congressman's need to communicate effec-
tively with many people who are not interested in an elaborate statement
of assumptions, conditions, or qualifications; because the complexity,
multiplicity, and subtlety of the components of decision and their rela-
tions usually defy easy explicit formulation; and because it is often

impolitic to comprehensively state one's criteria; congressmen only

partially state the rationale of specific decisions. Because he has not

25
made it clear that under different conditions he would act differently
(yet consistently), but instead has appeared to unqualifiedly support
some policy position, it is easy for the unsympathetic journalist to
criticize him for "inconsistency" when he (consistently) changes position
on some measure because of change in one or more relevant but unexplicated
variables. It is reasonable to hypothesize that congressmen adhere more
consistently to their relatively comprehensive and complex though largely
unexplicated cognitive and evaluative structures than to the brief sine
plistic ad hoc rationales offered to the public;[9 Thus it is also
reasonable to expect to find patterned decision behavior through longi-
tudinal analysis.

It will be demonstrated in another chapter that great changes in
rank on a dimension of policy behavior should not be arbitrarily equated
with "inconsistency" or "error" or absence of relatively enduring atti-
tudes. Such changes in rank can be associated with consistent attitudes

open to longitudinal exploration.

 

19It is necessary that we do not equate a change in position on a
policy dimension, or an apparent contradiction between a brief simplistic
ad hoc policy statement and a subsequent decision under changed circum-
stances with inconsistent or unpatterned behavior. It is tempting to
attribute unexplained "errors" to the respondent rather than to the
researcher's theoretical framework. But it is safest for the researcher
to initially assume that the "inconsistency" may be due to the limits of
his own perspective. It is partly by such an heuristic assumption that
his theoretical framework will be develOped.

III. THE ADVANTAGES OF LONGITUDINAL ANALYSIS

A General Summation of the Advantages

 

One may speak of the advantages of longitudinal analysis in either

general or specific terms. When speaking in general terms and thus sume

ming up the many specific advantages of longitudinal analysis, it may be

hypothesized that longitudinal analysis promotes:

l.

the discovery of recurring behavior and relatively
enduring attitudes so necessary to behavioral science;
cumulative science (now so conspicuously lacking);

the specification of concepts;

isomorphism;

the study of change and the conditions of change, and
thus helps identify multivariate relations that are
less likely to be identified in cross-sectional studies,
if they can be identified at all in cross-sectional
studies;

the development of concepts with both empirical and

theoretical import.

Obviously the points are interrelated. Several of them are little more

than different perspectives on the same matter. But perhaps the emphasis

and variety of expression serve to communicate more fully the nature of

the general advantages of longitudinal analysis.

26

27

Longitudinal Analysis Promotes the
Specification of Concepts

 

The first step in concept formation is typically the formation of
a vague image generated by intuitive combination of informal observations.
But concept formation must not stop with such a vague image. Fruitful
scientific concepts usually must be developed by repeated reference to
data. If the concept concerns phenomena assumed to endure or recur, it
is vital that reference be made to data from different points in time.
Failure to refer to data at different points in time may result in serious
misconceptions.
During this process of developing concepts by repeated reference
to data, longitudinal analysis often helps clarify:
1. the dimensionality of the behavior;
2. the level of the analysis (for example, manifest behavior
or underlying attitudes);
3. the aspects of behavior referred to (for example, valence
or rank order);
4. the degree of differentiation.
Whether a political scientist conceives of the opposition to
Supreme Court decisions (expressed in congressional voting 1957-59) as
opposition by Congress, a pappy (Republican), a ylpg of a party (Southern

Democrats), an inter-party coalition, or as a varied series of often

 

 

overlappipg vote groppings the exact nature of which varied considerably

with the specific issue (appointment of a judge, reversal of a decision,

 

limitation of a decision or of the jurisdiction of the Court) depends on
whether he relies on a few votes at one point which suggest opposition
by a party, or a few votes at another point which suggest Opposition by

a wing of a party, or a few at another point which suggest opposition by

28
a coalition, or careful longitudinal analysis which cautions the researcher
against any of the preceding and reveals a need to view the opposition in

multidimensional terms with the makeup of opposition varying with the

 

variation in specific issues. Thus, too narrow a time perspective often
reinforces a tendency toward oversimple conceptualization.

A cross-sectional approach permits the researcher to unconsciously
select a point in time which seems to substantiate his initial concep-
tualization of the phenomena under study. But, longitudinal analysis
often brings data to view that does not fit well with a priori notions.
Thus longitudinal analysis is more likely to force one to refine his
initial conceptualization. This may be frustrating in the short run but
rewarding in the long run.

If a researcher conceives of the attacks on the Court in the late
fifties as just another expression of "know-nothingism" and is content

' such as

to merely list a few other examples of "know-nothingism,'
McCarthyism, to suggest how he conceives the phenomenon, he fails to
learn what a simple longitudinal analysis could have shown--that much of
the anti-Court behavior in and out of Congress was not highly related to
support for McCarthy and that some of this behavior was even inversely
related to support for McCarthy. Such a longitudinal finding forces the
careful theorist to re-examine his initial concepts (or vague images).
In this case longitudinal analysis reveals that the proffered examples
of "know-nothingism" do not constitute a single behavioral dimension.
Learning this forces the scientific theorist to either abandon the vague
concept or to specify more clearly its meaning. Thus, longitudinal

analysis, by revealing empirical relations that contradict the vague

notion of recurring "know-nothing" behavior, encourages the theorist to

29

distinguish between levels of analysis: (1) manifest decision behavior,

 

and (2) attitudes that come together to generate that behavior. If the
theorist wishes to persist in using the term "know-nothingism," he must
recognize that any clear-cut continuity is not in the diverse patterns
of behavior. At most he can claim that each of the different examples
of "know-nothingism" are behavioral products of many factors, some common
and some unique. And, one of the common factors is an attitude labelled
"know-nothingism" that supposedly played a part in the generation of each
pattern of behavior (Each behavioral event [attack on the Court, McCarthy
support] is different either because of different loadings on the common
factors or because of the presence of unique factors or both.)1

In addition to promoting the specification of concepts by enabling

the researcher to learn more of the dimensionality of the manifest be-

 

havior, and by encouraging him to clarify the leyel of analysis, longi—
tudinal analysis promotes Specification by forcing the researcher to
state more clearly what aspect of behavior he assumes recurs or changes
(at a given level of analysis). After an event like Pearl Harbor the
valence or magnitude of most congressmen may change greatly on the selec-

tive service policy dimension but the rank order may change little. It

 

is essential that a researcher specify which aspect he assumes will
recur (or change).

As has been noted above (p. 8 ) longitudinal analysis also cautions
the theorist against conceiving policy behavior in dichotomous terms if

longitudinal evidence reveals that policy positions are more differentiated

 

lLongitudinal analysis is not the only unsung tactic for aiding the
process of conceptualization. The study of measurement models and the
(miniature) theories of behavior that underlie them can greatly clarify
problems of conceptualization. This is so even if calculations are never
made.

30
than this in the real world. Imagine that a student of legislative be-
havior concludes from an analysis of data in the second session of the
79th Congress that there were prg_vieWpoints," "twp sides," "lwp_groups
divided sharply on their attitudes toward organized labor." If other
political scientists assume that this minimum level of differentiation
is representative of congressional behavior, there will be serious con-
sequences for theory. If it is pgl_noted that a number of congressmen
who sided with labor on GPA or other questions on which labor's demands
seemed moderate and timely, voted against organized labor on a pro-labor
substitute for the Taft-Hartley Act and on overriding the veto of the
Taft-Hartley Act, it may be falsely concluded that congressmen are either
the agents of labor or of management and that none are in any sense ar-
biters. An interesting conclusion in light of the function of legisla-
tures as agencies for interest aggregation.

On each separate vote representatives must either support or not
support the claims of labor. In some sessions or Congresses one can
obtain "hard empirical evidence" that nearly all representatives side
with one or the other of two polar interest groups. And, representatives
have values, advocate and determine policies, and thus are not apart from
the group struggle. But if we take off our cross-sectional blinders we
can see that the position of many representatives on a recurring dimension
is not that of any particularistic interest group. Some representatives
do not serve as agents of particularistic interests (on all dimensions)
but play the crucial legislative role of aggregator of interests (if only
on some dimensions). But, this is not seen clearly in cross-sectional
analyses. To find evidence of interest aggregation often requires a longi-

tudinal effort to obtain a more realistic degree of differentiation.

31

The Advantage of Longitudinal Analysis for Replacing
Plausible but Scientifically Deficient Ex Post
Facto "Explanations" with Scientific
Explanations

 

 

 

 

A fairly typical and apparently accepted pattern of explanation
in cross—sectional studies of legislative behavior involves the calcula-
tion of a large number of correlations in a single session or Congress
and the selection of a limited number of these--the significant correla-
tions-~for use in "explanations." This method of "explanation" consists
of two parts. First, the researcher points to the statistical signifi-
cance of the selected correlation. Second, he notes the plausibility of

such a relation in light of an ex post facto selection of largely favor-

 

able quotes from a contradictory literature or an ex post facto selection

 

of those elements in the behavioral situation which seem intuitively or
self-evidently compatible with the statistical relationship.

The weakness of such ex post facto explanations revolves, in part,

 

around the selection of a fraction of the correlations. If one hundred
correlations are calculated (but not necessarily reported) and five are
said to be significant at the five per cent level of confidence (would
occur by chance not more than 1/20 times), any of the correlations re-
ported as significant could very well have occurred by chance, although
one would not realize this if he were not told of the many correlations
winnowed out.
If we calculate one hundred correlations and find that ten are

significant at the five per cent level, what can we conclude? Perhaps
one half of these relations should be attributed to chance. Combining

such statistical evidence with ex_post facto rationales does not provide

 

us with scientific explanations. However, two strategies of scientific

32
explanation often enable us to use the above sort of correlations and
yet have a firmer basis for deciding whether the relations occurred by
chance or represent behavioral regularities. First, if before he com-
putes the correlations for time Y, the researcher records predictions of
behavior at time Y based on analyses at time X and records them in such
a manner as to make them more subject to disproof than to rationalization;
the researcher has less reason to fear that he has unconsciously selected
one of several possible rationales, interpretations, or explanations be-
cause it is the one that fits his particular findings. Second, uppredicted
significant correlations which one does not know whether or not to attri-
bute to chance, but for which there is some substantive support, may be
treated as promising hypotheses--that is, predictions of relations to be
found in another Congress. Whether the researcher selects the first or
the second strategy, longitudinal analysis is essential.

With the advent of high-speed computers, the increased availability
of data checks containing over 100,000 responses for a single Congress,
and the use of computer programs such as MIT's Pattern Search and Table
Translation Technique, the chances have increased astronomically that
researchers will have the computer calculate hundreds or thousands of
correlation coefficients and then select for further analysis that small
fraction which are deemed statistically significant. If students of con-
gressional behavior are not to confuse the combination of correlations

which meet a statistical test of significance (yet may easily be due to

 

 

chance) and exypost facto selection of rationales with scientific explana-
tion, there must be more recognition of the need for longitudinal

analysis.

33
Guttman and others have, of course, previously warned against the
practice of inferring some "causal" connection from items from the same
cross-section of time. In 1950 Guttman noted that:
Theoretically one can never infer "cause" from the
cross tabulation of two attitudes or opinions

gathered at the same time. However, 0 inion ana-
lysts make this inference quite often.

The Advantage of Longitudinal Analysis for
Testing the Validity of Measures

If one is asked how he knows that some measure is valid (actually
measures what it purports to measure), he faces a question that often is
not explicitly raised and that seldom is answered in studies of legisla-
tive behavior.

One obvious reason why little or no evidence of the validity of a
measure is offered is that there may not be any other (relevant concur-
rent) behavior, in the time period of the study, which can act as a
criterion (i.e., which can allow a pragmatic test of concurrent validity).
Usually studies of legislative behavior cover a period of only one or two
years. Thus, no adequate test of concurrent validity may be possible
(given limited research funds, reluctance of congressmen to submit to
many kinds of tests, and limited roll call or interview data for a short
time period).

An alternative test of pragmatic validity and one that has two
advantages over a test of concurrent validity is predictive validity. By
predictive validity we mean the adequacy of a measure (such as Rosenau's

measure of indiscriminate hostility) to distinguish individuals (senators)

 

Measurement and Prediction, Vol. IV of Studies in Social Psychology
in World War II (Princeton: Princeton University Press, 1950), p. 155.

 

34

who will differ in their subsequent behavior (1953 or later).3

For example, the predictive validity of Rosenau's measure of
indiscriminate hostility may be tested by seeing if it usefully distin-
guishes senators who will differ in their subsequent support of, or
opposition to, McCarthyism or those who practice it. If Rosenau's
measure is valid, a disproportionate number of those he classified as
"Indiscriminates" should show up among the most extreme supporters of
McCarthy during the censure efforts. We would expect, for example, that

' who were still in the Senate in

nearly all of the "Indiscriminates,'
1954, to be among those in the most extreme pro-McCarthy ranks (on a
scale of support for McCarthy).4

Predictive validity, I have said above, has two advantages over
concurrent validity. First, it makes a test of validity possible even
if the period from which the original measure is derived does not offer
a relevant criterion. Because it frequently is the case that one can-
not find an adequate criterion in the same time period as that from.which
the measure to be tested is derived, it is a longitudinal search for a
criterion that offers hope for a test of validity.

The second advantage of predictive validity (i.e., the use of a
criterion derived from a different time period) is that whether our im—
mediate interests are primarily scientific or practical, we need to
identify behavior that endures for more than one or two years. While a

longitudinal search for a criterion may offer evidence of enduring or

recurring behavior, a concurrent criterion does not.

 

3James N. Rosenau, "Senate Attitudes Toward a Secretary of State,"
in Legislative Behavior: A Reader in Theory and Research, edited by J. C.
Wahlke and H. Eulau (Glencoe: Free Press, 1959), pp. 332-47.

4

dictions.

 

Chapter IV will offer evidence of the feasibility of such pre-

35

The Advantage of Longitudinal Analysis for
Suggesting the Relative Importance
of Variables

 

 

Students of legislative behavior who argue the importance of
(nominal) party affiliation often offer as evidence the predictive power
of party affiliation. Frequently the claim is stated as James A.
Robinson does in his Congress and Foreign Policijaking. There he says
that ". . .certain roll call studies have indicated that party is the
best single predictor of a legislator's vote."5 Thomas and Lamb claim
that:

Probably the single most important clue to the
attitudes of a legislator is his party membership.
David B. Truman's study of party influence in the
Eighty-first Congress found that "the party label
evidently is the single most reliable indicator of
congressional voting behavior. . . .6

The predictive value of a variable may be viewed as one measure
of the variable's importance. But, there is less justification for as-
suming that the predictive value or power can be gauged by a cross—
sectional study. There is even less justification for assuming that party
is the pe§£_predictor of roll call behavior, when one has not even at-
tempted to test the predictive power of such constructs as bloc member-
ship or scale type. When one does what Turner, Truman, Thomas and Lamb,
and others did not do--1ook at the dimensions of congressional behavior
and compare the predictive power of the ready-made formal aggregates

(nominal party affiliation, region, etc.) with the predictive power of

behavioral types or groupings--one finds that their oft repeated

 

5(Homewood, 111.: Dorsey Press, 1962), p. 100.

6Congress: Politics and Practice (New York: Random House, 1964),
pp. 36—37.

 

36

professional lore does not hold up. There is no convincing evidence
that party affiliation is the best predictor of roll call behavior on
any recurring substantive dimension or combination of dimensions. Al-
though prediction from party is generally superior to prediction from
socio-economic background, socio-economic status, and constituency vari-
ables; it is inferior to prediction from scale-type and from multidimen-
sional types derived from recurring behavior patterns.7

If one wishes to preserve the good name of party as a predictor,

he may, of course, claim that party, behaviorally defined, is generally

 

a better predictor than socio-economic background, socio-economic status,
constituency variables, or nominal party affiliation. 8

Longitudinal analysis of congressional behavior, however, reveals
that nominal party affiliation is less important as a predictive variable
than the above mentioned authorities claim, but more important than
Miller and Stokes assume to be the case in the making of foreign policy.9

The Advantages of Lopgitudinal Analysis
for the Study of Change

 

 

The theoretical advantages of longitudinal analysis, particularly
of the panel study, for the analysis of change require little comment.
Most readers will readily agree that unless we measure individual be-

havior over time we have to rely on a comparison of marginals to estimate

 

7See Chapters IV—VII below.

8Both nominal party affiliation and behavioral party membership may
often be useful for practical prediction and perhaps aid in the development
of hypotheses. But, those who are interested in "party" as an independent
variable would do well to distinguish between nominal party affiliation and
party identification. One type of evidence for such a distinction will be
offered in Chapter VI.

9See The American Political Science Review, Vol. LVII (March, 1963)
pp. 45-56 and Chapters IV-VII below.

 

37
change. But the marginals can only tell us of the net shifts, not the
actual turnover. The net shift in the marginals may be negligible and
suggest little or no change, but the unrevealed turnover may be so large
that the user of marginals has been seriously mislead}.

