Three pairs of eyes are better than one« or »Too many cooks spoil

Friedrich Funke 2002
[email protected]
A three-dimensional model of RWA
ISPP Annual Meeting Berlin 2002
Friedrich Funke
University of Jena
»Three pairs of eyes are better than one«
or
»Too many cooks spoil the broth«?
Validating a three-dimensional model
of right-wing authoritarianism
Please, do not quote parts of this script,
as the presented version may differ
from the written form.
Refer to this document as a whole.
Funke, F. (2002). »Three pairs of eyes are better than one« or »Too many cooks spoil the broth«? - Validating a
three-dimensional model of right-wing authoritarianism. Paper presented at the 25th Annual Meeting of the
International Society of Political Psychology. Berlin 16th July, 2002.
Abstract
A three-dimensional model of authoritarianism is presented and contrasted to the
classical unidimensional approach. Special
emphasis is laid on empirical validation of
the alternative measurement model based
on several studies relating the subdimensions of right wing authoritarianism
(aggressiveness, submissiveness, conventionalism)
to various political attitudes (xenophobia, attitudes to military actions, punitiveness). It can be
shown, that the three-dimensional approach can
explain additional variance, and – what is even
more interesting from the differential point of
view – the subdimensions have specific explanatory power for diverse dependent variables.
[email protected]
C:\Dokumente und Einstellungen\funke\Eigene Dateien\1_Funke\9 Veröffentlichungen\ISPP berlin\ISPPnew.doc
Friedrich Funke 2002
[email protected]
A three-dimensional model of RWA
ISPP Annual Meeting Berlin 2002
The title of my talk is somewhat puzzling, but I am sure that some of you guess, what I am
having on my mind.
Therefore it is always a good idea to touch at least SOME of the points promised in the abstract, in order not to frustrate the audience. I will try to do my best …
My major thesis today is the fundamental contradiction between the three-dimensional
theoretical conception of authoritarianism and the one-dimensional measurement with the
RWA-scale. I will NOT spend too much time on the theory. Instead I intend to demonstrate the benefit of my alternative approach with a handful of empirical examples.
But give me a couple of minutes to prepare the theoretical ground, so that we can harvest
the empirical crop in the end.
In the kindergarten of authoritarianism research we all have learnt that the authoritarian
personality comprises the following well-known 9 facets:
1. Conventionalism
2. Authoritarian Submission
3. Authoritarian Aggression
4. Anti-intraception
5. Superstition and Stereotypy
6. Power and "Toughness"
7. Destructiveness and Cynicism
8. Projectivity
9. Sex
Not all of them are necessary, none of them is a sufficient condition for the diagnosis of authoritarianism.
Staying in this metaphor in primary school, then, we have been taught that Bob Altemeyer
brought the nine facets down to the »Big Three«, which are nowadays the common denominator: His RWA-scale taps the attitudinal cluster of authoritarian aggression, authoritarian submission and conventionalism. Richard Christie noted in 1991: {quote} “The RWA
scale is the best current measure of the essence of what the authors of TAP were attempting
to measure” {unquote} (Christie, 1991) S.552. Love it or hate it, Altemeyer’s RWA-scale has
become the standard in the assessment of right-wing authoritarianism.
As a logical consequence of this conceptual “trinity” it seems almost trivial, that the items
of the RWA-scale should produce three highly correlated but still distinct factors: one for
aggression, one for submission and one for conventionalism. If truth be told, they actually
do not.
[email protected]
C:\Dokumente und Einstellungen\funke\Eigene Dateien\1_Funke\9 Veröffentlichungen\ISPP berlin\ISPPnew.doc
2
Friedrich Funke 2002
[email protected]
A three-dimensional model of RWA
ISPP Annual Meeting Berlin 2002
What you might find in principal component analyses is a two-factor solution. The data
will tell another story: the story of item wording effects. All protraits will load on one factor, the contraits on the other.
