Coding Instructions

CENTRALIZATION of POWER
[This material is condensed from Chapter 10 of Kenneth Janda’s Political
Pa flies: A Cross-National Survey. New York: The Free Press, 1980.]

‘CE%TRALIZATION or powER’’ as a concept in the analysis the locus of power within a party with eight basic varia
of political parties relates to Anderson’s dimension of bles:
‘‘control” as a concept in organizational theory (1968,
pp. 392—396). More specifically, Anderson means the
I.’! Ntona!izjtion ci Structure
distribution ol control instead of the volume or sources 9.02 Se,ect,ng the Nt:cnaI Leader
of control, in this sense, it is identical with Duvercer’s 9.03 Selecting Parliamer.:ary Candidates
concepts of ‘‘centralization and decentralization,” 9.04 Allocating Funds
9.C5 Formuiattng Policy
which Duverger says, “define the way in which power is
9.06 Controlling Communications
distributed amongst the different levels of leadership”
9.07 Administering Discipline
(1963, p. 52). There is some tendency within the parties 9.03 Leadership Concentration
literature to confuse “centralization” with “organiza
lion,” or at least to neglect drawing clear distinctions
between the two. Duverger cannot be blamed for this Basic Variable 9.01
conceptual ambiguity, for he takes pains to distinguish Nationalization of Structure
between centralization and articuLation (one of our mea
Stcuctural arrangements of parties can attach more or
sure of organization). Nevertheless, discussions of party
less emphasis to national, regional, or local organs in in
politics frequently equate “strong’’ party organization
traparty politics. For example, if no national party or
with centralization of power.
gans exist independently of meetings of regional party
Perhaps the confusion is due to the empirical relation
leaders, there cannot be a basis for centralization of
ship between a high degree of organization and central
power or effective national direction. Occasionally, par
ization of power. Duverger himself asserts this relation
ties feature a highly differentiated set of party organs
ship, and it appears explicitly in the writings of other
yet leave authority relationships among these organs un
authors (see Sorauf 1964. pp. 160—162). Notwithstand
specified, which is conducive to “nationalization” of
ing any empirical relationship that may obtain between
structure but “decentralization” of power, Variable
organization and centralization, the concepts should be
8.01, “structural articulation,” measures functional
separated in comparative analysis. In this way, one can
differentiation at the national level without regard to
determine precisely the extent of their relationship and
hierarchical distribution of power. Variable 9.01 seeks
also analyze deviant cases: parties that are high in orga
instead to measure hierarchical distribution of power
nization but low in centralization or vice versa, The exis
without regard to functional differentiation,
tence of federalism, for example, might explain why
The Least nationalized structural arrangements—and
some highly organized parties are also decentralized,
thus the most decentralized—occur when local orga’i
and low levels of institutionalization should account for
zations, commonly at the municipal or county level, are
some parties that are centralized but loosely organized.
effectively independent of higher organs. One govern
Viewing centralization of power as the location and
ment condition that may facilitate this type of local
distribution of effective decision-making authority with
independence is federalism, although federalism is ex
in the party, we define the national party organs as our
pected to have its greatest impact in separating state par
reference point for “central” location. Thus a central
ty organs from national organs.
ized party is one which features the concentration of ef
fective decision-making authority in the national party Operational Definition. The preceding conceptual
organs, with a premium placed on a smaller number of considerations are incorporated in this scale. Parties
individuals participating in the decision. We seek to tap were assigned the lowest applicable code.
 Centralization oi Power