The skeptic has reason, however, to ask what specific research
problem related to an important area of theory requires longitudinal
analysis of change. One vital area of theory which has recently become
much more open to exploration through longitudinal analysis of legisla-
tive behavior is the process of conflict. One might hypothesize, for
example, that in the period preceding the Civil War there was an increas-
ing simplification of the dimensions of congressional decision. Substan-
tive issues that were not initially perceived and determined primarily in
terms of the developing conflict between North and South became increas-
ingly tied to the sectional conflict as the latter became more salient.
Even if the original purpose of a legislative proposal were unrelated to
the North-South conflict and if the effect of the proposal on that contest
were slight, that contest was so close and the stakes so large that such
policy objectives were increasingly subordinated. Logically consistent
with this is the hypothesis that increasing bipolarization was associated
with the above simplification of the issues.

An anti-behavioralist would be right in arguing that no single

measurement model even if used longitudinally could offer a complete view

 

10For example:
Second Time

First Time Pro Anti Total
Pro 30 17 47
Anti 18 29 47

Total 48 46 94

38
of the above conflict. No model can fully portray reality. As scientists,
however, we do not seek to recreate historical reality in all of its de-
tail but to identify regularities which map (not reconstruct) reality.
A map that contains much less than one per cent of the details of the
area to be traversed is much more useful than one that contains so much
detail that the path to one's goal is obscured. The following measure-
ment models if used longitudinally to study the process of developing
conflict in the pre-Civil War period offer the following different but
complementary perspectives:
1. At a point in the anti-bellum period in which content

analysis of verbal behavior suggests that the conflict

was not yet explicit and intense, Guttman or MSA scales

of roll-call behavior should reveal many scale dimensions

which are not highly correlated with other scale dimen-

sions.l; As content analysis of verbal behavior reveals

more explicit and intense conflict, the scales of roll

call behavior should reveal fewer scale dimensions that

are uncorrelated with the over-all conflict. Fewer scales

should be unimodal and a larger percentage should be in-

creasingly bipolar. As the researcher approaches 1861

he may find fewer middle ranks containing more than two

members, and the Opposite or extreme ranks should be

separated by differences on more items or on items that

involve larger stakes.

2. The factor analysis of congressional roll call items should

 

11MSA = Multiple Scalogram Analysis. See Appendix B.

39
reveal that fewer orthogonal (uncorrelated) factors
are required to account for most of the variance, as
the nation moved closer to Civil War. Alternatively,
the R—factor whose content best repesents the North-
South conflict will account for an increasingly larger
percentage of the variance.

3. The factor analysis of senators or representatives should
reveal that the bipolar Q-factor that best reflects
North—South conflict should account for an increasing
percentage of the variance.

4. If the researcher applies Kruskal's multidimensional
scaling or smallest Euclidean space analysis to a longi-
tudinal series of matrices of correlation or agreement
among congressmen, he should find that a decade or so
before the Civil War there were many diverse clusters
and isolates, but that as 1861 approached the points
(representing legislators) become more closely clustered
into two widely separated groups. Furthermore, he will
find fewer distinct clusters in spite of the fact of
increasing membership.

The above loose hypotheses may seem obvious to those readers whose

common sense or social psychology corresponds. But the plausibility of

hypotheses and their merit are not identical.12

 

12George Galloway in his History of the House of Repgesentatives
states that "party discipline and the power of the caucus were weakened by
sectional disputes over the slavery question as the irrepressible conflict
approached" (New York: Thomas Y. Crowell Company, 1961), p. 131. Others have
been more emphatic in discussing "the sharp split" or "the breaking into
fragments" of the Democratic party. But we must await longitudinal measure-
ment before we can answer questions about how much, how fast, and in what
manner congressional party unity was weakened, and how much, how fast, and
in what manner the North-South congressional bipolarization took place.

 

40

The following generalizations by V. 0. Key suggest something of
the content of one type of cleavage that preceded the North-South con—
test and may be interpreted by some as implying that the simplifications
of the dimensions and bipolarization during the period of East-West con-
flict was comparable to that during the North-South conflict:

As the slavery issue grew in intensity and salience,
the old cleavage between East and West was replaced by
a new sectionalism. The pattern of sectional politics
became one of conflict between North and South. As the
"solid West" disintegrated, "rival societies, free and
slave, were marching side by side into the unoccupied
lands of the West, each attempting to dominate the
back country."13
Thus, Key's remarks suggest that we moved from two—sided conflict be-
tween East and West to two-sided conflict between North and South. The
previous set of generalizations, however, assumed less simplification
of the dimensions and less bipolarization during the period of East-
West conflict than during the period of North-South conflict.

The purpose of these few pages on the advantages of longitudinal
measurement for the study of change has not been to endorse a set of
hypotheses about the process of conflict but to suggest the difficulty
of scientifically choosing among or refining hypotheses without engaging
in longitudinal analysis.

It should also be noted that while the more accurate measurement
of change may greatly aid science, scientists are not interested in change

per se, Instead, scientists are interested in the increased opportunity

(afforded by the study of change) for inference about the regularities

 

that underlie change.l4

 

13V. 0. Key, Politics, Parties,and Pressure Groups (New York: Thomas
Y. Crowell Company, 1958), p. 254.

 

1""An example of how the study of changes in a condition (party control
of the administration) can reveal underlying regularities will be offered in
Chapter VI.

The Subject Matter of Chapters IV, V, and VI

Thus far I have (I) discussed the general problem, (II) described
the nature of longitudinal analysis, and (III) hypothesized advantages
of longitudinal analysis. Chapters IV, V, and VI offer original examples
of longitudinal analysis intended to serve as evidence of advantages
hypothesized in Chapter III.

In each of these examples successive events are hypothesized to
involve stimuli which are similar enough in some respect to elicit pat-
terns of response that are also similar (in some hypothesized respect
such as rank order, or a significant difference between members of dif-
ferent parties). That is, in each chapter I hypothesize that recurring
aspects of stimulus situations elicit (measurable) recurring aspects of
response behavior. The three examples are also similar in that each
(1) has United States senators as its population, (2) uses roll call
data, (3) seeks differentiation through ordinal measurement, (4) com-
pares measures at different points in time, and (5) is relevant to the
study of decision making, party, and political strategy.

These examples differ, however, in that each is designed to offer
a different application of longitudinal analysis and to emphasize a dif-
ferent type of advantage. Thus it is hoped that Chapters IV, V, and VI,
taken together, will evidence the broad utility of longitudinal analysis
for measurement, prediction and explanation.

The first example (Measurement of Right-Wing Indiscriminate Hostil-

ity) will abstract recurring aspects of behavior (rank order) common to

41

42
such events as the 1949-52 attacks on Secretary of State Acheson and the
December, 1954 censure vote support for McCarthy. It will emphasize the

utility of longitudinal analysis for measurement.

 

The second example (An Exploration of Urban Affairs Decision Making)
will abstract recurring aspects of behavior related to both the 1962
vote on the Department of Urban Affairs and the 1965 vote on the Depart-
ment of Housing and Urban Development. It will emphasize the utility

of longitudinal analysis for prediction.

 

The third example (Continuity and Change in Foreign Policy Position)
will abstract recurring aspects of behavior common to the 1952-53 Truman-
Eisenhower transition and the 1960-61 Eisenhower-Kennedy transition. It
will emphasize the utility of longitudinal analysis as an aid in

explanation.

 

The stimulus situation in all of its detail is not perfectly re-
peated in any of these examples: and it is because of this that such
analyses are relevant to the study of politics in the real world. Uptll.
political scientists can identify recurring patterns of behavior elicited

by recurripg types of real-world stimuli, there can be no science of

politics.

IV THE MEASUREMENT OF MCCARTHYISM:
A LONGITUDINAL ANALYSIS

The general, long—term purpose of this chapter is not "history
for history's sake" but the development of intellectual tools to
better measure future expressions Of indiscriminate hostility.

The specific purposes of this paper are to reveal how longi-

 

tudinal analysis can offer:

1. Increased differentiation so important to conceptualiza—
tion and the expression Of relations.

2. A test of the relations among four different types of
indicants (and thus Offer evidence Of the validity of
scale analysis Of roll call votes and content analysis
of speeches).

3. Evidence that the analyzed behavior is neither random nor
ephemeral but patterned and recurring and thus Open to
fruitful scientific analysis.

4. Evidence of the feasibility of scientific prediction Of
"representatives" voting behavior from their verbal be-
havior (a type of prediction that is important in a com-
plex representative democracy).

The four types of data which will be used are:
1. 1954 Senate roll calls on the censure of Senator McCarthy.

2. James N. Rosenau's dichotomization of senators as either

43

44

indiscriminately hostile or p93 indiscriminately hostile
toward Truman's Secretary Of State (based on the content
analysis Of their statements in the Congressional Record
1949—52).l

Margaret Chase Smith's 1950 (anti—McCarthy) "petition"
data.

Richard Rovere's rating Of Welker, Jenner, and McCarthy as
the three most extreme indiscriminate right-wingers or
McCarthyites (based on his intensive qualitative analysis

of many observations not included under 1, 2, and 3).2

The subjects are Republican senators common to the period Of

(1954).

the Rosenau study and to the period of the McCarthy censure votes

The scale hypothesis and test of relations among indicants are

as follows:

 

The three Republican senators that Rovere identifies as most

extreme will be among the thirteen that Rosenau identifies as indis-
criminately hostile;3 these Indiscriminates will be among the fifteen-
to-twenty most extreme Republican supporters of McCarthy identified

by the roll call analysis of the McCarthy censure votes, and these

 

$92, cit., p. 346.

2Senator Joe McCarthy (New York: Harcourt Brace and World, 1959).

3Subsequent study revealed that only ten of the thirteen identified

by Rosenau as "Indiscriminates" were still in the Senate at the time Of
the censure votes.

45

in turn will be included among those who did not endorse Margaret
Chase Smith's "Declaration of Conscience." In other words there
should be a cumulative pattern of behavior on the four types Of
indicants (Rovere's dichotomy, Rosenau's dichotomy, the scalable
censure roll calls, and Smith's petition data). Figure 3 portrays
the hypothesized relations.

The coefficient Of reproducibility (CR) will exceed .90.

Specification of the Concept

 

McCarthyism is related to a broad set Of style concepts. These
include Parsons' distinction between affectivity and affective-neutra—
lity (one of the pattern variables borrowed by Almond),4 Eric Hoffer's
"true believer,"5 Rokeach's "Opinionated rejection,"6 Shils' frus-
trated and hostile legislator,7 Pritchett's Know—Nothingism,8 aspects

Adorno's authoritarianism,9 and Lowenthal & Guterman's distinction

 

4Talcott Parsons, The Social System (Glencoe, 111.: Free Press,
1951).

 

5Eric Hoffer, The True Believer (New York: Harper, 1951).

 

6Milton Rokeach, The Open and Closed Mind (New York: Basic
Books, 1960).

 

7Edward A. Shils, "The Legislator and His Environment, " The
University_of Chicago Law Review, Vol. 18 (1950— 1951), pp. 571-584.

C. Herman Pritchett, Congress Versus the Supreme Court 1957-

wVwfirfi Vfi._fi_fi_‘, va ‘7

1960 (Minneapolis: University Of Minnesota Press, 1961).

9T. W. Adorno, Else Frenkel—Brunswik, D. J. Levinson, and R. N.
Sanford, The Authoritarian Personality (New York: Harper, 1950).

744—039 .510 ”:5, Cu. —Anvmwn.:..~:.nw WOCQN- ”UPTSN ”Tea: .JNH

Hafiz. I1 I . - J . 1‘...
fi_unn...p‘..53 3‘. .fi.,:.p.:.iru..l aw za‘nliufl—nvxufipu HuHLHUfiuxn—th-e-POU

n p.»— — IadI—nunh...nbn has. I

u.-~v.,..-.. a

em

mwoonrmmHNma memnHOSm >Bonm Hsawombmm mom mmbmnonm

mocw Hwomm om HDmHomnOHm

Hammomm. mHHmon HomemoHHBHomnm mnemonnmm GHQ non mononmm
mmomnmw mcooonn womnHHHnw nOI Znnmnnrw oanIZoomHnrw
mow mom onmnnwnm ammo >ormmoo on omomcnmm amowmnmneoo
Om Znomflnrwwma
m H Ennmunrwwnmm + + + +
H
m a
2 HH Hoawmoswawomnmm I + + +
w a
m m HHH SOHO UHmnnHBHomnm I I + +
m Zoomnnrw mnemonnmmm
n
o o H< rmnm AHDanaHnm :mmen I I I +
m ammnvmnzv ooooDmonm
n

m < mmmpw owooomsnm I I I I

 

mwOOmCmm nrmnm mum BCHnHon omomcnm nOHH omHHm zen: mwmmmneom smfimHanm to own ownmwo
mammnmn mwmmmnmonemnwoo arm: nrm mH<m smuon nwomm Ho anew mHBOHm mwmcnm.

carmmm mum nrm HmmmIZOQmHnrwwnm Hoawmnwwawsmnmm om mommomclm mncmw. Hmnrmm mono mHH
nrennmmo Om wommomclm Hoawmonwswomnmm.

wwmcnm w

47

between the anti-democratic agitator & the reformer."lo McCarthyism
is closest, however, to Rosenau's concept of indiscriminate hostility.
Indiscriminate hostility does not merely mean Opposition or criticism,

but refers to the quality (or style) Of Opposition. Rosenau's impor-
tant distinction between discriminate hostility and indiscriminate
hostility is based on the percentage of hostile references that are in
the personal qualities and symbol-collectivity categories.11 He states
that not all of the critics of Secretary Acheson "evidenced an inclina-

tion to treat him as a scapegoat for the insecurities Of a troubled

world."12

A significant distinction can be made between those who tended
to confine their criticism to Acheson's performances and those
whose hostility tended to be more indiscriminate and embraced
him in all his capacities. These tendencies have been opera—
tionalized by attributing a discriminately hostile attitude

to those Senators who recorded less than 10.3 per cent of their
hostility potential in the personal qualities and symbol-
collectivity categories, and by ascribing an indiscriminately
hostile attitude to those who registered more than 10.3 per cent
of their unfavorable references in these two categories. 3

 

10"Portrait of the American Agitator," Public Opinion Quarterly,
1948, 12, pp. 417-429.

11"The symbol-collectivity category was established to account for
references in which Acheson was identified neither as an individual
possessing certain qualities nor as an Official performing certain acts,
but as a symbol Of an ambiguous phenomenon ('the Truman-Acheson policy')
or as a member Of an indeterminate collectivity ('the Acheson crowd')."

lzlbid., p. 337.

l3Ibid., p. 337.

48

Rosenau Offers the following examples Of indiscriminate

hostility:

[The Indiscriminates] perceived that "the Secretary of State is
managing our Armed Forces, our political life, our private press
[Jenner]," that "Mr. Acheson today is not only writing our
foreign policy but he more than any other man in the United
States is dictating our domestic economy to a process which
put "Mr. Acheson in full charge of the throttle [Malone],"
so much so that it was "turning . . . the Government of ours
into a military dictatorship, run by the Communist—appeasing,
Communist-protecting betrayer of America, Secretary of State
Dean Acheson [Jenner]."
I have studied Acheson's public utterances sidewise, slant—
wise, hindwise, and frontwise; I have watched the demeanor
Of this glib, supercilious, and guilty man on the witness
stand; I have reflected upon his career, and I came to
only one conclusion: his primary loyalty in international
affairs runs to the British labor government, his secondary
allegiance is to the Kremlin, with none left over for the
country of his birth. The only trouble Acheson ever en-
counters is when Socialist-British and Russian-Communist
policy diverge. . . . Then he reluctantly follows the lead
from Socialist-London [McCarthy].14

Rosenau also categorizes McCarthy's descriptions of Acheson as

"the great Red Dean of fashion" and "the elegant and alien Acheson-

Russian as to heart, British as to manner,‘ as examples of indis-

criminate hostility.15

An example of indiscriminate hostility that refers less to

personal qualities than to some vague collectivity is McCarthy's ref-

16

erence to "the Acheson—Hiss—Yalta crowd."

14Ib1d., pp. 340-341.

lslbid., p. 341.

l6Ibid., p. 341.

49

"In addition to positing the Operation Of hyphenated collectivi-
ties as explanatory of all that was undesirable, the Indiscriminates
perceived that hidden groups and mysterious forces were causing untold

damage."l7

. . . Acheson was viewed as "the very heart of the octopus

[McCarthy]," as the expression of "a sinister, ruthless

undertow [Malone]," as the product Of "the inner circ1e .

. . with its hidden rooms and hidden corridors [Jenner]."

For the Indiscriminates, in short, the course of events was

determined by "a rather sinister monster of many heads and

many tentacles, a monster conceived in the Kremlin, and

then given birth by Acheson, with Attlee and Morrison as the

midwives, and then nurtured into Frankenstein proportions by 8

the Hiss crowd, who still run the State Department [McCarthy].