Pattern Matrixa
Component
1
2
P_A_21
.860
P_A_17
.827
P_ASC_32
.819
P_A_28
.772
P_AC_23
.732
P_AC_34
.720
P_ASC_7
.695
P_A_11
.658
P_AC_15
.653
P_S_30
.653
P_AC_5
.631
P_SC_9
.590
P_SC_12
.454
P_C_18
.430
P_A_26
N_C_22
.713
N_C_19
.710
N_C_25
.669
N_C_10
.645
N_C_24
.643
N_C_16
.628
N_C_6
.610
N_SC_8
.608
N_C_14
.603
N_C_13
.599
N_C_31
.555
N_SC_29
.545
N_SC_27
.490
Component
N_SC_20
.405
1
1.000
.540
N_C_33
.309
2
.540
1.000
Component Correlation Matrix
1
2
[email protected]
C:\Dokumente und Einstellungen\funke\Eigene Dateien\1_Funke\9 Veröffentlichungen\ISPP berlin\ISPPnew.doc
3
Friedrich Funke 2002
[email protected]
A three-dimensional model of RWA
ISPP Annual Meeting Berlin 2002
Bob Altemeyer explains the failure of three-factor solutions with the process of scale construction:
{quote}
“The answer is, that the three components are thoroughly intertwined among the items on
the test. All of the items tap at least two […]. It would therefore be nigh impossible to find,
say an authoritarian submission factor in such a tangle. (But the ‚tangle‘ is consistent with
the definition of right-wing authoritarianism as the covariation of the three.)”
{unquote}
The dark side of such items, however, is a methodological one: if more than one dimension
is touched in an item, we call it “double-barreled”. This is at severe drawback and a proverbial textbook-example for misconstruction of items, as you never know, which part of the
item stimulated the respondent to answer in exactly the way he or she did.
If we have the intention of dividing the indivisible Gordian knot and to separate the subdimensions from one another, we need to develop and apply more sophisticated measurement models or even three-dimensional scales. This is the main field of my past research,
but not of my talk today. I will not waste your time going down into the abyss of methodology and number-twisting. I will not explain, how I tackled the problem. I rather decided
to share with you the far-reaching theoretical implications.
I will not give theoretical proof neither. Talking about “proof” I say “The proof of the
pudding is in the eating”.
Therefore let me invite you to look into a handful of empirical examples. The data have
been collected in not quite a dozen of studies with more or less 2000 German participants.
Most but not all participants have been recruited in the internet.
Whenever I refer to the RWA scale I mean my German RWA³D-scale. It consists of 12 rephrased items from Altemeyers 1996 RWA scale, half of them are protraits, half of them are
negatively keyed. The items are not double-barreled, but tap only one of the subdimensions.
The benefit of a three-dimensional measurement can be demonstrated in several ways. Let
me draw your attention on just two of them:
1. Differential effects of the subdimensions
2. Differential effects depending on the field of research
Let us consider the classical situation of one predictor – namely RWA mean score x– and
one dependent validation criterion, e.g. prejudice y.
RWA
β
e.g.
prejudice
[email protected]
C:\Dokumente und Einstellungen\funke\Eigene Dateien\1_Funke\9 Veröffentlichungen\ISPP berlin\ISPPnew.doc
4
Friedrich Funke 2002
[email protected]
A three-dimensional model of RWA
ISPP Annual Meeting Berlin 2002
Plausibly there is only ONE regression coefficient β between x and y.
RWA
Aggress
RWA
Sub
RWA
Convent
β1
β2
β3
e.g.
prejudice
β1 ≠ β2 ≠ β3
(1.1)
Now imagine that we had three dimensions instead of one. Consequently we could build
the regression with THREE predictors. We could examine THREE regression coefficients
β1 through β3 and would gain the chance to discuss, why aggression explains more variance
than, say, conventionalism. This is scenario 1.
A second scenario could compare the predictive potential of the subdimensions for different
validation criteria. In other words: it might be, that authoritarian aggression is a powerful
predictor for racial prejudice, but not for sexism, where conventionalism could be a
stronger predictor.