3 Local organizations, defined as constituency/mu 6 There is a discernible carsy hierarchy that has a
nic’pal/commune,county ievei or lower, are the single national council or executive committee at
only discernible structural element in the party: the top acting directly on me local organizations
tnese organizations may indeed demonstrate oil without interposing regional organizations: there
ierent farms of organization among themselves: are only national organs.
there is no formal provision for higher party organs.
Local organtzations remain the most discernible
Coding Results. Tables lOla and lOib present the
structural element in the party, but there are formal
provisions for federation of local organizat:ons that results of coding parties on 8V901, “nationalization of
involve their representation in higher party organs structure.’’ We coded about 80 percent of the parties on
at a regional or state level, although these organs this variable, had adequate means for AC901, and
are not elfectiveiy superior to the local organiza found no significant relationship between BV9Ol and
AC9OI. Nevertheless, the operationalization of this var
iable was somewhat wanting, as most of the parties were
2 There are dtscernibie regtona. party organs that
exercise their auihor:ty cver local orcan:zat:Dns, clustered at the upper end of the scale, with the top two
but tnere are no formal national orqans; naticnwide positions embracing about 70 percent of the parties.
coordination ci party activities 15 handled. it is SN Perhaps a scale that is moreThensitive to distinctions
tempted al all, through informal meetings of rec:on among parties at this end of the continuum could be
party leaders. constructed. On the other hand, perhaps the scale was
There are discerntbie national party organs thaI adequate but parties simply do not display much varia
provide tor formal representatior. of re:onal or tion within the concept.
state organs (Cr iocal organs in the absence of re
gional or state organs), but these national organs Basic Variable 9.02
are not effectively superior so the regiona ones. Selecting the National Leader’
whch in practzce can and do defy me natlonai or
gantzation. At a minimum, a party’s “national leader” is the per
4 There are discernible national party organs that are son who acts as the primary spokesman of the party in
more powerful than regional and local organiza the country’s communications media. At maximum, the
tions, but these national organs themselves consti national leader may not only fulfill the symbolic func
tute competing power centers rather then a single- tion of personifying the party, but he may also be the
peaked hierarchy: includes situations in whtch the real locus of power in the party and its effective leader,
parltamentary party organization challenges lead issuing orders that are regarded as “legitimate” by par
ershIp of other organs. ty members.
3 This 5 a d:scern:ble party hierarchy that runs from This variable isolates the set of procedures used to se
a single national council or executive committee lect the national leader. Selection procedures reflect
tnrough regional party organs down to local organi processes of negotiation, compromise, conflict, and co
zations: parliamentary organization is subordinated
to national organs. ‘Donald Sylvan assisted in writino his section,

TABLE lO,la TABLE 1O.1S

SUSFILE FLRSI suertta sscoo
BVUQL NATIONALIZATION OF STRUCTURE HAYIOHALIZATION OF STRUCTURE
Bygot

RELATIVE AOJUSTEO RELAI1.VC AUJUSTED

SOSOtUTE FREQUENCY FREQUENCY ISSOLUTE FREOUO4?? FREQUENCY
tATEGORT LA9EL coot FREQUENCY CERCENII FREQUENCY IPERCEN I IPERCENTI
IPERCENTI CATEGORY LABEL CODE
ONLY LOCAL ORGANS 1. .7 0 .8
.9 CMLI LOCAL ORGANS
MAINLY REGIONAL ORGS 2 2.2 2.7 .7 .8
MAINLY REGIONAL ORGS t
BOTH REG •
NAIL ORGS 3 tT 1.2.6 15,5 BOTH REC • InTL ORGS 3 21 lb-S 2.7.5
COMPETING NAIL ORC5 1.3 9.6 CONPETEN InTL 0R55 13 1.1
HIERARCHY • REG ORCS 5 46 3,j •1.8 HIERARCHY • lEG 0R55 5 55 31.4 45.9
DIRECT InTL CCNTRCL 5 30 22.2 S 29 19.7 24.2
27.3 DIRECT HAIL CONTROL
OLANE
._..J!_ u.S HISSING 0LAN 2_ ?!:. .!!H2
TOTAL 2.35 100:0 100:0 TOTAL 147 1.20.0 LiO.0

MEAN 4.718 510 DCV u .1st MEAN 4.757 510 DCV 1.124

VARIAOL€ *0901 VARX*8LE *0901.