The scale variable hypothesized to be common to the four kinds
of data is conceived as a behavioral product Of multiple attitudes;
not as an attitude. Although there appears to have been widely shared
perception of what underlying issue was common to the (above—mentioned)
petition, speeches and votes, it is assumed that the behavioral re-
sponses to that issue were generated by more than a single attitude.

We are not concerned here with all expressions Of right-wing
indiscriminate hostility. Any cumulative ordering obtained below will
not adequately measure right-wing indiscriminate hostility toward

Blacks, Union leaders or Jews. The concern is with indiscriminate

hostility directed at "leftist" management of national security

questions.

17Ibid., p. 341.

181b1d., p. 341.

50

The Theoretical Import of the Concept

It is hypothesized that the conceptualization Of indiscriminate
behavior at this level of generalization Offers a more nearly optimal
juncture of theoretical import and empirical import. Clearly this
level of conceptualization Offers more theoretical sc0pe than a
specific measure of senatorial behavior toward Acheson or McCarthy.
And if the four different types of indicants scale together, the
increase in theoretical sc0pe is not at the expense of empirical im—
port. TO seek even greater theoretical scope by treating race,
religion, national security, and labor questions as one issue—
dimension, would sacrifice empirical import. Congressmen simply are
not arrayed along but one dimension on these issues.19 And neither
are their constituents.

The theoretical importance Of the concept can better be seen by
noting the sc0pe of the relations that it has been hypothesized to
enter into as both a dependent and an independent variable. The

following inventories of hypotheses are not meant to be exhaustive,

 

19Duncan MacRae, Jr., Dimensions of Congressional Voting
(Berkeley and Los Angeles: University of California Press, 1958), pp.
212ff. Warren Miller and Donald Stokes, "Constituency Influence in
Congress," The American Political Science Review, Vol. XVII, No. 1
(March, 1963), pp. 45—56. Charles Farris, "A Method of Determining
Ideological Groupings in the Congress," Journal of Politics, Vol. 20.
NO. 2 (May, 1958), pp. 308-338.

 

 

20Miller and Stokes, ibid. Donald E. Stokes, "Spatial Models Of
Party Competition," The American Pplitical Seience Review, Vol. XVII,
NO. 2 (June, 1963), pp. 368—377. Michael Rogin, "Wallace and the
Middle Class: The White Backlash," Public Opinion Quarterly, Vol. XXX,

NO. 1 (Spring, 1966), pp. 98-108.

 

51

but only suggestive of the sc0pe of relations involving indiscriminate
hostility.
When concerned with the explanation of senators' indiscriminate

hostility—~that is with IH as a dependent variable--we may relate IH

 

to:
I. The Situation as Immediate Antecedent of Action (including
opportunities, limits and prudential considerations)
A. The Position Of
1. Constituents on
a) Indiscriminate hostility
(1) Direction of Opinion
(2) Saliency Of opinion
b) Other factors that influence electoral outcome
(probably net effect Of constituents' party
images, issue and candidate orientations on
electoral outcome—-victory or defeat--if
candidate does not engage in indiscriminate
hostility)
2. Party leaders (for example, Taft's calculated

tolerance or encouragement Of McCarthyism)21

 

a v w w

21"Once McCarthyism took hold, Taft . . . encouraged its author
to keep on with his accusations on the ground that the law of probabili-
ties could not in the long run fail him. 'If one case doesn't work, try
another,‘ he told McCarthy." Rovere, Op. cit., p. 136. McCarthy ". .
was a pure delight now to the campaign committees, and the Republican
organizations were in hot competition to have him comeiruwith a load Of
documents on anyone who was giving them trouble. Approval came from high
places. Robert Taft gave it out that in his Opinion 'the pro-Communist
policies Of the State Department fully justified Joe McCarthy in his
demand for an investigation." Rovere, p. 179.

52

Opinion-makers (for example, the ". . . group Of

reporters . . . who worked closely with [McCarthy]
and his staff," and " . . . the reporters, es—
pecially those from the wire services, [who tried]

to stimulate conflict [subversive to the Senate

folkways])"22

B. The relation of the Senator's party membership to that

of the target Of indiscriminate hostility (for example,

Acheson) and to that Of those expressing indiscriminate

hostility (McCarthy, Jenner, and so forth)23

c. The clarity, vagueness or ambiguity of situational norms

II. The Senator's Personality Processes and Dispositions

A. Motivational Bases Of Attitudes

l.

2.

3.

Object appraisal (reality testing)
Mediation of self—other relationshps (social ad-
justment)

Externalization and ego defense

B. Attitudes Toward

l.

2.

3.

Communists and those accused of being communists
The norms of democracy, fair play, etc.

The importance Of party victory

 

22

Matthews, Op. cit., pp. 202, 217.

23Rosenau, Op. cit.

53

III. The Senator's Social Environment as Context for the

Acquisition, Maintenance, and Change Of Attitudes

A. The Information that His Environment Provides about
Relevant Social Objects (e.g., targets and exponents
of McCarthyism).

B. The Information that His Environment Provides about the
Prevailing Social Norms (of positive and negative re-
ference groups) Concerning Relevant Social Objects
(e.g., targets and exponents Of McCarthyism).

C. Life Situations Conducive to Formation of "Authoritar—
ian Personality," Deflating Self—esteem, Provoking Dis-
placed Aggression
1. Personal life situations (for example, the senator's

position, accomplishment, prestige do not match
that of other senators, or meet the standard set by
a starving ego)

2. Defeats or decline for his (subnational) group
(party, ideological group, socio-economic group-
ing)24

3. National reversals (defeats, retreat, stalemate,
other evidence of declining national power) that
frustrate and humiliate those whose self-esteem

strongly depends on national power and respect

 

24Rovere, Op. cit., p. 21.

54

(examples——loss of Chine, American retreat in
Korea)25
IV. Remote Facts of History, Politics, Economics, etc., that
Contribute to the Distinctive Features of the Senator's
Environment and of the Immediate Situations that He

2
Encounters.

Any of these classes of determinants could be more thoroughly
developed. For example, Table 1 hypothesizes types of pressures
(on senatorial choice of style) generated by the permutations of I,
A, l, a), and I, A, 1, b). Many of McCarthy's supporters in the Senate
perceived the position of constituents on indiscriminate hOStility as
favorable, and the probably electoral outcome (without indiscriminate
hostility) as unfavorable (see lower left cell in Table 1) . As
Rovere puts it:

Into [the world of the daft and the frenzied] came large

numbers of regular Republicans who had coolly decided

that there was no longer any respectable way Of unhorsing

the Democrats and that only McCarthy's wild and conscience-
less politics could do the job.27

III, C, 2, points tO an additional less conscious, less direct

 

25Rosenau, o . cit., p. 342.

26Adapted from M. Brewster Smith, Determinants of Anti—Semitism:
A Social-Psychplogical Map (New York: Anti—Defamation League of B'nai
B'rith, 1965,) and Rosenau, Op. cit.

27Rovere, Op. cit., p. 21.

0‘0 71

WHO‘S)

O

(D B

3‘ n H: 2:

C O

('1‘

HO-ZJI-l

m
*< n H- F4 P-

55

TABLE 1

TYPES OF CONSTITUENCY INFLUENCE ON SENATOR'S CHOICE OF STYLE

 

 

m

['1’

H- CO

Favorable

(victory)

Unfavorable

(defeat)

Position of Constituents on Indis-

criminate Hostility

 

 

 

 

Favorable Unfavorable

IH. 227 Electoral
necessary situation dis-

couraging IH

E

lectoral NO advantage
situation in IE
encouraging

1H

 

56

sort of constituency influence on senatorial style. Those senators
who strongly identified with the Republican party, whose self—esteem
was deflated by the defeats from 1932-1948, and who felt subjected to
the arbitrary authority of a misguided electorate and a liberal
Opposition party, were more likely to engage in indiscriminate hos—
tility. The greater the policy differences between the "losers" and
the "winners" and the more salient the issues on which they differ,
the greater the frustration of the "losers" and the greater the pro-
bability that they will seek emotional satisfaction through indis-
criminate attacks. The extent to which frustration leads to indis-
criminate hostility rather than socially useful "reality testing" and
coping with national problems, depends in part on (1) how well—
equipped, intellectually, the senator is for "reality testing," and
(2) how well he has internalized the vital norms that help keep
competition on substantive issues from being severely compounded by
personal conflict.28

The purpose here is not, however, to thoroughly develop each
point in the above outline, but simply to suggest the sc0pe Of the
relations into which the concept enters.

As an independent variable indiscriminate hostility has been

hypothesized to have effects on all three branches Of government, the

 

28V. 0. Key, Jr., Ppblielppinion S Americen Democracy (New York:

KnOpf, 1961).

 

57

electorate and foreign publics. Beer et al.,29 Shils,30 Matthews,31

and Almond32 are among those who have discussed the relation between
style of expression and the functioning of legislative bodies. Almond,
for example, argues that both the function of interest articulation
and the function of aggregation are performed in part by legislative
bodies and that the style of articulation has important effects on
aggregation.33 He states that

With regard to the style Of interest articulation, the more
latent, diffuse, particularistic, and affective the pattern of
interest articulation, the more difficult it is to aggregate
interests and translate them into public policy. Hence a
political system characterized by these patterns of interest
articulation will have poor circulation between the rest of the
society and the political system, unless the society is quite
small and has good cue-reading authorities. On the other hand,
the more manifest, specific, general, and instrumental the
style of interest articulation, the easier it is to maintain
the boundary between the polity and society, and the better
the circulation of needs, claims, and demands from the society
in aggregable form into the political system. A political
system with an interest articulation structure and style of this
kind can be large and complex and still efficiently process raw

 

29Samuel Beers, ep_el., Patterns Of Government (1st ed.; New
York: Random House, 1958), pp. 22-25. '

 

3OShils,.ep. cit.

3102. cit., pp. 97—99, 217.

32Gabriel Almond and James S. Coleman, The Politics of the
Developing Areas (Princeton: Princeton University Press, 1960).

 

33Ibid., pp. 33, 36, and 40.

58

demand inputs from the society into outputs responsive to
the claims and demands Of that society.34

Key,35 Kris and Leites,36 Lipset,37 Shils,38 Stouffer,39 and

others help us relate style Of leadership to the electorate's:

 

1. Personality Processes and Dispositions4O
A. Motivational Bases of Attitudes (for example style
of leadership can foster reality testing or ego
defense)41
B. Attitudes, beliefs, expectations, images

1. Feelings about the effectiveness Of democracy42

 

341bid., p. 36.

35V. 0. Key, Jr., Public Opinion and American Democraey pp, cit.

36Ernst Kris and Nathan Leites, "Trends in Twentieth Century
Propaganda," reprinted in Berelson and Janowitz (eds.), Public Opinion
and Communication (Glencoe, 111.: Free Press, 1950), pp. 278—288.

 

37Seymour Martin Lipset, Political Man (Garden City, N.Y.:
Doubleday and Company, 1960).

 

38Shils, pp, cit.

39Samuel Stouffer, Communism, Conformity, and Civil Liberties
(Garden City, N.Y.: Doubleday, 1955).

40Important leaders affect the attitudes of this and the next
generation of voters through relatively direct effects on their social
environment as the context for acquisition, maintenance, and change of
attitudes. Such leaders have quite indirect effects on the voters of
the (remote) future. That is, present—day leadership behavior will be
among the remote antecedents Of electoral behavior in the distant future.

41Kris and Leites, pp, cit., especially pp. 286-288.

42Lipset, pp, cit.

59

2. Feelings about the legitimacy Of democracy43

3. Feelings about the legitimacy of specific office-
holders44

4. Faith in people45

5. Images Of the U.S.S.R., Socialist Britain, etc.46

11. Behavior47
A. Voting Behavior

B. Mass Action

Perhaps the theoretical import of indiscriminate hostility (as
an independent variable) is best seen in the work of Key and that of
Shils. The latter, after discussing the sources, objects, and conse—

quences of indiscriminate hostility, concludes that:

 

43David Easton, A Systems Analysis of Political Life (New York:

Wiley, 1965), pp. 278-310. Also, Lipset, pp. cit.

44Easton, ibid., speaks Of the Objects Of legitimacy as authori-
ties and regimes. Lipset has also spoken broadly of participants in the
political process as Objects Of legitimacy. Specifically, Lipset
has viewed Birchers and other extremists as denying the legitimacy of
those in the democratic game who differ with them; paper on extremism
delivered at the State University of New York at Binghamton, October
6, 1966.

5For measures of faith in people see Gabriel Almond and Sidney
Verba, The Civic Culture (Boston: Little, Brown and Co., 1965), pp. 212—
214. For the relation between faith in people and political coopera—
tion see pp. 227-231.

 

46For a distinction between image and attitude see Herbert C.

Kelman (ed.), International Behavior (published for the Society for the
Psychological Study Of SOcialfiISSEes, New York: Holt, Rinehart, and
Winston, 1965), pp. 25-26. For images as dependent variables see

"Part One" of Kelman.

47In addition to affecting the voters' behavior through effects

on their attitudes toward objects, leaders influence behavior through
their effects on the voters' situation.

60

. .each accusingly worded question or hostilely intended gen-
eral statement sets the stage for more bitter accusations and more
violent denunciation. The pattern of discourse already too pre—
valent in American political life-—that a point cannot be made
unless it is overstressed and reiterated in the strongest possible
terms——has been . . . made into the standard currency of American
legislative and political argument. Legislators who feel relative—
ly little animosity against their Opponents use this language be-
cause it has become a convention of their profession or because
to be heard in the clamor Of sensational words, they too must
Speak sensationally.

The deepe damage consi ts in what it i doing to the tone an
etiquette Of American political life. The tolerance and calm

which are necessary for rational discussion Of the extremely
complicated and difficult alternatives confronting us are greatly
diminished by low standards of political discourse . . . . This is

a time when it is most desirable and necessary that the Older crudi—
ties of mind and sentiment should be replaced by dispassionate
reflection and carefully measured statement. The tact and self
restraint which are essential to the political life Of a demo—

cracy are stunted in this atmosphere.

By their disrespect for the prOprieties, without which an ef-
ficient democratic government cannot function, the reputation of
the political profession is further besmirched and the quality of
its performance is reduced.

The nature of the tasks facing politicians today however is so
taxing to every moral and intellectual resource that the question-
able luxury of political savagery can no longer be afforded.

In a situation which seems very much like a vicious circle at
least part of the remedy must be sought in greater tactfulness

and restraint. It is not too much to hope that if legislators,
politicians, and citizens can be made more aware of the delicate
balance on which the free society rests, some Of them will change
their behavior and Others will no longer support or tolerate
practices which are harmful to the democratic system.

The Ten Indicators Used ip the Test of the ScalelHypothesis

The first indieatpr in the following scale is a dichotomy that

 

separates McCarthy and . . . his two great supporters in the Senate,

 

Viffiwww‘ 1“ fifi—v—V V‘V fifir ‘fi‘ V W

4822. at” pp. 583-584.

61

William Jenner, of Indiana, and Herman Welker, Of Idaho. . .49 from
the other senators (see Column 1 of Figure 4). These three were rated
as having given more intense, direct, and general support for Mc-
Carthyism then the other senators.

The second indicatgr is a dichotomy that distinguishes thirteen
senators who were indiscriminately hostile toward Secretary of State
Acheson (1949—52) from senators who were not indiscriminately hostile.
The number and proportion of "indiscriminate" references by those
who were hostile to Acheson are reported in the last two columns Of
Table ZIO The thirteen hostile senators who had more than 10.3
percent Of their unfavorable references in the personal qualities
and symbol-collectivity categories were labeled "Indiscriminates" by

Rosenau. The hostile senators with less than 10.3 per cent in this

category are labeled "Discriminates."

 

49Rovere, Op. cit., pp. 56-57. Although Rovere's classifica—
tion of Jenner and Welker as McCarthy's "two great supporters" is
accepted here, there were some "near greats" in the Senate. Bridges,
for example, supported McCarthy and McCarthyism in many ways in-
cluding motions on Senate floor. The most extreme roll call stimulus
in the following scale was shpaed by Bridges.