RWA
Aggress
RWA
Sub
RWA
Convent
β11
e.g. racism
β12
β13
β21
β22
β23
e.g. sexism
β11 ≠ β12 ≠ β13
β21 ≠ β22 ≠ β23
β11 ≠ β21
β12 ≠ β22
β13 ≠ β23
[email protected]
C:\Dokumente und Einstellungen\funke\Eigene Dateien\1_Funke\9 Veröffentlichungen\ISPP berlin\ISPPnew.doc
(1.2)
5
Friedrich Funke 2002
[email protected]
A three-dimensional model of RWA
ISPP Annual Meeting Berlin 2002
Example I: Prejudice and Xenophobia
This is the prime example for a classical question of authoritarianism research.
The zero order correlation of RWA and a racism scale is .67. Together with the political
left-right-orientation the adjusted R² amounts to .50.
Unstandardized
Coefficients
Model
1
B
Standardized
Coefficients
Std. Error
Correlations
Zeroorder
Beta
Partial
Part
(Constant)
-.481
.191
LI_RE Politische
Orientierung
.289
.042
.261
.500
.316
.236
RWA Right-wing
Authoritarianism
.853
.059
.553
.666
.576
.499
a. Dependent Variable: AF Ausländerfeindlichkeit Kurzskala
If we enter the sub-dimensions of RWA instead of the global mean score R² increases
minimally, but another interesting result catches the eye:
Coefficientsa
Unstandardized
Coefficients
Model
1
B
(Constant)
Std.
Error
stand.
Beta
Correlations
t
Sig.
8.699
.000
Partial
Part
11.916
.000
.500
.500
.500
-2.161
.031
.000
.500
.340
.247
9.204
.000
.591
.408
.306
.290
6.789
.000
.579
.313
.226
.050
1.288
.199
.355
.062
.043
1.453
.167
.554
.046
(Constant)
-.405
.188
LI_RE Politische
Orientierung
.304
.041
.275
7.431
RWA_A Right-wing
Authoritarianism
aggressiveness
.364
.040
.356
RWA_S Right-wing
Authoritarianism
submissiveness
.395
.058
.060
.047
LI_RE Politische
Orientierung
Zeroorder
.500
RWA_A Right-wing
Authoritarianism
aggressiveness
RWA_S Right-wing
Authoritarianism
submissiveness
RWA_C Right-wing
Authoritarianism
conventionalism
2
RWA_C Right-wing
Authoritarianism
conventionalism
a. Dependent Variable: AF Ausländerfeindlichkeit Kurzskala
[email protected]
C:\Dokumente und Einstellungen\funke\Eigene Dateien\1_Funke\9 Veröffentlichungen\ISPP berlin\ISPPnew.doc
6
Friedrich Funke 2002
[email protected]
A three-dimensional model of RWA
ISPP Annual Meeting Berlin 2002
Only two of three sub-dimensions are significant predictors of prejudice; to be precise authoritarian aggression is the strongest predictor, followed by submission. Conventionalism
has no significant impact.
eps4
eps3
eps2
eps1
eps8
eps7
eps6
eps5
eps12
eps11
eps10
eps9
.65
RWA2AP
.50
RWA8AP
.25
RWA5AN
.07
RWA11AN
eps21Modellname:
.80
.71
.50
.26
.17
Aggress
congeneric RWA / congeneric AF (Standardized estimates)
Chi²=468.254 (143 df) p=.000 Chi²/df-ratio=3.275
IFI=.879 Tucker-Lewis=.837 AIC=600.254
RMSEA=.071 close fit=.000
GFI, AGFI, PGFI, cAIC nicht definiert bei FIML;
.68
.54
.63
RWA6SP
.59
RWA12SP
.12
RWA9SN
.04
RWA3SN
.79
.77
.35
.21
.20
RWA4CP
.43
RWA10CP
.27
RWA7CN
.06
RWA1CN
.45
.66
.52
.25
AF3_P
.40
eps20
Zeta AF
.51
.16
Submiss
.68
.21
AuslFeind
.76
.04
eps19
.24
Convent
.41
.74
.69
.66
.66
.78
.72
AF5_P
AF6_P
.48
.43
.44
AF1_N
.61
AF2_N
.52
AF4_N
eps18
eps17
eps13
eps14
eps15
eps16
.22
.40
.49
Politische Orientierung
The story continues when we direct our attention on another dependent variable:
Political Intolerance.