6 • 44 MEAN 8.392
 VARIABLES, CODES, AND SutuAY STATISTICS

operation within the party, providing an indicator of its suit of “power struggies within tne tnp leadership
distribution of power. The most decentralized or “dem group.
ocratic” method of selection would involve [he direct 8 He :s selected by his predecessor; selections in
election of the leader by party members of even party buds the s:tuation to wh:ch the party was created by
‘‘identifiers’’ or supporters. The most centralized of the incumbent leader and no apparent means for
“oligarchic’’ method of selection would involve the transferring :esership have been established.
leader naming his successor. The range of alternative
between these points on the continuum might involve a Coding Results, Coding parties on their procedures
variety of “open” or ‘‘closed” procedures, including for “selecting the national leader” proved to be relative
out-and-out skulduggery. In this variable, we are less in ly manageable. As revealed in Tables lO.2a and lO.2b,
terested in the proprieties of the process than in the 80 percent or more of the parties were rated for BV902,
numbers and credentials of people who participate in the data quality codes averaged above 7, and there was
the selection. no significant correlation between BV902 and AC902.
Operational Definition. The higher the score on this Moreover, the operationalization captured considerable
scale, the more “centralized” is the party on leadership variance among parties in their manner of selecting
selection. leaders. While about half the parties featured centraliza
tion in the choice of leaders with little opportunity for
a No national party leader can be identified. rank-and-file participation (codes 7 and 8), more than
The leader is selected by vote of party identifiers or one-fifth selected leaders through national conventions
supporters. with the delegates representing local organizations (code
2 He ts selected by vote of party members, a smaller 3). Note that the proportion of parties that gave the cur
group than above. rent leader the power to select his successor (code 8) in
creased somewhat from the beginning to the end of the
3 He is selected by a national convention or cDr.gress.
the delegates to which represent regional or local l950s. This was because the new panics that arose dur
party organizations. ing the second half of our time period had not yet con
fronted the issue of leadership change. Perhaps these
4 He ts selcoted by the parliamentary delegation of
the party. newer parties should be filtered out of any analysis
involving 8V902 as an indicator of centralization of
5 He is selected by a national convention or congress,
power.
the delegates to which are appointed or primarily
determined by the national party organization,
6 He is selected by the national executive committee Basic Variable 9.03
or party council sub3ect to ratification by some Selecting Parliamentary Candidates
lower levels of the party.
Our concern with the distribution of control over the
7 He is selected by the national execjttve committee selection of parliamentary candidates stems directly
or party council withcut further review of the deci
sion; selections include change in leaders as a re ‘This section was drafted by Gilbert Roikin

TABLE IO.2a TABLE lO.Zb

SUB Ft L C FIRST SUIFILE SECONO
8 VS 0 A SCLECJ’ING THE NATIONAL LEADER 0V902 SELECTING THE NATIONAL LEADER

RELATIVE A OJ US I E C RELATIVE A 0.3 U S 7€ 0

ABSOLUTE F REDuENc, FREQUENCY ABSOLUTE FREQUENCY FREQUENCY
CATEGORY LASEL CODE FREQUENCY EeC E N I I IF CRC E NT I CATEGORY LABEL CODE FREQUENCY IF C BCE NT I PERCENT I
NO NATIONIL LEADER £ 3.7 NO WATONAt LEADER 2 3 2.0 2,4
ST PARTY NENOERS 2 .9 BY PAR1’Y MEMBERS a t .7
NAIL CONY. REG REP 3 27 23.9 WAIL COW, REt REP 3 32 at-a 26-b
PARL DELEGATION PARL DELESATIOR ‘. to S. I 7.9
NAIL CONY, NAIL SEL 5 I ¼.b 5.3 WAIL CONY, NATL SEL S r “I 5.6
NAIL CON • LOWER 0 6 7 5.2 6.2 NAIL CON • LONER ON 6 5 5.’ 6.1
NAIL COMMITTEE ALONE 7 30 22 • 2 26.5 NAIL COMMITTEE ALONE 7 35 aia 2’.. 6
CHaSEN By INCUNOENT a 26 19.3 23. i CHOSEN BY INCUMBENT fl-a
BLANN 22 16.3 MISSING BLANK 21 U,. I HISSING
TOTAL 535 2_SD .1. t30.O TOTAL ji,7 5.4.0 176.0

MEAN 5.510 STO CCV 2.295 MESh 5,635 570 0EV 2.201

VARIABLE AC9O2 VARIABLE AC0Z

N EAR 7.527 MEAN 7.02’.
 Centralization of Power

but
ition of parties as “Organ izations that 7 Selection is done by tr.e national organization,
from our defin or affiliat ed
ntatives in the selection must be approved by local
pursue a goal of placing their avowed represe organizations.
e for
government positions.” The salience of this variabl ss
Schatt S Selection is determined by a national party congre
parties research has been cogently expressed by or caucus.
selecting)
schneider: The nominating (he.. candidate or
has become the crucial process of the party . 9 Seleotton is determined by a national committee
process . . .

ines the party council.