50Table 2 is based on Table VII of Rosenau's "The Senate and
Dean Acheson: A Case Study in Legislative Attitudes," unpublished
Ph.D. thesis, Princeton University (Princeton, 1957), (This is the
only reference that will be made to Rosenau's thesis.) All other
citations are to Rosenau's "Senate Attitudes Toward a Secretary of
State," pp, pip,

 

 

 

 

 

I I HO 0.00 HO 0.00 OH I I O .3: O HHOOOOO O I
H.H H OO 0.00 HN H.HO OH O.N H ON .HO O amppaouz O z
O.N H OO 0.00 OO H.OO OO O.H H O .Oz O OOHOHOO H H
N.N H OO 0.00 OO 0.00 OO H.OH OH O .O O panes O O
H.O H OO O.NO OO H.HO NO I I OH .32 O OOOOONOO 2 O
N.O O OOH H.HO OO 0.00 HH H.H OH O .O O OamHaocO H O
H.O O HNH N.OO OO 0.00 OH 0.0 H O .O O OHOO H
N.O O OO O.HN OH 0.00 HO H.O O NH .OO O HOOO3OHO O
O.NH O HO O.HN OH 0.00 HO O.HH O NH .3: O NOOOOH O
O.HH H OO 0.0N HH O.OO OO H.H H O .3: O seen: O I
O.ON O ON 0.0N H O.OH HN I I O .32 O HOOOHNO H z
0.0N O OO O.NN OH O.HH OO I I O .32 O HHOOOOOO O H
H.ON HN OOH O.HN NO 0.0H HHH I I O .32 O BOO z
O.OH OO OON O.ON OO H.OH OON O.O H O .3: O Hugues H
0.00 ON OH 0.0H OH 0.00 OO I I OH .mz O OOOOHHO O
0.0H OH NO 0.0H OH H.OO OO I I OH .32 O HOOH O
H.NN O ON 0.0H O 0.00 NN I I N .32 O OOOONHO O
O.NH OO OOO N.O NO 0.00 OHO I I O .HO O maonz H
0.00 ONH ONO 0.0 OH 0.00 OOO I I O .32 O Oeupmouz O
O.ON HO OHN O.O O N.OO NON I I O .3: O HOOOOO z
H.OO H OH I I O.OOH OH I I N .OO O HOOHOO H
H .oz .02 N .02 H .oz N .oz
G
.I , II II OOOH o H
hao>auooaaoo AHOHOGOOom on O u
ounces w H Houmoom
IHonaam O OO pom xuwawumomv mo o m
mOHuHHmoo Hmoomuom HOOOH Hmuusoz manmuo>mwaa manmuo>mm mumow m m

 

mm omouooom

mmooouomom OHano>mmop
mo coauuoamum pom Honsoz

mm ooouooom

moooouowom mo nowuuomoum pom Honaoz

 

OOOHHHOOHZ OHHumom mnu mo csooxmmum w N OHLOH

63

The third through the ninth indicators (Columns 3-9 of Figure
5) are McCarthy censure (Senate Resolution 301) roll call votes

taken in December, 1954.

Indicator §Q_ Number51 Description

 

3 6 (p. 472 McCarthy Censure. BRIDGES (R.N.H.)
substitute for Committee amendment,
resolving that McCarthy violated
no rule or precedent in failing to
appear before the Privileges and
Elections Subcommittee. Rejected
(12/1/54), 20—68

4 l (p. 473) McCarthy Censure. Committee amend-
ment rewording first count Of re-
solution condemning McCarthy for
failure to cooperate with the
Privileges and Elections subcom-
mittee investigating his finances
in 1952 and for abusing the Sub-
committee. Agreed to (12/1154),
67-20

5 4 (p. 472) McCarthy Censure. DIRKSEN (R 111.)
substitute.&or'the above committee
amendment) resolving that McCarthy's
conduct did not warrant formal

censure or condemnation. Rejected
12/1/54) 21-66.

6 5 (p. 473) McCarthy Censure. Adoption of
resolution as amended. Adopted
(12/1/54), 67—22

7 3 (p. 473) McCarthy Censure. BENNETT (R Utah)
amendment to committee amendment,
to delete second count, concerning
Zwicker, and to substitute a count
condemning McCarthy for his charges
against members of the Committee
which recommended censure and for
calling the Senate censure session a
"lynch bee." Agreed to (12/2/54)
64-23.

 

5100ngressional,Quarterly Almanac, 1954. Washington, D. C.: Con—
gressional Quarterly News Features, 1954, pp. 472—473.

 

64

8 4 (p. 473) McCarthy Censure. Committee
amendment as amended to eliminate
reference to Zwicker and sub-
stitute condemnation for abuse
of the censure Committee.

Agreed to (12/2/54), 62—24.

9 2 (p. 473) McCarthy Censure. Committee
amendment rewording second count
of resolution, condemning Mc-
Carthy for abusing and denounc-
ing Gen. Ralph W. Zwicker for his
testimony before McCarthy on
security at Ft. MOnmouth, N.J.
BRIDGES (R N.H.) motion to table
Committee amendment. Rejected
(12/2/54), 33-55. (Second count
of censure resolution thus was
kept alive.)

The tenth indicator (column 10) distinguishes those who endorsed

 

Margaret Chase Smith's 1950 statement of principles from those who
did not. The signers of this anti-McCarthy declaration, which became
known as "A Declaration of Conscience," were Charles W. Tobey (R N.H.),
George D. Aiken (R Vt.), Wayne MOrse (R Ore.), Irving Ives ( R N.Y.),
Edward J. Thye (R Minn.), and Robert C. Hendrickson (R N.J.), and
(the author) Margaret Chase Smith (R Maine).52

The statement said in part that

Certain elements Of the Republican Party have [promoted]
confusion in the hOpes of riding the Republican Party to

victory through the selfish, political exploitation of fear,
bigotry, ignorance and intolerance.

It is high time that we stopped thinking politically as
Republicans and Democrats about elections and started think-
ing patriotically as Americans about national security based
on individual freedom.53

 

52Congressional Quarterly Almanac, 1950 (Washington, D.C.: Congres-
sional Quarterly News Features, 1950), p. 452. Also see footnote
"a" Of Figure (below).

53Ibid., p. 452

65

TO the extent that the ten indicators discussed here form a
cumulative pattern the resulting measure demonstrates that longi—
tudinal analysis can promote the scientific measurement of significant

behavior.

The Test Of the Scale Hypothesis

Figure 5 orders the items according to their extremeness
(magnitude, or "difficulty"). The resulting scalogram has a C0-
efficient Of Reproducibility of .99. Although further refinement
is possible (e.g., through content analysis Specifically designed
to break ties), Figure 5 does reveal that those rated as most
supportive of McCarthyism by Rovere are included among those who
were indiscriminately hostile toward Acheson, the latter are in-
cluded among those who were most supportive of McCarthy on the
censure vote and these are included among the nonsupporters of
Margaret Chase Smith's anti-McCarthy Declaration Of Conscience.

(See Figure 5.)

Implications

 

It was noted above that Often the data are not available in a
single session of Congress for constructing a scientific measure. It
was assumed, however, that finding sufficient data for a measure that
offers differentiation, reliability, and validity, becomes more likely
if we (1) take off our cross—sectional blinders, and (2) simultane—
ously Opt for the complementary use of different types Of data,
rather than restrict ourselves to but one type of indicant.

The scale hypothesis (based on qualitative content analysis)

that the ten indicators drawn from a six—year period actually tap

m whomfim

 

 

 

z + N N I I I I I I Ouoxmn .m II Ommo munooommo VHOE

mucumomm
+ + + + I I I I I I mumswaoo II msmwaafiz mucooommo OHHE wH
+ + + + + I I + I I opOHOHoo II eHOHHOzOHOHHOOOOO OHHe HmeH m:
+ + + + + + + I I I OOOOH II Omemuosn
+ + + + + + + + I I muoxma .2 II mosey
+ + + + + + + + I I .aemO II aHHHmz
+ + + + + + + + I I OHOHOOHHOO II OOOHSOOO
+ O + N + N N + I I oowouo II doouoo
+ + + + + + + + I I OOOOOO II HmOOmonum
+ + + + + + + + + + OSOH II Honooaooxowm mHmuuoemom hnuumuoz
z + + + + + + + I I OOOHOHOO II HOHHOO ppmaHaHpomHO one:
+ + + + + + + + + I OHOOOH .2 II HOOOOH
+ + + + + + + + + I OHOOOH .O II neon:
+ + + + + + + + + I OOOHOOH II HHOOOOOO
+ + N N + + + + + I OHHO II HOOUHHO
z + + + + + + + + I OHOOHHHH II OOOOHNO OHOHHOOOOO
+ + + + + + + + + I OOO>Oz II OOOHOO Oeuumouz
+ + + + + + + + + I .O . 2 II OOOOHHO meOHaHNUOHOOH
+ + + + + + + + + + OOOHOOH II Hmeemm
z + + + + + + + + + OHOOH II HmOHm:
+ N N N N N N N + + OHOOSOHO II 398%: OBHOfiHOOuz
OH O O H O O O O N H Iwmmw. gum 4&3

mamHH

EOOOSOHmoOz no

mnoumcom omowanommm mo Hmvno xomm

66

nonnauooo I m ouomwm

.mmouwooo oumw onu wowuoo unmoooaoooH on mp3 OOHOZQ

 

.ommH .onoomm Hmooammomwdoo moms .m mom Nowmm mafia:
mo Sufism .mnz :wowauhno>o saws unnamoumm oouumosoaona: no unoaomuoooo
m.nuflsm .Hz mo ouOOOH>O mom .OOOHS o ooowammm OH Oomuom 3oz mo Sufism

.mnoowam msu ONO no woman owanoo m HOOHOOHO mm xoou on omomommm

 

.HHOO do Hoswam Ho monsooom no:
can :.uoomoum: .HHOQ.HOMOOOm .ucomn< N

 

.m .0 .OEHH wasp um mmonwooo mo Honaoa m uoz u z

I I I I I I I I .ooom II «won

I I I I I I I I moan: II Sufism

I I I I I I I I muommocfiz II oan

I I I I I I I I monumh 3mz II oomxowuooom

I I I I I I I I xHow 3mz II mm>H

I I I I I I I I cowouo II omnoz

I I I I I I I I Oomnom 3oz II Sufism

I I I I I I I I I Hoosuo> II ooxfl<
+ I I I I I I I I Hoosuo> II mumndmam
z I I I I I I I I can: II uumdawm
z I I I I I I I I mmmomx II domaumo
+ I I I I I I I I emu: II moOxumz
+ I I I I I I I I amwwnoflz II downwumh
+ I I I I I I I I .mmmz II Hanumoouamm
+ I I I I I I I I onmemama II Oomph
OH w H O m O m N H oumum Houmoom

 

mEOuH

muoooommo Oaumm

muoooommo Oumq

mmkH mamum

67

68

"McCarthyism" ("right-wing indiscriminate hostility directed at
'leftist' management of national security issues"), is well

supported by the empirical analysis. Not only is the reliability
impressive but the fact that the ten indicators were generated in

four different ways.clearly demonstrates that the rank order is not

an artifact peculiar to one type of data. Also the fact that the roll
call items fit so well among three other types of data argues against
the assumption that roll call data are necessarily artifactual or
invalid.

When the interpretation Of roll calls or other data is in
dispute, a longitudinal search for a criterion Of validity can often
resolve the dispute. For example, when the editors Of the H22.
Republic and William S. White offer conflicting interpretations of
the McCarthy censure votes, we need not conclude that the process
of interpreting such phenomena is so complex that it is necessarily
arbitrary. The fact that three other types of data support the

New Republic rather than the Citadel interpretion is intersubjectively

 

transmissible evidence; not subjective opinion.
The complementary use of data (made possible by longitudinal
analysis) also fosters and pinpoints reexamination Of measures and

the assumptions underlying the use of data and measuring techniques.54

 

54According to Rovere, Bridges was not among the three most extreme
senators, but Bridges was judged third most extreme by Rosenau's method
of estimation (percentage of unfavorable references in the personal quali-
ties and symbol—collectivity categories). This directs one's attention
to a comparison of Bridges' remarks with those Of Jenner to see if
Rosenau's raw data is inconsistent with Rovere's rating, or if Rosenau's
failure to distinguish between extremely indiscriminate and mildly indis-
criminate references accounts for his relatively low rating of Jenner.

69

For example, conflict between Rosenau and Rovere on the ordering of
Jenner and Bridges is found to be a consequence of Rosenau's decisions

to (1) estimate indiscriminate hostility from the_percentage rather

 

"personal

than the number of unfavorable:referencesrecorded in the
qualities and symbol-collectivity categories," and (2) treat all
references in these categories as equal. The consequences Of the
latter decision, to give no more weight to Jenner's extremely indis—
crinimate references than to Bridges relatively mild ones, are made
explicit by the systematic comparison of the different types Of
data. When one uses Rosenau's new data on Jenner and Bridges and gives
more weight to relatively extreme references than to mild ones, the
results are quite consistent with Rovere's rating.55

The above measure also demonstrates that longitudinal analysis
can increase the differentiation among respondents. If the 1949—
1952 data were not linked to the 1954 roll-call data, we could not
differentiate among the twenty who were most supportive of McCarthy.
The importance of increased differentiation to the development of
political theory has not, however, been sufficiently recognized;
perhaps because the vital role of differentiation in the interplay
Of theory and research has not been made explicit. If one were not
aware Of the range of behavior revealed by the above scale, it might

seem quite reasonable to diehotomize the senators on the roll call

on adoption of the Senate censure resolution (indicant 6);

 

5 O

Rosenau's data on Jenner and Bridges are also conSistent with
Rovere's rating if some weight is assigned to the number Of indis-
criminate references as well as to the percentage.

70

labeling those who voted "Nay" McCarthy supporters, and those who
voted "Yea" McCarthy Opponents. Such a dichotomy has the appeal Of
both simplicity and convention. Unfortunately this common practice
treats men who differ importantly as though they were the same
type. For example, such a dichotomization would lump together those
who had different positions on indicants 1, 2, 3, 4, and 5 as "McCarthy
supporters." And it would categorize as one type ("McCarthy
Opponents") senators who greatly differed on items.7, 8, 9, and 10.
Examples such as this must have inspired Hempel to write that
. . . comparative or quantitative concepts will Often prove so
considerably superior for the purposes of scientific descrip-
tion and systematization that they will seem to reflect the
very nature of the subject matter under study, whereas the

use of classificatory [either . . . or] categories will seem
an artificial imposition. (Emphasis added)56

 

 

 

When behavioralists use dichotomies to represent behavior that
can be shown to have a broad range, it gives some substance to the
criticisms that behavioralists do not measure important though
subtle differences—~do not respect the uniqueness or complexity of
political behavior. On the other hand, the above scale (which is Open
to even greater differentiation) is part of the evidence that suggests
that the more extreme anti-behavioralists have prematurely concluded
that science cannot sufficiently differentiate among the politically
important and that subjective intuition is preferable to "futile

counting."

56Hempel, Aspects of Scientific Explanation (New York: Free
Press, 1965), p. 56.

V URBAN AFFAIRS DECISION MAKING IN THE SENATE:
A PREDICTIVE STUDY

Congress is making an increasing number of decisions directly
affecting urban governments and is increasingly worthy of attention
from those concerned with all arenas of decision affecting urban
affairs. Congress, however, because Of its size and complexity can
not be as fruitfully studied by a traditional approach as the
smaller arenas for urban affairs decisions. A reSearcher may gain
considerable insight from intimate acquaintance with all members of
a city council. He may be able to retain a large percentage of
their reaponses in his mind and imaginatively order these into
patterns of behavior. This is not as true, however, when our re-
searcher turns his attention tO urban affairs decisions flowing
from Congress, and this may explain the relative poverty of generali-
zations in this area. Statements such as "Congressmen tend to think
small about city housing,"1 may suggest something of the value
orientation Of the author but do not provide descriptions that may
serve as a basis for explanation, prediction, and control.

One requisite of the scientific study of Congressional decision-
making on urban affairs issues is the clarification of the dimensions
of such behavior. The a priori lumping together of all measures such

as urban renewal, housing, community facilities, airports, District

 

1Daniel Seligman, The Exploding Metrppolis, The Editors Of

Fortune (Garden City: Doubleday and Co., Inc., 1958), p. 102.

71

72

of Columbia affairs, and mass transit as urban affairs policy may
allow only the roughest description of policy behavior. A more
precise identification of the correlated but multiple dimensions
will promote explanation, prediction, and control.

A related requisite of the scientific study of policy behavior
is the identification of enduring or recurring dimensions. This
search for regularities that hold up over time requires that time-
bound case studies of single sessions be supplemented with longi-
tudinal analyses which can distinguish the ephemeral from that which
is more enduring and thus of scientific interest.

But what are the assumptions about the nature of decision-
making in the area of urban affairs that lead us to expect the
recurring behavior necessary to scientific study? If it is typical
that a problem appears, a decision is made, and the problem is
solved; there may be no recurring or enduring dimension and limited
prospects for scientific study. If, however, urban problems are
complex and our economic and intellectual resources are limited, the
problems are likely to require a lengthy series Of related attacks
-—constituting an enduring dimension amenable to scientific study
through longitudinal analysis. This Chapter assumes that corporate
decision-making by Congress on urban affairs is overtly serial and
incremental. It also suggests a method for clarifying this
assumption.

If most of the decisions on urban affairs can be placed on a
few recurring dimensions, the Kennedy administration should have

been able to avoid its embarrassingly inaccurate prediction of

73

Senate behavior on the Department Of Urban Affairs proposal.2
Furthermore, if this decision which was described by journalists as
a unique result of many factors is found to be part of an enduring
pattern of decisions, so should most of the more routine urban

decisions.
The Department of Urban Affairs Decision

In February, 1962, proponents of the Department of Urban
Affairs received a setback in the united States. The unexpected
42-58 defeat not only signified the weakness of urban forces or
strategy but also proved to be one Of the key events giving Presi-
dent Kennedy his image as weak in managing Congress.3 An apprOpriate
test of the hypothesis that this decision was not idiosyncratic but

a predictable part of a larger pattern Of behavior on urban measures

 

2"Administration leaders ... wanted it voted on first in the

Senate where they thought it would be approved." 1962 Congressional
[Quarterly Almanac, p. 380.