I used the Content-controlled Measure of Political Intolerance (Sullivan et al., 1979; Sullivan,
Piereson, & Marcus, 1982). Political tolerance is defined as »a willingness to permit the expression of ideas or interests one opposes.«.
Authoritarians show a tendency to be intolerant towards political opponents. This is especially interesting, as usually the majority of the sample has right-wing parties in mind, when
they are thinking of the least preferred party. In other words, in this special case the usually
»good guys« are the wicked.
But what about the differential effect of the sub-dimensions? Once again we se, that we see
more with three pairs of eyes:
[email protected]
C:\Dokumente und Einstellungen\funke\Eigene Dateien\1_Funke\9 Veröffentlichungen\ISPP berlin\ISPPnew.doc
7
Friedrich Funke 2002
[email protected]
A three-dimensional model of RWA
ISPP Annual Meeting Berlin 2002
Coefficientsa
Unstandardized
Coefficients
Model
1
B
Std. Error
(Constant)
4.367
.184
Politische Orientierung
-.054
.051
stand.
Correlations
Beta
t
Sig.
23.754
.000
-.051
-1.055
.292
Zeroorder
Partial
Part
-.051
-.051
-.051
Right-wing Authoritarianism
aggressiveness
Right-wing Authoritarianism
submissiveness
Right-wing Authoritarianism
conventionalism
2
3.608
.249
14.50
.000
Politische Orientierung
-.141
.055
-.134
-2.560
.011
-.051
-.124 -.120
Right-wing Authoritarianism
aggressiveness
.278
.053
.289
5.237
.000
.229
.248
.246
Right-wing Authoritarianism
submissiveness
-.067
.078
-.052
-.852
.395
.062
-.042
-.04
Right-wing Authoritarianism
conventionalism
.035
.063
.031
.563
.574
.023
.027
.026
(Constant)
a. Dependent Variable: Content-controlled Measure of Political Intolerance (Sullivan et al.,1982)
Now only one of the three sub-dimensions is a significant predictor: aggression takes it all.
eps4
eps3
eps2
eps1
eps8
eps7
eps6
eps5
eps12
eps11
eps10
eps9
.68
RWA2AP
.48
RWA8AP
.23
RWA5AN
.07
RWA11AN
.64
RWA6SP
.60
RWA12SP
.11
RWA9SN
.03
»congeneric RWA / congeneric Intol« (Standardized estimates)
Chi²=544.569 (143 df) p=.000 Chi²/df-ratio=3.808
eps21
.82
.69
.48
.27
.17
Aggress
.40
eps20
.80
.78
.33
.18
IFI=.821 Tucker-Lewis=.756 AIC=676.569
RMSEA=.079 close fit=.000
GFI, AGFI, PGFI, cAIC nicht definiert bei FIML;
.67
.16
Submiss
.05
-.24
PolIntol
.76
.16
RWA3SN
.20
RWA4CP
.45
RWA10CP
.25
RWA7CN
.06
RWA1CN
Zeta AF
.29
eps19
.45
.67
.50
.25
.24
Convent
.41
.56
.41
.32
.51
.87
.89
.32
INTOL1_P
.17
INTOL3_P
.10
INTOL5_P
.26
INTOL2_N
.76
INTOL4_N
.79
INTOL6_N
eps18
eps17
eps13
eps14
eps15
eps16
-.17
.40
.49
Politische Orientierung
A third example confirms our belief, that it might be a fruitful idea to use three subdimensions instead of the global mean score: punitiveness.
The participants of the study have been asked to put themselves in the position of a judge
or prosecutor and call for a prison sentence in four hypothetical criminal cases (e.g. sexual
abuse, professional misconduct of policemen). The term of imprisonment was the dependent variable.