 

3That the Kennedy image did suffer from his Congressional defeats

is suggested by the following comparison:

Public Approval for Johnson and Kennedy in Key Areas Of Action

Per cent approval for: Net

Kennedy Johnson K—J

Keeping government clean 55 51 +4
Handling Khrushchev 66 74 -8
Handling race problems 49 60 -11
Handling Vietnam 56 72 —16
Keeping economy healthy 59 77 -18
Getting Congress to act 44 77 -33

(Kennedy figures based on early November, 1963, results,
Johnson figures based on August, 1964, results?
Newsweek, August 31, 1964, p. 27.

74

is found by relating the 1962 decision to the previous relevant set
Of decisions--those on the 1961 Housing scale. If there is con-
siderable regularity (order, structure, patterning) in urban affairs
decision-making, the 1962 prOposal Of a department centering around
the FHA should either scale with the 1961 housing items or be
highly correlated with the 1961 housing scale. Furthermore, if the
hypothesized regularity is confirmed, we will have a basis for
understanding the options Open to urban stategists in each period.
The following scale (Figure 6) reveals not only that the 1961

housing decision (items 1, 3-9) form a unidimensional pattern but

also that the 1962 decision(item 2) fits rather well on the same dimen-

 

 

sion.

1 2 3 4 5 6 7 8 9
Clark Pa. + + + + + + + + +
Kefauver Tenn. + + + + + + + + +
Bartlett Alaska + + + + + + + + +
Carroll Colo. + + + + + + + + +
Douglas 111. + + + + + + + + +
Hart Mich. + + + + + + + + +
McNamara Mich. + + + + + + + + +
Humphrey Minn. + + + + + + + + +
Long NO. + + + + + + + + +
Smith Mass. + + + + + + + + +
Williams N. J. + + + + + + + + +
Morse Ore. + + + + + + + + +
Neuberger Ore. + + + + + + + + +
Pastore R. I. + + + + + + + + +
Pell R. I. + + + + + + + + +
Jackson Wash. + + + + + + + + +
Magnuson Wash. + + + + + + + + +
Byrd w. Va. + + + + + + + + +
Randolph w. Va. + * + + + + + + +
Case N. J. + + + + + + + + +
Javits N. Y. + + + + + + + + +
Yarborough Texas + + + + + + + + +
Proxmire Wisc. + + + + + + + + +

Figure 6

 

75

 

Symington
McCarthy
Engle
Mansfield
Metcalf
Mess
Hickey
McGee
Burdick
Long
Young
Muskie
Gruening
Church
Hartke
ANDERSON
CHAVEZ
DODD
BIBLE
CANNON
SPARKMAN
HAYDEN
HILL
JOHNSTON
BOGGS
LONG
FONG
Keating

FULBRIGHT

Gore

Kerr
Cooper
Ellender
Kuchel
Wiley
Jordan
Ervin
Talmadge
MOnroney
Scott
Merton
McClellan
Allott
Bush
Goldwater
Williams

Mo.
Minn.
Calif.
MOnt.
Mont.
Utah
Wyo.
Wyo.
N. D.
Hawaii
Ohio
Maine
Alaska
Idaho
Indiana
N. Mex.
N. Mex.
Conn.
Nev.
Nev.
Ala.
Ariz.
Ala.
S. C.
Del.
La.
Hawaii
N. Y.
Ark.
Tenn.
Okla.
Ky.
La.
Calif.
Wisc.
N. C.
N. C.
Ga.
Okla.
Pa.
Ky.
Ark.
Colo.
Conn.
Ariz.
Del.

1

+++—+++++
++-++++++
-++++++++
—++++++++
—++++++++
-++++++++
-++++++++
-++++++++
-++++++++
-++++++++
—++++++++
—++++++++
-++++++++
-++-+-+++
-+-++++++
--+++++++
--+++++++
--+++++++
—-+++++++
--++—++++
-—-++++++
---++++++
---++++++
—--++++++
---+-++++
+---+++++
+---+++++
+++ ----- +
--—+--+++
++ ------ +
-+--+--——+
+-+ ------
--——+---+
_____ + _ _ _
--+ ----- +
-------- +
-------- +
-------- +
_+ _______
_+ _______
+ ........

Best prospects for
winning pep_votes

in the 87th and 88th
Congresses (if issue
not posed as race

issue)

 

Figure 6 - Continued

76

 

 

 

 

 

Holland Fla. _________
Smathers Fla. _________
Russell Ga. _________
Dworshak Idaho _________
Dirksen Ill. _________
Capehart Indiana —————————
Hickenlooper Ia. —————————
Miller Ia. _________
Carlson Kansas _________
Schoeppel Kansas _________
Smith Me. _________
Beall Md. _________
Butler Md. _________
Saltonstall Mass. ? ————————
Eastland Miss. _________
Stennis Miss. _________
Curtis Nebras. —————————
Hruska Nebras. —————————
Bridges N. H. _________
Cotton N. H. —————————
Young N. D. _________
Lausche Ohio _________
Thurmond S. C. _________
Case S. D. _________
Mundt S. D. _________
Tower Texas ? — — ? ? ? ? ? -
Bennett Utah _________
Aiken Vt. _________
Byrd Va. _________
Robertson Va. _________
Prouty Vt. _________
T 3 4 4 4 5 5 5 5 5
O l 2 4 7 0 0 l 1 9
t Nonscale Responses
3 CR = 1 -
1 Total Responses
5 CR = 1 - 24
893
CR = .97
+ = Pro-urban vote

Anti—urban vote

* Senate rules required that Senator Randolph, the only Senator
with perfect support for the dimension in 1961 not to vote for the
Department of Urban Affairs, vote nay after acceding to Majority Leader
Mansfield's request that he move for discharge of

Figure 6 — Continued

77

the resolution of disapproval (1962 CO Almanac, p. 383).

Identification Of Scale Items:

1 2 3 4 5 6 7 8 9
1961 CO Roll Call 68 185 62 71 72 7O 74 85
1962 CO Roll Call 14

It should be noted that support for urban measures drOpped
precipitously after the middle of the distribution. Thus it was
necessary (in the 87th and 88th Congresses) that the proponents
garner five to eight of the votes of the moderate Southerners or
their sympathizers from Nevada, New Mexico, and border states. Con-
siderable support (five to eight votes) from that range of the dis—
tribution that extended from Anderson (N. Mex.) to Fulbright would
not have been an unrealistic hOpe (barring the premature announcement
Of a Negro to head the department). To have expected to Offset
losses among these crucial marginal senators by searching for the
support of Northerners further down in the distribution appears
unrealistic. Advocates of a cabinet level Department Of Housing
and Community Affairs in the early sixties did not have a promising
alternative to recruiting support from these identifiable marginal
Senators.

While the above scale provides a very useful basis for identi-
fying the best prospects in 1962—64 for votes 43-50 for a Depart-
ment of Housing and Urban Development, there are alternative per-
spectives for exPlanation or prediction of such a decision. The
relationships between constituency variables, nominal party

affiliation, the Senators' personal characteristics, and policy

78

position on the 1962 measure are as follows:

Constituency Variables and Policy Position

As one should expect there is a positive correlation between
"Z urban" and position on the 1962 measure. The more interesting
finding, however, is that neither "Z urban" nor "Z rural farm" is

as strongly related to policy position on this issue as are eight

other variables.

Table 3

CORRELATION OF THE 1962 VOTE WITH
CONSTITUENCY VARIABLES*

 

 

Z urban . . . . . . . . . . . . . . . . . . .2887
Z rural farm . . . . . . . . . . . . . . . . -.2226
Z nonwhite . . . . . . . . . . . . . . . . . -.2525
Z foreign born . . . . . . . . . . . . . . . .3265
Z native of foreign or mixed parentage . . . .3480
Median income . . . . . . . . . . . . . . . . .3933
Z Of families with incomes under $3,000 . . -.3909
Z Of families with incomes $10,000

and over . . . . . . . . . . . . . . . .2958
Z completed less than 5 years Of school . . -.3440
Z unemployed . . . . . . . . . . . . . . . . .3253
Z white collar . . . . . . . . . . . . . . . .3476
Z using public transportation to work . . . .2848

 

* Calculated from Table 2, 1962 County City Data Book

While none Of the constituency variables provides a basis for
prediction or political strategy that compares with the above scale,
the relationships suggest hypotheses for further testing. (For

example, two Of the correlations suggest that persons who have

79

received their political socialization in foreign countries or from
parents rasied in foreign countries are more accepting of central

government efforts in urban affairs.

Party Affiliation and Policy Position

 

Nominal party affiliation correlated .45 with position on the
Department Of Urban Affairs. The 38—26 split among the Democrats
would have made any attempt to predict policy position from nominal
party affiliation somewhat disappointing.

If, however, we define party behaviorally by using Louis

 

McQuitty's "HiClass" to dichotomize the Senate on the basis Of its
1961 roll call behavior, we find that 93Z (38/41) of one behavioral
party supported the Department of Urban Affairs while 93Z (53/57)

of the other behaviorally defined party opposed the 1962 measure.

 

The Senators' Characteristics and Polipy Position

 

Table 4

SUPPORT FOR THE DEPARTMENT BY ORDERED
SUBCLASSES OF SENATORS

 

 

Pro Con
Age 62-71 3 26 (most anti)
Agriculture 2 l4
Business/Banking ll 18
(Senators 42 58)
Teachers 8 6

Age 37—50 17 5 (most pro)

 

80

While much Of the association between age and policy position
is incidental to or stems from the association of age with more
fundamental variables, not all of it is incidental to differences in
constituency. When we examine the fourteen states in which "same-
state senators" split on this measure we find that in these fourteen
comparisons where legal constituency is held constant the probability
is nine to five that the younger of the "same—state senators" will
support the Department while the older will Oppose. Thus while age
is not a major factor it may be necessary to a comprehensive explana-
tion.4

None of the tOp 22 Republicans on seniority and only one of the
top 16 Democrats favored the bill.5 This was in small part, a
consequence of how the administration defined the issue and not simply
a consequence of implacable Opposition of all of these men toward the
basic urban dimension.

In the case of the 1962 Department of Urban Affairs vote it
appears that for purposes of prediction, assessment of the political
prospects Of urban interests, and develoPment of political strategy,

the ordering obtained by longitudinal scale analysis is more

 

“Subsequent examination of the 1965 data revealed that in 12
of 16 states that split on HUD the younger Senators were pro-HUD.

SThis reinforces doubts about the capacity of the seniority
system to anticipate and cape with urban problems.

81

instructive than the alternative perspectives. But is this a

unique case--a fluke? It has been suggested above that-~because
journalists considered this vote as more complex or more nearly
unique than most votes on urban affairs-—evidence that it fits on

a dimension may be considered evidence that most of the more routine
urban affairs decisions are predictable parts of highly patterned
behavior.

As an additional check on the hypothesis that urban affairs
decisions are predictably patterned, two predictions were attempted
from the 1963 Urban Mass Transit scale (Figure 7 , items 1, 2, 4, 5,
7—12). The first was a prediction of position on the l9p§_Urban
Mass Transit Act (Figure 7, item 3).

The first Of these predictions (that the 1393 Urban Mass
Transit Act vote would fit on the same dimension with the l9p§_Mass
Transit items) was made with the most confidence because of the
apparent homogeneity of content. Whether using correlation or error
criteria the items from these different sessions were found to con-
stitute one recurring or enduring dimension.

The second of these predictions (the higher the scale position
Of a group of senators on Urban Mass Transit the higher the pro-
bability that they will support reapportionment) assumes that most
senators, especially Northerners, will respond to both Urban Mass
Transit and the reapportionment rider largely in terms Of a more
general attitude of support for or hostility to the claims of urban
areas. Some attenuation of the relationship between the two issues

may result from Southerners and perhaps some of their sympathizers

82

Figure 7

l 2 3 4 5 6 7 8 9 10 11 12
++++++++++++
++++++++++++
++++++++++++
+++++++++~+++
++D++D++++++
-+++++++++-++

Conn.
Mass.
Conn.
Tenn.
Minn.

Pa.

 

 

Clark
Ribicoff
Kennedy
Dodd
Kefauver
McCarthy
Case

-++++++++ + + +

J.
Haii.

--++++++++++
-—++++++++++
—-—++++++++++
--++++++++++
--++++++++++
--++++++++++
--++++++++++
--++++++++++
--++++++++++
-—++++++++++
-—++++++++++
--++++++++++

Ill.
Md.
Md.
Mich.
Mich.
Minn.
M0.
M0.
Mont.

Inouye
Douglas
Brewster
Beall
Hart
McNamara
Humphrey
Long
Symington
Metcalf
Williams
Javits

+++
+++

--+++++++
--+++++++

YD

Keating

--++++++++++
—-++++++++++
--++++++++++

Ore.
R.

w.

McGovern

Burdick

Morse

Pastore

Va. --++++++++++
Wisc.

Randolph
Nelson

--+++++++ + + +
——D++++++ + + +

Calif.
Tenn.

R.

Engle

Gore

?
+
+

-—+++++++ + +

Pell

+
+

—-++++-++ + + +

—-+++++++
--+++++++

Pa.

Scott

Ore.

Neuberger
Jackson

?++++++ + + +
--+?+++++ + + +
——+?+++++ + + +
--+++-+++ + + +
--—++++++ + + +
——-++++++ + + +
---++++++ + + +
—-—++++++ + + +
-——++++++ + + +

+
+

Wash.
Wash.
Ind.
N.

Magnuson

Hartke

Mex.

Anderson
Bible

Nev.
Alaska
Nev.

Gruening
Cannon

++
++

++++ +
.++++ +
---++++-+ + + +

—-2
...—'7

Mont.
Okla.
Ohio
Alaska

Mansfield
Monroney
Young

Bartlett
Johnson

C.

 

Figure

83

7 —~ Continued

 

 

Edmondson
Bayh
Yarborough
Sparkman
Hayden
Muskie
Church
Kuchel
McGee
Byrd
Fulbright
Long
Russell
Talmadge
Hill
Aiken
Proxmire
Smith
Cooper
Boggs
Fong
Ellender
McIntyre
Jordan
Saltonstall
Prouty
Ervin
Miller
Young
McClellan
Lausche
Eastland
Stennis
Byrd
Robertson
Williams
Morton
Cotton
Smathers
Holland
Jordan
Dirksen
HickenlOOper
Carlson
Pearson
Curtis

Okla.
Ind.
Tex.
Ala.
Ariz.
Maine
Ida.
Calif.
Wyo.
W. Va.
Ark.
La.
Ga.
Ga.
Ala.
Vt.
Wisc.
Maine
Ky.
Del.
Haii.
La.
N. H.
N. C.
Mass.
Vt.
N. C.
Ia.
N. D.
Ark.
Ohio
Miss.
Miss.
Va.
Va.
Del.
Ky.
N. H.
Fla.
Fla.
Idaho
Ill.
Iowa
Kansas
Kansas

Nebras.

--++-++++
--++-?+++
--??+++++
--?+++——+
---+++--+
--++-—++-
-—+-+++++
——+-+++++
-—2 --+2+
----++++—
--+-+--?+
--+-++—-+
---—++-—+
----++-—+
--?-++-—-
----- +++—
—-+-—-++-
...... + + _
...... + + _
...... + + _
...... + + _
—+ ------ +
---+—-—-+
_______ + _
_______ + _
_____ 7..--
_____ + - _ -

+++++++++++++++++++

I + + I I

+u++++u++++++++++++~+++~++++

u++++n++n++++++++++++++++++++~++

84

Figure 7 -- Continued

 

 

 

Hruska Nebras. ————————— _ _ _
Mechem N. Mex. _________ _ _ _
Thurmond S. C. _________ _ _ _
Mundt S. D. ————————— _ _ _
Tower Tex. ————————— _ _ _
Bennett Utah ————————— _ _ _
Simpson Wyo. _________ _ _ _
Goldwater Ariz. ————————— _ _ -
Allott Colo. _________ _ _ _
Dominick Colo. _________ _ _ -

+ = Pro-urban

- = Anti-urban

D = Dead at time of vote

CR for the eight 1963 Urban Mass Transit scale items 1, 2, 4, 5,

9—12 (not containing the explicit states rights issue clearly con-
tained in items 7 and 8) and the 1964 Mass Transit vote (6) = .97.

When items 7 and 8 (containing the explicit states right
component) are included in the analysis, the CR is lowered to .95.

1 2 3 4 5 6 7 8 9 10 11 12
1963 CQ Roll Call 17 15 18 21 11 10 16 13 9 19
1964 CO Roll Call 284 198

Column 3 (CO Roll Call 284) contains the reapportionment rider
referred to in the text.

85

reacting in terms of displeasure with the Court.
By dividing the senators into five groups on the basis of their
degree of support for Urban Mass Transit we find the following

divisions on the reapportionment issue:

 

 

Pro— 9 Anti-
Reappor ' Reappor
Group 1 Most Pro Urban Mass Transit ( 4) 4 O 0
Group 2 ( 2) 2 0 0
Group 3 (41) 29 5 7
Group 4 ( 9) 4 2 3
Group 5 MDst Anti (41) l O 40
97* 4O 7 50

 

*Engle, Kefauver, and Mbss not included in this analysis.

Thus there is little reason to believe that the relatedness of
the 1961 and 1962 items was simply a fluke. Relations among 1963
and 1964 items are quite clear. But perhaps the most convincing
evidence of a predictable patterning of urban affairs decision comes
from the test of a prediction of 1965 behavior. COpies of such a
prediction were submitted to Professors Spaeth and Adrian in July,
1965 (before the HUD vote). The basis of this identification of the
Senators most likely to provide a majority for HUD was a scale de-
rived from the 1965 housing items and the 1962 vote on the Depart-
ment of Urban Affairs (Figure 8 ). The 1965 housing items had been

chosen as the basis for updating the persPective on potential

86

support for HUD before the 1965 housing votes took place.6 The
1961-64 analyses had given considerable reason to expect that the
1965 housing items would clarify the Options Open to HUD prOponents
in the 89th Congress. Just how sound a basis the longitudinal view

(of housing and "Department of Urban Affairs," 1961-65) provided for

 

practical political analysis and for academic understanding of the
options open to urban strategists is evidenced by the fact that not
one of the fifty Senatorsgprodicted to favor HUD subsequently voted

against it.7

 

Some Additional Implications and Extensions

Urban affairs decision—making in Congress is quite open to
longitudinal analysis scientific study.

1. Urban affairs decision behavior is measurable at the level of
ordinal (not just nominal) measurement. The finer distinc-
tions made possible by ordinal measurement are not sought for
the sake of descriptive subtlety in itself but because greater
descriptive flexibility or differentiation increases the

opportunity to conceive, state, and test generalizations.

 

6And before the designation HUD had replaced the tentative label

"Department of Housing and Community Affairs." Appendix B of "An
Exploration of Urban Affairs Decision-Making in Congress (paper sub—
mitted to Charles Adrian in completion of Political Science 805).

7At the request of one professor probably Opponents were also
identified on the above prediction. The correlation between the pre-
dicted and the actual results is .9 (Yules Q).

Senator
Hart
Mbrse
Clark
Pastore
Ribicoff
Kennedy
Bartlett
Gruening
Church
Douglas
Hartke
Muskie
McNamara
McCarthy
Long
Mansfield
Metcalf
Williams
Case
Javits
Burdick
Young
Monroney
Neuberger
Pell
Yarborough
Moss
Jackson
Magnuson
Proxmire
McGee
Scott
Symington
Kennedy
McIntyre
Mbntoya
Harris
McGovern
Bass
Nelson
Bayh
Brewster
Tydings
Inouye
Mondale
Gore
Byrd
Randolph
Hayden
Long

HHH Pres. of Senate

State
Mich.
Ore.
Pa.
R. I.
Conn.
N. Y.
Alaska
Alaska
Idaho
Ill.
Ind.
Maine
Mich.
Minn.
Mo.
Mbnt.
Mont.
N. J.
N. J.
N. Y.
N. D.
Ohio
Okla.
Ore.
R. I.
Texas
Utah
Wash.
Wash.
Wisc.
Wyo.
Pa.
Mo.
Mass.
N. H.
N. M.
Okla.
S. D.
Tenn.
Wisc.
Ind.
Md.
Md.
Hawaii
Minn.
Tenn.
W. Va.
W. Va.
Ariz.
La.

87

Main Bases of Prediction

(7/65)
161

n++++++

(l962)(7/65)

l4

I++zzzzzzzzzzzz+++++++++++++++++++++++++++zz++++

+

159

++++++++++++++++++++++++++++++++

n++++++++++++u

21+ +A+ I

5
({ég )(

+++++++++++++++++++++++++++++++

|+++++++++++++I

z+++1

Pred.

7/65)
Pred.

z++++++++++++++++++++++++++++++++++++++++++++++++++

Actual

(8/65)
179

z++++++++++++++++++++++++++++++++++++++++++++++++++

 

=Not eligible to vote.

Figure

(l

O

88

Most of the hypotheses given below about change, the dis—
tribution of power, the relation between individual and
corporate decision, and political strategy could not be
based on classificatory concepts that merely indicate the
presence or absence of a prOperty but not the order or
degree.

The dimensions of urban affairs behavior can be clarified
—-divided into related but not identical variables-~thus
allowing more precise description, explanation, and pre-
diction.

Urban affairs decision behavior is sufficiently enduring
to be the subject of generalizations that hold over more
than one session of Congress and thus allows the

generality of inference over time associated with science.

Recognition of the kind of distribution (whether it approaches a

normal or a U-shape) existing for urban affairs dimensions is a

prerequisite for understanding: (1) the relation between individual

and corporate decision; (2) the potential for change in corporate

decision; (3) the relative power of marginal individuals near the

midpoint of the distribution; and (4) the relevance or superficality

of concepts of "average" or "typical" behavior of senators.

l.

The relation between individual and corporate decision varies
greatly with the kind of distribution. With a normal dis-
tribution there is a high relationship between the corporate

decision pattern and that of many individuals. With a U—shaped

89

distribution, however, a large percentage of the individual
members may be extreme yet the corporate decisions be moderate.
On the Housing and Department of Urban Affairs dimension each
of 79 senators cast all or all but one of his nine votes in
either a consistently pro—urban or consistently anti-urban
direction, but the decision pattern for the Senate as a cor-
porate body was — — — - + + + + +: quite different from that
of 79% of the senators. Mbre impressive evidence for the mean-
ingfulness of the distinction between individual and corporate
decision is the fact that not a single senator on this urban
dimension had the same decision pattern as did the Senate as

a corporate body.

The potential for change in the corporate position of a

 

decision-making body such as the Senate also depends on the
nature of the distribution of its members. With a normal dis-
tribution many individuals must change to bring about signifi—
cant change in the corporate position. With a U—shaped distri-
bution, however, such as is found on urban affairs, a change

in a few individuals near the midpoint of a distribution may
bring considerable change in corporate decisions.

The relative importance or "power" of the individuals near the
midpoint of a distribution to contribute to change in corporate
decision also depends on the shape of the distribution. While
with either type of distribution it is the pivotal individuals
who would be considered by the Shapley—Shubik school as the

most powerful, another consideration is the percentage who are

90

near the midpoint. On the 1961-62 Housing dimension the
potentially pivotal members were fewer and thus more "powerful"
than the potentially pivotal members of a normal distribution.
(An individual's power stemming from his pivotal position in a
U—shaped distribution is not a general personality trait or
personal possession capable of equal expression in all policy
areas, but rather is a consequence of a relation with others
and greater freedom to act that may be unique to the dimension
under consideration. It is a source or kind of power that is
highly specific.)

4. Statements such as "Congressmen . . . think small about city
housing,"8 suggesting that congressmen think alike about hous-
ing are less plausible if we note that on the relevant dimen-
sion the distribution of congressmen simply does not conform
to the above assumption. For example the 1961 housing distri-
bution is nearly the reverse-—a bimodal distribution with a
large percentage of the subjects at or near the extremes.
Given such a distribution descriptions of "average congressmen"
invite serious misunderstanding. They serve neither as a
descriptive base for science nor as a practical base for
political strategy.

Viewing urban affairs decision-making in Congress from the perspec—

tive of Braybrooke and Lindblom's classification of types of

 

8Seligman, op, cit.

91

decision—making (page 21 ), we can conclude that it does not involve

 

the large sudden change associated with revolutionary or utopian
decision-making, nor the high understanding associated with some
administrative or technical decision-making. It is less rational-
deductive problem solving, than a lengthy series of limited decisions
(or attacks or steps) of an incremental nature——that is decisions
whose consequences differ only incrementally or marginally from the
status quo.

If Braybrooke, Lindblom, and Simon are correct in suggesting
that experience enlightens us more about prOposals that differ only
marginally from those with which we have had experience than about
more grandiose proposals, we may conclude that although incremental
decision-makers do not have as much understanding as is required
for utopian or revolutionary decision—making, they have more under-
standing of the probably consequences of their policy moves than
the typical utopian or revolutionary actually has of his. The
relative modesty of the incremental decision-maker is based on a
realistic respect for the complexity of problems and the limits of
our resources including our intellectual resources.

(Incremental decision—making need not have a conservative
coloring. It would, for example, be entirely compatible with this
type of decision—making for Congress to repeatedly expand the
urban mass transit program if previous investments are judged
worthwhile. In fact the initial mass transit program was clearly

intended not as a solution but as an experiment which if promising

92

could lead to a series of more extensive attacks on urban con-

gestion.)

Advice to the (urban) Prince(s)

 

Some of the implications of the above that are relevant to the
practical political strategy are:

l. Clarify the dimensions of the behavior. See how the issues
are related in the minds of the decision-makers--what
general evaluative dimensions are employed.

2. Identify dimensions which recur or endure and thus allow
prediction.

3. Use ordinal measurement to obtain finer differentiation
among decision—makers than obtained by dichotomies such as
pro and con or rural and urban. Increased descriptive power
not only reduces the tendency toward stereotypy but increases
the potential for empirical generalizations.

4. Identify the type of distribution and the related con—
sequences (four of which are discussed above).

5. Identify the decision-makers near the midpoint of the dis-
tribution who are the best prospects for providing the
winning votes for a coalition.

6. Identify how far these marginal voters were willing to go
(on the previous relevant dimension) and using this and
perhaps some interview data estimate how far they may be
willing to go in the foreseeable future.

7. Careful identification of the marginal members not only

93

suggests which representatives to concentrate on and how far
they may be willing to go but what tactics of persuasion are
the most apprOpriate. It appears from an examination of the
1961—64 distribution on housing and the Department of Urban
Affairs that the tactics supported byginnumerable political

scientists who believe that the President as representative of

 

urban interests must get Congress to act by taking issues to
the people, raising the level of popular awareness, thus in—

creasing pOpular pressures upon congressmen, can be grossly

 

inappropriate. The senators who were the best prospects for

 

providing the winning votes on this dimension before 1965
came from states where the wishes of John F. Kennedy were more
likely to stimulate a negative than a positive response. The
bulk of these constituencies were certainly not inclined to
follow the presidential lead. To have made their senators'
actions more visible on issues seeming to involve more federal
power in local affairs, race, more taxes, expenditures, and
deficits would have invited negative pressures.

Realistic strategies must be based on realistic analyses.

The latter often require a longitudinal perspective.

VI. CONTINUITY AND CHANGE IN FOREIGN POLICY POSITION:
LONGITUDINAL ANALYSIS AS AN AID TO CAUSAL INFERENCE

Critics of the scientific approach to political behavior have
argued that an intractable problem in the scientific study of
politics is that of finding recurring patterns in the changing be—
havior of politicians, particularly of sephisticated politicians
faced with complex and changing conditions (for example, U. S.
senators). No doubt changing behavior can complicate the tasks of
description, explanation, and prediction. But do changes in the
behavior of the politically important defy scientific generaliza-
tion?

The degree of difficulty in identifying regularities under-
lying change depends on what kind of change is being analyzed and
what intellectual tools (for example, research design) are brought
to bear on the problem. For this reason it is useful to explicitly
distinguish several types of change and then to consider what type
of analysis is appropriate to each.

Among the kinds of change that may at first frustrate the ob-
server of legislative behavior are:

1) Changes in the direction of response (from on extreme

 

proposals to "+" on more moderate prepositions) that are

 

simply a function of the different valences (magnitudes)

of the items stimulating a cumulative pattern of response.

 

Such a cumulative pattern of roll call behavior may have
been generated by either:

a) a single univocal attitude (which is measured at

94

2)

3)

4)

The
analyses
may seem

however,

95

the ordinal level by the roll call scale).

b) the systematic interaction of multiple attitudinal
components (none of which is measured by the rank
order on the cumulative pattern of behavior).

Change over time in rank order on an issue because of:

 

a) change in relevant condition(s).

b) change in underlying attitude(s).

Changes in the valences of congressmen on a dimension of
policy behavior with little or no change in rank order
(for example, shift of the congressional distribution in
a pro—selective service direction after Pearl Harbor).

Change in the empirical relation between variables (for

 

example, clear change from one Congress to another in

the correlation between policy behavior on an issue and

a constituency variable because of a shift in rank order

on policy produced by a change in a relevant condition).
last three kinds of change obviously call for dynamic

(panel or before-and-after). But the first kind of change
to be open to cross-sectional analysis. A major difficulty,

is that the researcher often cannot tell from a cross-

sectional analysis how to interpret a cumulative pattern of voting.

No matter how high the coefficient of reproducibility, the cumula-

tive pattern may have been generated by either a single univocal g;

a systematic interaction among multiple attitudinal components. If

we perceive the problem in terms of an individual compensatory

composition model in which an excess in one attitudinal component

96

can compensate for a deficit in another, it is obvious that a
number of congressmen can have the same rank on voting behavior
without having the same rank on any of the attitudes that contri-
buted to that behavior. Thus the researcher using the cross-
sectional approach does not know whether to consider the rank order
on the cumulative pattern of behavior as the rank order on an under-
lying attitude (as the measure of an attitude), or merely as the
behavioral product of the systematic interaction of several under-
lying attitudes.

This chapter's example of longitudinal analysis, a simple
before-and—after design, focuses on "type 2a" change, but some
aspects of the problem treated here are also relevant to "type 1"
and "type 4" changes. The following methodological hunches are
explored:

1) Dynamic longitudinal analyses (panel or before-and—after
designs) aid in the identification of patterns of change
in behavior associated with changes in conditions.

2) Thus such longitudinal analyses provide a basis for em-
pirically disciplined inference about regularities under—
lying change in behavior.

3) Among the specific problems requiring inference that are
made more amenable to scientific analysis are:

a) the problem of whether single or multiple attitudes
underlie a cumulative pattern of voting behavior
(foreign policy position).

b) what the content of one or more of the underlying

97

attitudes is:
c) whether the condition hypothesized as relevant is
indeed an important condition affecting behavior.
d) the identification of the subset of politicians who
vary their behavior with the change in condition
(party control of the administration).

This chapter's substantive problem is the appraisal of alterna—
tive models of congressional delegation to the executive on foreign
policy roll call decisions. It is concerned with the identification
of an external condition and an attitude, or attitudes, that under-
lie congressional foreign policy making.

Miller and Stokes have argued that decision making in the
domain of foreign policy conforms to Burke's conception of represen—
tation in the sense that the Congressman looks elsewhere than to
his district (they report a correlation of —0.09 between the policy
position of representatives and those of districts).1 They add that
congressional foreign policy—making deviates from the Burkean model
in that the representative relies on the executive branch which is
deemed to have superior information with which to calculate the
public interest. Their concluding paragraph warrants close
attention.

It would be too pat to say that the domain of foreign
involvement conforms to the third model of representa-

tation, the conception promoted by Edmund Burke. Clearly

it does in the sense that the Congressman looks elsewhere

than to his district in making up his mind on foreign
issues. However, the reliance he puts on the President

 

cit., p. 56.

is?

98

and the Administration suggests that the calculation of

wwv~vwwv“~

 

Vwfiffiw

 

legislative initiative in foreign affairs has fallen
victim to the very difficulties of gathering and
appraising information that led Burke to argue that
Parliament rather than the public ought to hold the
power of decision. The background information and the
predictive skills that Burke thought the people lacked

 

 

are held primarily by the modern Executive. As a result,

 

the present role of the legislature in foreign affairs
bears some resemblance to the role that Bruke had in mind
for Ehe elitist, highly restricted electorate of his own
day.

This assumption, that congressmen rely on the executive because

of an essentially bipartisan consideration, belongs to the first of

the following three models of congressional delegation to the

executive.

Bipartisan Delegation MOdel

Given:

the condition that a transition in party control of the

administration is attended by no abrupt change in executive foreign

policy (for example, 1952-53, 1960-61)

If:

I. Constituencies are not likely to produce important short
run changes in the rank order of congressional roll call
support for executive foreign policy positions
(Because:

A. Constituencies have less effect on foreign policy
than on other major issues such as race and welfare

B. The rank order of the constituencies as perceived

 

2Ibid., p. 56. The executive branch does not take a position on

all of the foreign affairs roll calls. But it does often enough to make
possible the degree of reliance upon the executive branch that Miller
and Stokes hypothesize.

99

by the congressmen is not likely to change importantly

in the short run

1. The rank order of_the constituencies is not likely
to change importantly in the short run

2. Although congressional perceptions of constituency
position are relatively accurate, evidence of
clearcut change in constituency position is usually
required to alter perceptions held by men busily
attending many other phenomena)

II. Congressmen's (private) substantive attitudes on foreign
policy are not likely to produce important short run change
in the rank order of congressional support for executive
foreign policy (See "Reasons for Expecting . . . Consis-
tency," p. 17)

(Because:

A. The private attitudes of the congressmen are not likely
to change importantly in the short run

B. The few congressmen whose private policy attitudes
change importantly are likely to restrain [moderate]
the expression of this change, rather than abruptly
and obviously flipflop)

III.Congressmen rely on the executive because of its enduring
institutional superiority in background information and

predictiyefiskills

 

Then Expect:

I. Littlefigr no change in rank order of congressional support

100

for executive foreign policy positions after a transition
in party control of the administration

II. Whatever changes that do take place in rank order of
congressional support of the executive to be bipartisan
(that is, expect Eg_significant difference between the
proportion of Democrats and Republicans included among those
who increase their support of the executive and the pro-
portion of the Democrats and Republicans included among

those who decrease their support).