[email protected]
C:\Dokumente und Einstellungen\funke\Eigene Dateien\1_Funke\9 Veröffentlichungen\ISPP berlin\ISPPnew.doc
8
Friedrich Funke 2002
[email protected]
A three-dimensional model of RWA
ISPP Annual Meeting Berlin 2002
Unstandardized
Coefficients
Model
1
B
stand.
Std. Error
Correlations
Beta
t
Sig.
8.080
.000
Zero-order
Partial
Part
(Constant)
6.535
.809
Right-wing Authoritarianism
aggressiveness
1.257
.212
.348
5.937
.000
.298
.301
.300
Right-wing Authoritarianism
submissiveness
-.372
.344
-.072
-1.083
.279
.078
-.057
-.055
Right-wing Authoritarianism
conventionalism
-.213
.240
-.053
-.890
.374
.000
-.047
-.045
a. Dependent Variable: Haftstrafe in Jahren (Mittelwert über Delikte)
As in the study presented before only aggression serves as a statistically significant predictor.
This pattern remains stable in all four crimes.
On balance after those few examples we can record, that it is worth having a closer look at
the sub-dimensions as measured with the RWA³D-scale:
1. They explain more variance of the dependent variable than the traditional mean score
does.
2. The predictive potential differs between the sub-dimensions. In the extreme case only
one of them is statistically significant. In this case the global correlation might be suppressed beyond recognition.
3. Dependent upon the research question the relative importance of one or the other subdimension may differ.
You might get the feeling now that the appetite grows with the eating, so I have prepared
another appetizer.
Let us have a closer look at the sub-dimensions against the background of the Five-FactorModel of personality, the »Big Five« or O.C.E.A.N.-model.
A
Φ
Aggression
S
C
.796
.595
Submission
N
E
O
A
C
-.517
-.001
-.169
-.462
.079
.307
.269
.140
-.177
.840
Conventionalism
Neuroticism .19
.06 -.05
Extraversion -.01 -.03 .00
Openness -.59 -.60 -.54
Agreeableness -.13 -.12 -.06
Conscientiousness .18
.27
.555
.33
I don't want to overinterpret the findings, as the differences are not dramatic. But they give
an interesting insight.
[email protected]
C:\Dokumente und Einstellungen\funke\Eigene Dateien\1_Funke\9 Veröffentlichungen\ISPP berlin\ISPPnew.doc
9
Friedrich Funke 2002
[email protected]
A three-dimensional model of RWA
ISPP Annual Meeting Berlin 2002
Not very surprisingly Openness to Experience is negatively associated with authoritarianism. This reminds us of the concept of closed-mindedness (Rokeach, 1960; Vacchiano et
al., 1968; Shaver & Richards, 1971; Helwig & Smallie, 1973). There is no visible difference
between the sub-dimensions.
Matters are somewhat different with conscientiousness. Hard-working, thorough perfectionists are not very aggressive, but they are high in conventionalism. Likewise neurotic persons tend to see the world as a dangerous jungle, they are easily threatened and this triggers
their authoritarian aggression.
Summing it all up: Limitations and possible misinterpretations
The first results are not dramatically revolutionary, but still justify a guarded optimism.
The tragedy of such ideas is, that once the trump card of a three-dimensional scale is
played, the idea becomes trivial. Being in your place I could understand if you said: oh, so
what, I knew it all along.
The problem here is not in the idea. But it is the easy thing difficult to put into action. I do
not plead for my RWA³D-scale as the last word in scale construction. I has a couple of
problems: some items have to be rephrased because they become outdated, they are very
skewed and therefore show a low selectivity. Another problem is, that the scale has not been
tested in English.
Anyway, I believe that I am on the right way, but it seems to remain a long one. To cut the
long story short:
It seems to be a fruitful idea to work with three-dimensional scales. This solves the fundamental contradiction between the three-dimensional theoretical conception of RWA and
the traditional one-dimensional measurement.
For didactic reasons I used my preliminary RWA³D scale and could collect some promising
evidence, that the three dimensions
1. explain more variance of the dependent variable than the traditional mean score
does.