The following are examples of tables that involve small numbers
of changers and no significant differences between parties in direc—

tion of change. Both tables are consistent with the above model.

Party Party
of of
Changers Changers
D R S R
Direction of Change 1 l 2 increase 2 l
in the Support of the l 1 decrease 2 1
Executive across
Administrations 2 2 4 4 2

The above model is designed to explain congressional delegation
to the executive in terms of the dependence of the congressmen upon
the executive branch's superior infomation and predictive skills.
Numerous statements by academicians and journalists about the infor-
mation advantages if the executive and of bipartisanship in foreign

affairs seem to support the above model.

101

Some writers, however, have hypothesized that congressional
delegation to the executive depends on partisan considerations.
Professor Kesselman, for example, suggests that ". . . the President's
party label plays a significant part in explaining congressional voting
on foreign affairs."3 At no point does he argue that congressmen
delegate because of the informational advantages of the executive.
This explicit assumption, that congressmen rely on the executive be-
cause of the relation between his party label and theirs, belongs to
the second model: the "nominal party affiliation model." Although
the only assumption in the model that is different (from the assump-
tions of the bipartisan model) is the assumption about the reason for

delegation, note the effect of change in this assumption (III) on the

predictions.

Nominal Party Affiliation

 

(VParty Label?) Model

 

Given: the condition that a transition in party control of the admin—
istration is attended by no abrupt change in executive foreign policy
(for example, 1952-53, 1960-61)

If: I. Constituencies are not likely to produce important short

run changes in the rank order of congressional roll call

support for executive foreign policy positions

 

3Mark Kesselman, "A Note: Presidential Leadership in Congress
on Foreign Policy," Midwest Journal of Political Science, V, No. 3
(August, 1961), 284.

 

II.

102

(Because:
A. Constituencies have less effect on foreign policy than
on other major issues such as race and welfare
B. The rank order of the constituencies as perceived by
the congressmen is not likely to change importantly
in the short run
1. The rank order of the constituencies is not likely
to change importantly in the short run
2. Although congressional perceptions of constituency
position are relatively accurate, evidence of
clearcut change in constituency position is usually
required to alter perceptions held by men busily
attending many other phenomena)
Congressmen's (private) substantive attitudes on foreign
policy are not likely to produce important short run change
in the rank order of congressional support for executive
foreign policy ("Reasons for Expecting . . . Consistency,"
p- 17)
(Because:
A. The private attitudes of the congressmen are not likely
to change importantly in the short run
B. The few congressmen whose private policy attitudes
change importantly are likely to restrain [moderate]
the expression of this change, rather than abruptly and

obviously flipflop)

103

III. Congressmen rely on the executive because hisgparty label

coincides with their own

Then Expect:

I. Considerable change in the rank order of congressional

support for executive foreign policy positions after a

transition in party control of the administration4

11. Changes in rank order of congressional support of the

executive to be partisan (that is, expect a significant

 

difference between the proportion of nominal Democrats

 

and Republicans included among those who increase their

support of the executive and the prOportion of Democrats

and Republicans included among those who decrease their

support).

Kesselman reports the following pattern of change and consis—

tency for representatives common to the 8lst and 86th Congresses:

Demo—
crats
Number shifting toward isolationism 25
Number consistently isolationist 21
Number shifting toward internationalism 7
Number consistently internationalist 60
Total in both Congresses 113-

Repub—
licans

2
13
16
15

46

 

4Kesselman classified 50 of 159 House members (31%) as shifters,

pp. 286-87.

51bid., p. 287.

104

The table for the changers (shifters) reveals a significant

difference between the parties:

D R
Toward isolationism 25 2
Toward internationalism 7 16

One difficulty, however, in Kesselman's "before—and-after"
design is that the time lapse between the first ("before") measure
(8lst Congress) and the change in condition (January, 1953), and the
time lapse between the change in condition (January, 1953) and the
second ("after") measure (86th Congress), are so large that the
Congressmen's private attitudes and the constituencies' attitudes
may have changed importantly. Thus Democrats shifting away from in—
ternationalism and toward isolationism may have been responding to
factors other than the change in party control of the administration
of foreign policy. In fact, Rieselbach's work gives critics reason
to suSpect that a good part of the change reported by Kesselman did
not take place across administrations (from the 82nd to the 83rd
Congress), but within the Eisenhower administration.6 Rieselbach
reports that the percentage of Democrats who were isolationists
more than doubled (from 13.4% in the 83rd to 29.8% in the 85th) and

that the percentage who were internationalists declined from 72% in

 

v Vavv—VfiV wfifi“ wi—w—v i VVV

6Leroy N. Rieselbach, "The Demography of the Congressional
Vote on Foreign Aid, 1939-1958," American Political Science Review,
Vol. LVIII, No. 3 (September, 1964).

105

the 83rd to 47% in the 85th Congress.7 These figures add weight to
Kesselman's comment that:

It might have been more desirable, for the purpose of

controlling changes other than that of presidential

succesion, to have comgared the Eighty—second Congress

with the Eighty—third.

Another difficulty raised by Kesselman's analysis is less read—
ily remedied. In fact, it involves a problem in conceptualization

and measurement which in itself constitutes a challenging subject

for a dissertation. Attempting to explain delegation to the execu-

 

tive in terms of party label may seem inadequate to those who
hypothesize that the 1954 desegregation decision and subsequent events,
such as the conflicts over the 1957 and 1960 civil rights preposals,
substantially weakened the party identification of those Southern
Democrats who were wholehearted segregationists. Not only can it be
assumed that the party identification of segregationist congressmen
was weakened, but in at least a few cases, a discrepancy between

party label and party identification developed by the end of the

 

fifties. The Southern Democrats who most fervently attacked

"federal encroachment, the welfare state," "federal regulation,"

"bureaucrats,' and "creeping socialism,‘ while fervently supporting

"balanced budgets,‘ and "economy," probably came to feel as close or

somewhat closer to the Republican party. This is particularly

 

 

7Ibid., p. 578. Unfortunately, because Rieselbach's study is
not a panel study and we do not know whether these figures represent
individual change, these figures may cast doubt on Kesselman's in-
ference but not settle the question raised.

392, cit., p. 285.

106

plausible if the Democratic congressmen who repeatedly made compari—
sons (of the parties) in a manner unfavorable to the Democrats, who
supported Republican candidates, or who switched parties, were a
subset of the preceding group (which can be identified by content
analysis).9

If we assume that party identification and nominal party
affiliation at times diverge and if we assume that congressmen dele~

gate to the executive because of their party identification rather

 

than their party label, we must derive predictions that differ from
those of the preceding model. The following "Party Identification

Model" does this.

Party Identification Model
Given: the condition that a transition in party control of the
administration is attended by no abrupt change in executive foreign

policy (for example, 1952—53, 1960—61)

 

9

The direct effect of the race issue has been to weaken the
party identification of segregationist Democrats, but this alone
would not cause them to identify with the Republicans. An indirect
effect of the race issue on segregationist Democrats is the rein-
forcement of images of "federal encroachment," "spiraling bureau-
cracy," "centralization," and so forth, that clearly support the
model Republican position on many economic issues.

Southern Democrats who have the least (private) resentment
of increased opportunity for Negroes, and who are relatively liberal
on economic matters, may have a strong (private) identification with
their party in spite of a hundred public differences with the majority

of their party on civil rights roll calls.

If:

II.

107

Constituencies are not likely to produce important short
run changes in the rank order of congressional roll call
support for executive foreign policy positions
(Because:
A. Constituencies have less effect on foreign policy than
on other major issues such as race and welfare
B. The rank order of the constituencies as perceived by
the congressmen is not likely to change importantly in
the short run
1. The rank order of the constituencies is not likely
to change importantly in the short run
2. Although congressional perceptions of constituency
position are relatively accurate, evidence of
clearcut change in constituency position is usually
required to alter perceptions held my men busily
attending many other phenomena)
Congressmen's (private) substantive attitudes on foreign
policy are not likely to produce important short run change
in the rank order of congressional support for executive
foreign policy (See "Reasons for Expecting . . . Consis—
tency," p. 17)
(Because:
A. The private attitudes of the congressmen are not likely
to change importantly in the short run

B. The few congressmen whose private policy attitudes

108

change importantly are likely to restrain [moderate]
the expression of this change, rather than abruptly
and obviously flipflop)

III. Congressmen rely on the executive because their party

 

identification coincides with the executive's
Then Expect:

I. Considerable change in the rank order of congressional
support for executive foreign policy positions after a
transition in party control of the administration

II. Changes in the rank order of congressional support of the

executive to be partisan (that is, expect a significant

 

difference between those who identify with the Democratic

party and those who identify with the Republican party).

Should the slight divergence between nominal party affiliation and
party identification in the Senate during the Eisenhower-Kennedy
transition become considerably greater by the time of some future
transition in party control of the administration, we would pp£_ex-
pect significant differences between those who wear different labels
and we would have to seek measures of party identification.10 But

in the Senate at the time of the Eisenhower-Kennedy transition it

appears that only a few senators (including Thurmond, Byrd, and

 

vvfiw—wafi‘ fi‘r VV

10Presumably if a greater discrepancy took place and a score or
two of men were less supportive of the party whose label they bore
than of the Opposition party, the discrepancy between label and iden—
tification would become more of an issue and more indicators of party
identification would be available.

109

Lausche) identified with a party opposed to their nominal party.
Thus for the 1960—61 transition models II and III should lead to
similar although not identical results. (With model III Thurmond
et. al. would pp£_be counted as deviant cases if they are found to
have decreased their support in 1961; with model II they would be

deviant cases.)

The Before-and-After Analysis: Purpose, Span, Indices and Results
We often hear of the difficulty or impossibility of measuring
all or even most of the variables that contribute importantly to
complex behavior. But it is not necessary to measure all or even
most variables if by research design we minimize the opportunity of
certain variables to contribute to the behavior being explained. In
order to minimize the effect of changes in the constituencies'
attitudes and the senators' private (substantive) attitudes on
foreign policy, we must compare measures of the senators' behavior as
shortly before and after the January 1961 transition as possible.
Because we do not want the classification of many congressmen as
changers to be determined by one or a few foreign policy roll calls,
it is more appropriate to compare net support scores in the Congress
preceding transition with the net support scores in the subsequent
Congress than to compare "sessions" offering half as much data.
This is often important when analyzing the House because of the
paucity of foreign policy roll calls (on which the President took a

position).11

 

1For example there were only eleven "Eisenhower-issue roll
calls in the field of foreign policy in 1960." 1960 Congressional
[Quarterly Almanac, p. 110.

 

110

Because the substantive hypotheses relating senators' and
representatives' foreign policy positions to various independent

variables are concerned with change and consistency in foreign policy

 

position (broadly defined), and not just foreign aid, it would seem
that the substantive concept (foreign policy position) is better
tapped by a general measure than by a narrow quasi—scale.12 The
above considerations and a concern for interpretability led to the
Index of Partisan (foreign policy) Change.13 The Index of Partisan
Change utilizes more information than a cross-tabulation of Congres-
sional Quarterly Support Scores or a cross-tabulation of C0 Opposi-
tion Scores. The Index of Partisan Change = net (foreign policy)
support for the new President — net support for the preceding Presi—
dnet. Change as measured by the Index can most appropriately be
labeled partisan when two conditions are met: (1) there is a signi-
ficant difference between the direction of change on Republicans and
the direction of change of Democrats, (2) "changers" are defined as

those who score :40 or more points on the index. The stronger the

relationship between direction of change and party membership, the

 

12In previous tests of the hypotheses my narrow quasiéscales
revealed, when crosstabulated, that the changers were partisan but
the small number of items in such scales offered insufficient
reliability. It is probable that Kesselman would have found most of
the changers to be partisan changers even if he had chosen a differ-
ent subset of foreign policy roll calls. But many details of a cross—
tabulation of quasi-scales composed of few items depend greatly on
the items selected.

13It was hoped that the index would be more easily interpreted
than the somewhat esoteric general (first unrotated) factor.

111

stronger the probability that the change was partisan. For any
transition a positive score indicates change favorable to the in-
coming President. For the January, 1961 transition changers with
positive scores switched in a direction favorable to the Democratic
President; those with negative scores switched in a direction pp:
favorable to John Kennedy.

If we apply the Index of Partisan Change for the 86th—87th
Congress (using :40 or more points as the definition of changer) we
find that more than a third of the Senators common to the two
Congresses changed. Much more important is the finding that all but
one of the thirty—two changers changed in an overtlygpartisan direc-
‘pigp. The one deviant (from the standpoint of the Nominal Party
Affiliation MOdel), Lausche of Ohio, was less clearly deviant in terms

of the Party Identification Model}4

D R
Senators Who Scored +40 or More 14 0
Senators Who Scored -40 or More 1 17

 

4LChaples has reported that Lausche was the only nominal Demo-
crat in the 87th Congress who was, in terms ofgpolicy behavior, one
of thirteen ”Dirksen-Type Republicans" on foreign Policy. Chaples
also found that Lausche was a "Bush—Type Republican" on twenty-eight
agriculture votes. Peterson previously had found that (in a factor
analysis of the roll call positions of 99 senators and President
Kennedy) Lausche's prime loading was the conservative Republicans. In
the 87th Congress Lausche's CQ Party Opposition score (64) was more
than two-and-a—half times his Party Support score (25). Earnest A.
Chaples Jr., "The Use of a MOdified Form of Elementary Factor Analysis
for Studying the Public Attitudes of United States Senators..," Twenty-
third Annual Meeting of the Midwest Conference of Political Scientists.
D. J. Peterson, Michigan Academy of Arts, Sciences, & Letters, 1964.

112

Several thoughts occur to those of us who are not used to see—
ing such nearly perfect results. Would very many "deviant cases"
have been added if the marginal changers (Senators with scores in the
upper thirties) had been added? Or what if the before-and—after
analysis had been based on the 1960 and 1961 Sessions rather than the
86th and 87th Congresses (and the lower reliability of the measures
Offset by more demanding definition of change)? What if the Index Of
Partisan Change had been applied to the House of Representatives?

Adding the nine marginal changers has the unusual effect of

increasing the significance of the findings, Only one of the 41

 

 

(clearcut and marginal) changers changed in a direction inconsistent

 

with nominal party affiliation and he appears to have changed direction

in a manner consistent with his (psychological) or latent party identi-

fication.1

D R
Senators Who Exceeded +35 22 0
Senators Who Exceeded —35 l 18

If we use the Index of Partisan Change for 1960-61 and a more
demanding definition of change (ifiO) because of the lower reliability
when using Sessions rather than Congresses, we find that 29 of 30

changers changed in an overtly partisan direction.

 

5See previous footnote. If the nature of Lausche's "same-state
Senator (Young)" and the 1968 rejection by the Ohio Democrats are con-
sidered along with the material in the previous footnote it seems
quite likely that Lausche was acting in a manner consistent with a
Republican identification.

113

Relation Between Party & Direction of Change 1960—61

D R
Senators Who Scored +50 or More 8 0
Senators Who Scored —50 or More 1 21

The Index of Partisan Change for 1960—61 also reveals that
eleven Democrats, but no Republicans, increased their support in such
a manner as to yield scores of +40 to +76. Twenty—four Republicans
and three nominal Democrats decreased their support sufficiently to

earn a score of —40 to —119.

D R
Senators Who Scored +40 or More 11 0
Senators Who Scored -40 or More 3 24

It is interesting to note that the three deviants (from the

 

standpoint of the Nominal Party Affiliation Model) are the only

 

Senators identified in the Prospectus (p. 70) and alprevious paper

 

as nominal Democrats who probably "identified with a party opposed

 

to their nominalparty."l6 This suggests that nominal party affilia-

 

tion is a very strong correlate of direction of change (in support

of Presidents on foreign policy) only as long as it is an indicator

fivfi“

"causallagent," and that when nominal parly affiliation does not

 

reflect (psychological) party identification the Party Identification

 

6Thurmond, Byrd, and Lausche; Peterson, Michigan Academy,
9p, cit.

114

Model will be seen t°1§§919131939:2!9d19t change better than the
Nominal Party Affiliation Model.

When we apply the Index of Partisan Change for the 86th—87th
Congress to the House offiReppesentatives as a check on the generality
of the above findings, we discover impressive evidence about the
impprtance of party to the explanation of change in support of Presi-
dents on foreign policy. Of the 86 House members who changed +40 or
morglpoints only one changed in a manner inconsistent with his nominal

affiliation.l7 One of the most extreme partisans among the Congres—

Party & Direction of Change For House Members 86th—87th Congress

R D
Congressmen Who Scored +40 or More 1 57
Congressmen Who Scored —40 or More 28 O

sional Republicans, Melvin Laird of Wisconsin (~75), was recently

 

17The deviant (from the standpoint of Nominal Affiliation Mbdel)

was Weaver of Nebraska. His increase in foreign policy support can
largely be attributed to his 1962 Foreign Policy Support score of 78
(and Opposition score of only 17). The other Nebraska Republicans had
Support scores Of 11, 22, 22, and Opposition scores of 89, 78, 78.
Weaver's 1962 Foreign Policy Support score was associated with an
extraordinarily low Party Unity score of 46 and relatively high Party
Opposition (46). The other Nebraska Republicans had Party Unity Scores
of 98, 81, 86, and Party Opposition scores of 0, l4, 4. While Weaver
had a net of zero Party Support the average of the other Nebraska
Republicans was a net of 82. Again it appears that the nominal deviant
is less clearly deviant in terms of party identification.

115

elevated to a position with great impact on foreign affairs. Because
of the almost perfect relationship between direction of change and
party, and because of the extreme magnitude (—75) of Laird's change

there is a very high probability that the new Secretary of Defense is

 

one of the men who was least concerned with the content of merits of
executive foreign policy proposals during the 86th-87th Congresses.l8
Longitudinal analysis of both houses during the Truman—Eisen-
hower and Eisenhower—Kennedy transitions has revealed that only a
minority of the men in Congress are clearcut changers and that this

rather significant minority is highly partisan. Party label was so

 

highly related to direction of change in support that it is hard to
believe that American parties are as "meaningless" as many Of their
critics claim. Deviant case analysis suggests, however, that if party
label and party identification become less highly correlated the

Party Identification MOdel will offer better explanation of change
than the Party Label Model.19

The preceding analysis also has serious implications for the

numerous cross—sectional analyses of "party" and "nonparty" votes.

 

18Another Republican who probably had less than average concern

with the content or merits of executive foreign policy proposals, but
enjoyed a position of considerable influence was Senator Dirksen of
Illinois.

19The cases of the two Democrats least identified with their
nominal party, Lausche and Thurmond, suggest that there are forces
at work (party organization and personal factors such as the desire
to reduce cognitive dissonance) that inhibit widespread and enduring
discrepancy between party label and party identification.

116

A roll call on which the parties do not overtly differ has been
assumed to be a roll call on which party had no influence. We now
have a theoretical and empirical base for denying that an Index of
Likeness of 100 means that party was unimportant. It is now apparent
that many of the Senate foreign policy roll calls of the Eisenhower
era, that have previously been classified as p93 being party votes,

had high indices of party_likeness precisely because 17-20 Senate

 

Republicans assigned important weight to party_when voting on Eisen-

 

hower prOposals (thus causing them to vote much like the rather inter—
nationalist Democrats). If "bipartisan cooperation" assumes that
Senators of both parties vote alike because of shared substantive
assumptions and/or common delegation to the President (because "the
background infomation and the predictive skills ... are held primarily

by the modern Executive...");20 the scope of "blpartisan COOperation"

 

during the Eisenhower era is seriously exaggerated by including the
seventeen clearcut partisan Republicans among those supposedly en-
gaged in "bipartisan COOperation."

Once again we have found that cross—sectional analyses can be
seriously misleading unless associated with longitudinal analyses.
Perhaps the most appropriate conclusion is that of Heinz Eulau:

Most political behavior research is cross—sectional..

... As a result, most political behavior research has been

ahistorical..... It is, of course, possible to compare the

behavior of cross sections ... through time and venture
inferences about change from such comparison. But the method

 

20Miller and Stokes, pp, cit.

117

is unsatisfactory because changes may be in opposite direc-
tions and compensatory, making only for marginal results

that indicate little or no change..... The dilemma of
studying change through time must ultimately be solved by
longitudinal studies. Longitudinal research on political
behavior, using the individual as the unit of analysis, is
probably the most dependable technique of studying the pro-
cess of cause and effect, for it permits description of the
direction, degree, rate, and character of political change.21

 

fl v VV" TVj f‘ww VV —r w 1‘ ' a

21"Political Behavior," International Encyclopedia of the
Social Sciences, New York: Macmillan, 1968.

 

VII CONCLUSIONS

 

In this dissertation the longitudinal (non-cross-sectional)
perspective has been defined (within the context of studies of
Congressional behavior) to include not only dynamic studies such as
before-and—after designs or panel designs, but also "static" studies
which make systematic use of data from more than one Congress.

The first three chapters discussed problems associated with the
predominant cross—sectional approach and the nature and advantages
of longitudinal analysis. The longitudinal analyses in Chapter IV-
VI were designed to broadly check the methodological hunches about
the utility of longitudinal analysis and to further examine substan-
tive assumptions about consistency in Congressional behavior.

It was found that the sharp limits of the cross-sectional per—
spective reduce opportunities for measurement, explanation, and pre—
diction. The longitudinal perspective, on the other hand, increases
the opportunities to reliably and validly measure otherwise elusive
behavior such astfluapractice of, support for, or Opposition to
McCarthyism (IV). The longitudinal perspective helps us to identify
predictors of recurring behavior, identify pivotal or marginal

decision-makers and thus helps clarify the strategic alternatives

open to political leaders (V). And the longitudinal perspective helps

to explain subtle behavior that is likely to be misinterpreted in
a crossruntional perspective. For example, the partisan change in

roll call voting between the 82nd and 83rd Congresses

118

119

(associated with the transition in party control of the administra—
tion) contributed to the appearance of "bipartisan" foreign aid voting
in the "Eisenhower Congresses." It was easy, from a cross-sectional
perspective, to assume that bipartisan reasons for delegating to the
executive ("the constitutional and informational superiority of the
executive branch") explained the apparent "bipartisan behavior." The
longitudinal analyses reported in Chapter VI revealed, however, that
the surface pattern of bipartisan voting was in great part a product
of partisanship. Bipartisan motives are not likely to explain the
behavior of partisan changers who like Bennett of Utah varied their
support depending on which party controlled the administration of
foreign policy. But the such partisan behavior was not evident to
cross-sectional students of an "Eisenhower Congress."

Probably the clearest advantagecnflongitudinal over cross-
sectional analysis is in the study of change in individual behavior.

If we accept consistency—change in rank order and consistencyuchange

 

in magnitude as two basic dimensions for develOping a typology of
longitudinal analyses of consistency—change in legislative behavior
-- each of the three types exemplified in Chapters IV, V, VI fits in
a different one of the three cells representing logically possible
analyses. Chapter IV, a static study, fits in the cell representing
consistency in both rank order and magnitude. Chapter V fits in the
cell representing change in magnitude but consistency in rank order.
And Chapter VI, a dynamic study, clearly fits in the cell represent-
ing change in both magnitude and rank order. It is felt that the

findings reported in these diverse but complementary analyses better

120

evidence the broad utility of the longitudinal perSpective than
examples that are redundant in the sense that they illustrate but one

type or but one advantage of longitudinal analysis.

Some Substantive Conclusions Regarding Consistency—Change

In addition to the methodological insights regarding the
analysis Of consistency—change offered in the above chapters there
were substantive hypotheses about consistency and change in Congres-
sional behavior. With one possible exception, the subsequent research
(reported in IV, V, and VI) did not negate these assumptions.

Most Senators have rather well formulated policy tendencies by
their first term in the Senate. These policy tendencies are supported
by numerous interrelated and relative enduring values, sympathies,
identifications, beliefs, and attitudes, which guide cognitive pro-
cesses such as perception, thinking, and memory. In sum the belief
system is rather thoroughly develOped and not likely to change greatly
in the short run.

Given the excessive cost of genuinely scientific polling (in
depth) and multivariate computer analysis of the responses, many
politicians (like the Senators studied by Matthews and those studied
by Brimhall and Otis) hesitate to change much from "winning positions"
taken in previous elections. Knowing that they won when they
stressed x, y, and z hardly constitutes precise scientific evidence
of why they won, but it may Often be salient information for those
who lack precise information. Furthermore, the rank order of con—

stituencies is not likely to change dramatically in the short run on

121

the constituency variables most relevant to the overall policy
positions of Senators (for example, percentage of the labor-force
that is organized in labor unions, region, population per square
mile, percentage foreign stock, or urban rural). And change in one
constituency variable often is sufficiently outweighed by consistency
in others to inhibit dramatic overall change. Thus constituency does
not frequently cause grand reorderings of conservatives and liberals.

There is an ideological bias in the social exposure and communi-
cation patterns of Senators that results in social reinforcement of
their policy tendencies. That is "Senators, once in office,establish
working relations with like-minded [people]" and tend to "hear" dis—
proportionately from those who agree. This bias in exposure or
differential contact is due both to the predispositions of represen-
tatives and the predispositions of lobbyists, constituents and others
who disprOportionately contact (and reinforce views of) politicians
who lean their way.

These and the other informed hunches about consistency-change
that appeared originally in the prospectus (and more recently in
Chapter 11) receive additional support from the research reported in
Chapters IV and V, which like the research cited in Chapter 11, re—
veal impressive consistency over time. At first glance studies such
as those by Kesselman might seem to directly contradict the above
assumptions about Senatorial consistency. After all, Kesselman
concluded that the change in rank he observed for scores of House
members represented change in attitude. And Price also suggested

that when Senators such as Bennett changed from one of the most

122

anti-foreign aid positions in 1952 to the very top pre—aid class in
1953 they were changing attitudes. Given the intellectual tools em-
ployed by scholars such as Coombs and Rokeach, especially the individ—
ual compensatory composition model, we can now see that there is no
basis for the inference (by Kesselman and others) that change in rank
on foreign aid voting (associated with transition in party control)
represents change in attitude. The evidence suggests that Senators
such as Bennett change in rank on foreign aid in order to remain con-
sistent partisans.

Such consistency (as was reported in Chapters II, IV, and V) has
enormous significance for both the theorist and practioner of demo—
cratic politics. This consistency over time no doubt makes the job
of the electorate more manageable and probably helps explain the
extremely significant correlations between Conservative Coalition
Support Scores of House members and electoral support for Goldwater
in their Congressional districts (1964) and between Conservative
Coalition Opposition Scores of Senators and electoral support for
Humphrey in their states (1968). It would appear that electoral
politics need not be as random or devoid of policy relevance as
cynics have suggested.

Although thsse general conclusions cannot substitute for the
more detailed ones in the above chapters, it is hoped that they
suggest something more of the broad advantages of longitudinal

analysis —- a method that Eulau rightly views as underutilized.

BIBLIOGRAPHY

BIBLIOGRAPHY

Books

Adorno, T. W., Frenkel—Brunswik, Else, Levinson, D. J., and Sanford,
R. N. The Authoritarian Personality. New York: Harper, 1950.

 

Alker, Hayward, Jr. Mathematics and Politics. New York: Macmillan,
1965.

 

Almond, Gabriel and Coleman, James B. The Politics of the Developing
Areas. Princeton: Princeton University Press, 1960.

 

Almond, Gabriel and Verba, Sidney. The Civic Culture. Boston:
Little, Brown and Co., 1965.

 

Beer, Samuel, g£_§l, Patterns of Government. lst ed. New York:
Random House, 1958.

 

Braybrooke, David and Lindblom, Charles. A Strategy of Decision.
New York: Free Press, 1963.

 

Congressional Quarterly Almanac, 1948-1969. (Washington, D.C.:
Congressional Quarterly News Features, 1948-1969.

 

Easton, David. A Systems Analysis of Political Life. New York:
Wiley, 1965.

 

Eulau, Heinz. The Behavioral Persuasion in Politics. New York:
Random House, 1963.

 

Galloway, George. History of the House of Representatives. New
York: Thomas Y. Crowell Company, 1961.

 

Hempel, Carl G. Aspects of Scientific Explanption. New York: Free
Press, 1965.

Hempel, Carl G. FHESEQEEF§l§le Concept Formation in Empirical

Science. Vol. II, No. 746f the International Encyclopedia of
United Science. Chicago Press, 1962.

Hoffer, Eric. The True Believer: New York: Harper, 1951.

 

Kelman, Herbert C. (ed.). International Behavior. New York: Holt,
Rinehart, and Winston, 1965.

 

123

124

Key, V. 0., Jr. PoliticallpParties, and Pressure Groups. New York:
Thomas Y. Crowell Company, 1958.

Key, V. 0., Jr. Publichpini n 8 American Democracy. New York:
Knopf, 1961.

 

Lazarsfeld, Paul and Rosenberg, Morris. The Language of Social
Research. Glencoe: Free Press, 1955.

 

Lipset, Seymour Martin. Political Man. Garden City, N. Y.:
Doubleday and Company, 1960.

 

MacRae, Duncan, Jr. Dimensions of Congressional Voting. Berkeley
and Los Angeles: University of California Press, 1958.

Margenau, Henry. The Nature of Physical Reality, New York: McGraw—
Hill, 1950.

Matthews, Donald R. U.S. Senators and Their World. New York:
Vintage Books, 1960.

 

Parsons, Talcott. The Social System. Glencoe, 111.: Free Press, 1951.

 

Pritchett, C. Herman. Congress Versus the Supreme Court 1957—1960.
Minneapolis: University of Minnesota Press, 1961.

Robinson, James A. Congress and Foreign Policy-Making. Homewood,
111.: Dorsey Press, 1962.

 

Rokeach, Milton. The Open and Closed Mind. New York: Basic Books,
1960.

 

Rovere, Richard. Senator Joe McCarthy. New York: Harcourt Brace and '
World, 1959.

 

Stouffer, Samuel. Communisml_Conformity, and Civic Liberties.
Garden City, N. Y.: Doubleday, 1955.

Thomas, Norman and Lamb, Karl. Congress: Politics and Practices.
New York: Random House, 1964.

Torgerson, Warren S. Theony;andeethods offScaling. New York: John
Wiley and Sons, 1958.

Wahlke, John and Eulau, Heinz (eds.) Legislatiye Behavior. Glencoe:
Free Press, 1959.

White, William S. Citadel: The Story of the U. S. Senate, New York:
Harper, 1956.

125

Articles and Periodicals

 

Brimhall, D. R. and Otis, A. S. "A Study of the Consistency in
Congressional Voting," Journal of Applied Psychology, XXXII,
1948.

Cantwell, Frank V. "Public Opinion and the Legislative Process,"
The American Political Science Review, LV, 1946, 924-35.

Farris, Charles. "A Method of Determining Ideological Groupings in
the Congress," Journal of Politics, Xx, NO. 2 (May, 1958),
308-338.

 

Feigl, H. "Philosophical Embarrassments of Psychology," The American
Psycholpgist. Vol. XIV (1959), p. 127.

Flinn, Thomas A. "Party Responsibility in the States: Some Causal
Factors," The American Political Science Review, LVIII, 1964.

Hamilton, Thomas. "Social Optimism and Pessimism in American Pro—
testantism," Public Opinion Quarterly, VI, 280-83.

Kesselman, Mark. "A Note: Presidential Leadership in Congress on
Foreign Policy," Midwest Journal of Political Science, V, No. 3
(August, 1961), p. 284.

Kris, Ernst and Leites, Nathan. "Trends in Twentieth Century Pro-
paganda," Public Opinion and Communication. Edited by
Berelson and Janowitz. Glencoe, 111.: Free Press, 1950,
pp. 278-288.

Lowenthal & Guterman. "Portrait of the American Agitator," Public
Opinion Quarterly, XII, 417-429.

Margenau, Henry. "Philosophical Problems Concerning the Meaning of
Measurement in Physics," Measurements: Definitions and Theories.
Edited by John Wiley. New York, 1959.

Miller, Warren and Stokes, Donald. "Constituency Influence in Congress,”
The American Political Science Review, XVII, No. 1 (March, 1963),

45-56.

Rieselbach, Leroy N. "The Demography of the Congressional Vote on
Foreign Aid, 1939-1958,” American Political Science Review,
LVIII, No. 3 (September, 1964).

Rosenau, James N. "Senate Attitudes Toward a Secretary of State,"
ngislative Behavior: A Reader in the Theory and Research.
Edited by J. C. Wahlke and H. Evlau. Glencoe: Free Press,
1959, 332-47.

126

Royce, J- R- "Factors as Theoretical Constructs,"

chologist, XVIII (1963), p. 524.

American_Psyf

Shils, Edward A. "The Legislator and His Environment," The Uni—
versity of Chicago Law Review. XVIII, (1950-1951), 571-584.

 

Smith, M. Brewster. Determinants of Anti—Semitism: A Social-
Psychological Map. New York: Anti—Defamation League of B'nai
B'rith, 1965.

 

Stokes, Donald. "Spatial Models of Party Competition," The American
Political Science Review, XVII, No. 2 (June 1963), 368-377.

 

Unpublished Material

 

Chaples, Earnest A., Jr. "The Use of a Modified Form of a Elementary
Factor Analysis for Studying the Public Attitudes of United
States Senators." Twenty—third Annual Meetings of the Midwest
Conference Of Political Scientists, Bloomington, Ind., 1964.

Rosenau, James N. "The Senate and Dean Acheson: A Case Study in
Legislative Attitudes," Ph.D. thesis, Princeton University
(Princeton, 1957).