2. The predictive potential differs between the sub-dimensions and
3. Dependent upon the research question the relative importance of one or the other
sub-dimension may differ.
So revisiting the topic of my talk I would say that three pairs of eyes see more than one. I
thank you for your attention and wish us all good luck with this idea.
[email protected]
C:\Dokumente und Einstellungen\funke\Eigene Dateien\1_Funke\9 Veröffentlichungen\ISPP berlin\ISPPnew.doc
10
Friedrich Funke 2002
[email protected]
A three-dimensional model of RWA
ISPP Annual Meeting Berlin 2002
The English version is only for illustration purposes. Please contact the author if you intend
to use the RWA³D-scale.
Variable
RWA³D
English
rwa1cn
»Man sollte seine eigenen Moralvorstellungen über People should pay less attention to the Bible and the other old
›Gut und Böse‹ entwickeln und weniger der Bibel traditional forms of religious guidance and instead develop
oder anderen alten, traditionellen Glaubenssätzen their own personal standards of what is moral and immoral.
Beachtung schenken.«
rwa2ap
»Was wir in unserem Land anstelle von mehr
›Bürgerrechten‹ wirklich brauchen, ist eine anständige Portion Recht und Ordnung.«
rwa3sn
»Die Zeiten, in denen sich Frauen ihren Männern
unterzuordnen hatten, sollten der Vergangenheit
angehören. Der ›Platz einer Frau‹ in der Gesellschaft sollte sein, wo immer sie möchte.«
rwa4cp
»Die Abkehr von der Tradition wird sich
eines Tages als fataler Fehler herausstellen.«
rwa5an
»Es gibt kein Verbrechen,
das die Todesstrafe rechtfertigen würde.«
rwa6sp
»Gehorsam und Achtung vor der Autorität sind
die wichtigsten Tugenden, die Kinder lernen sollten.«
rwa7cn
»Gleichgeschlechtliche Lebensgemeinschaften
sollten der Ehe gleichgestellt werden.«
rwa8ap
»Was unser Land wirklich braucht,
ist ein starker, entschlossener Kanzler, der das Übel zerschlagen und uns wieder auf unseren rechten Weg bringen wird.«
What our country really needs is a strong, determined leader
who will crush evil and take us back to our true path.
rwa9sn
»Es ist gut, daß die jungen Leute heutzutage größere Freiheiten haben, ›ihr eigenes Ding zu machen‹ und gegen Dinge zu protestieren, die sie
nicht mögen.«
It is wonderful that young people today have greater freedom
to protest against things they don’t like and to ‘do their own
thing’.
rwa10cp
»Tugendhaftigkeit und Gesetzestreue bringen uns
auf lange Sicht weiter als das ständige Infragestellen der Grundfesten
unserer Gesellschaft.«
rwa11an
»Es ist wichtig, die Rechte von Radikalen und
Abweichlern
in jeder Hinsicht zu wahren.«
It is important to protect fully the rights of radicals and deviants.
rwa12sp
»Die wahren Schlüssel zum ›guten Leben‹
sind Gehorsam, Disziplin und Tugend.«
The real keys to ‘good life’ are obedience, discipline, and sticking to the old and narrow.
A "woman's place" should be wherever she wants to be. The
days when women are submissive to their husbands and social
conventions belong strictly in the past.
Obedience and respect for authority are the most important
virtues children should learn.
[email protected]
C:\Dokumente und Einstellungen\funke\Eigene Dateien\1_Funke\9 Veröffentlichungen\ISPP berlin\ISPPnew.doc
11
Friedrich Funke 2002
[email protected]
A three-dimensional model of RWA
ISPP Annual Meeting Berlin 2002
Dipl.-Psych. Friedrich Funke
Institut für Psychologie
Methodenlehre und Evaluationsforschung
Steiger 3 Haus 1
07743 Jena
[email protected]
[email protected]
C:\Dokumente und Einstellungen\funke\Eigene Dateien\1_Funke\9 Veröffentlichungen\ISPP berlin\ISPPnew.doc
12