The nature of the nominating procedure determ
ations is the
nature of the party; he who can make nomin
. It is there fore one of the best points Coding Results. The proportion of parties that
owner of the party candi
on of power within the could be scored on “selecting parliamentary
at which to observe the distributi dates” was limited by the numbe r of countr ies with
to Ranney,
party,” (1942, p. 64). Moreover, according functioning legislatures during our time perio d as welL
“the candidate selecting process is . . central to party
.
lOib re
as by the availability of data. Tables l03a and
the se
structure and activity” (1968b. p. 142). Clearly, 70 percen t of the parti es were succ ess
or of cord that almost
lection of candidates offers an important indicat fully assessed for their methods of candidate selection.
party. For pur
the centralization of power within the requir
) se The most common procedure was local selection
poses of cross-national comparison, we focus on the al (code 5). Thy next most frequ ent
amen tary or legisla tive candid ates. Party ing national approv
lection of parli nal com
method was central determination by the natio
candidates for ocher government positions might have the other extrem e of the cont inuum.
mittee (code 9). At
been studied, but no other office or set of offices pro of more than 90 includ ed in this as
only two parties out
vides as much opportunity for comparability. For coun the radical techniq ue of allowi ng
sessment employed
tries which feature a popularly elected chief executive, s to select candid ates directl y. The
party supporter
the preceding variable, “selecting the national leader,’’ ary elec
s for the top nation al of method of candiate nomination through prim
often includes party candidate to the Democ ratic and Repub lican
tions seems unique
fice. States.
parti es in the United
Although the method of selecting candidates might be
conceived in terms or a distinc tion betwee n “elec tion”
Basic Valabi. Lot
and “designation.” we view this as a side issue and di Allocating Funds’
rect our attention instead to the structural position and
concept
functional composition of party organs which dominate Regardless of the sources of party funds, a
the process, whether one of competitive election or non- cmbraced in variable 7.01 • the levels at which collection
in es
competitive desig natio n. We are interes ted in the num and allocation of those funds occur are important
ts in the decisio n and their locatio n in g the distr ibuti on of power within the party . As
- her of participan tablishin
the organizational hierarchy. Hence, the more restricted Hcidenheimer notes, “Crucial to the nature of the sup
i
the privilege to participate in candidate selection, the port transfer that takes place is who controls the distr
ibute d
more highly centralized is the party. bution of funds, and whether they are distr
on the
among individual candidates or parties, and if so,
finit ion, t applica ble value
OperationaiDe The highes basis of what criteria” (1963, p. 804). The organiza
in a
was assigned from this set: tional level that controls the allocation of funds is
nme nt of
powerful position to set priorities for the attai
I Nominations are determined locally by vote of party party goals.
supporters, for example, in a direct primary. t
It is difficult to think of ‘‘allocating’’ funds apar
2 Nominations are determined locally by vote of party from “collecting” funds, for funds can hardly be allo
members, for example, by vote in local party meet cated unless they are in hand. Collection can occur at
es’
ings. different stages, however, with the collection agenci
3 Selectton is made by local party leaders whose se alloc ation by diffe r
transferring funds for subsequent
lection must be ratified in some way by party mem ent agencies. In general, we cont end that power resides
nsi
bers. mainly in the agency or level of the party with respo
4 Selection is made by local leaders with little or no bility for allocating funds obtain ed either by direc t col
participation by rank-and-file members. leccion or transference of funds collected at another
di
S Selection is made locally, but the selections must be level, although a premium is put on funds collected
approved by the national organizatIon: this in rectly rather than transferred.
cludes cases that provide for local ‘recommenda Recognition must be given also to the lack of estab
tion rather than selection of candidates. lished organizational mechanisms with responsibilities
6 Selection is made by associat ions affiliated with the for collecting and allocating funds. In some panics, the
party or regional associations, but the selection
must be approved by the national organization. Oonald Sylvan assisted in writing this section.
 VARIABLES, CODES, AND SUMMAF.Y STAr:S::c:

TABLE lO3a TABLE l0.3b

SUCFILE FIRST SUBFILE SECOND
BV933 SELECTING pALjAHzNTAqY CANDIDATES 6V903 SELECTING PAPLIAMENTARY CANDIDATES

RELATIVE AOJUSTEfl RELATIVE AOJUS TED

iBSOLUES FREQUENCY FREQUENcY A 650 LUTE FREQUENCY EGU C N C V
CATEGORY LACEL COCE FREQUENCY (PERCENt) PERCENT) CATEGORY L8QEL Colt tout I (PERCENT I (PERCENT
CV PARTY SUPPORrEES z 1.5 2.2 97 PARIT SUPPORTERS 1.b
67 PARTY r€M6ERS 2 7 1.1 BY PARTY MENBEPS i’D
LOCAL LEADERS • 3 U a-I 12.1 LOCAL LEADERS • Ot 11 7-S ti-S
LOCAL LEADERS ALONE 9 I,.? 9.9 LOCAL LEADERS ILONE Is 6.8 9.9
LOCALLY - NAIL O 5 32 237 LCCALLT • NLTL OK 35 23.8 34-7
GRoups WtTH NAIL C 6 3 2.2 3.3 GROUPS ATTN NAIL ON 2 N 2.7 4.0
NAIL ORG • LOcAL O 7 10 7.4 11.0 NAIL ORG • LOCAL OK S I. I
NAIL PARTY CONGRESS 8 2 1.5 2.2 NAIL PARTT CONGRESS i-b 2.3
NAIL PARTY COMMITTEE 9 21 13.6 23.1 HATL PARTY COMMITTEE 27 Ia.’. 25.1
EL*N •‘. 32.6 MISSING SLANt ‘.6 31.! HISSING
TOTAL Efl lOi.E :32.0 TOTAL T

MEAN 5.753 510 DC) MEAN 5.92 500 0EV 2.235

YAREABLE *095! VARIACLE 109i3

‘SAN MEAN 6.59.

collection and allocation of funds may be diffused 6 Fjnos ore cof.ec:eo sr;mariiy Dy the national orga
throughout the entire organization, so that virtually all rjzo:icn. ;‘.‘n,c’r. aso exerc:ses respcns:bii::y br al
levels participate in a helter-skelter fashion. This lack of ,ocating unos.
ztructure for allocating funds is thought to provide for
complete decentralization of power, even more than the
Coding Results. Few parties divulge information
situation which obtains when the collection and alloca
concerning the collection and allocation of funds that
tion of funds are prime responsibilities of the local orga
completely satisfies the need for coding BV904. In
nizations.
about half the cases, thc coders felt that they had a suf
Operational Definition. This scale incorporates a ficient grasp of party practice to venture coding this var
number of combinations of levels of party organization iable; in the other half, they refrained from hazarding
in the collection and allocation of funds. The higher the any guess at all. Although the means for AC904 (given
level at which the funds are both collected and distrib in Tables lO.4a and lO.4b) attest to a relatively low level
uted, the greater the centralization of power in the par- of confidence in estimating party practice on this varia
tv. The highest level applicable was recorded. ble. no relationship emerged between 8V904 and AC904.
Concerntng the half of our parties that were evaluated
0 Responstbilttv for coflectino and oiiccat:na funds is for allocating funds, there were widely different pat
diii used throunnout the pony: little or structure terns. A few parties were tagged 0 to indicate a total tack
:s mposed on this aspect oi party activity of structure in fund raising, but none drew code 1, which
I Funds are coileciec and allocated primar:iy cy local signifies both local collection and allocation of funds. If
organizations—defined as constituency/countvrnu funds were collected locally, the universal practice was
nlcipal/cornmune level or lower. to shunt them upward for allocation (code 2), and this
2 Funds are collected primarily at the local level. but pattern was observed in about 30 percent of the parties.
large amounts are transmitted upward for distribu Comparable percentages of the parties fixed responsibil
ion by either the regional (state) or national organi. ity for collecting and allocating funds at the national
:ations.
level (code 6), which was assumed to be most conducive
Funds are collected by all levels of the orqanization to the centralization of power.
nut are transferred to the regional level icr alloca
tion.
4 Funds ore collected primarily at the recional level Basic Variable 9.05
and allocated by regional organizations. Formulating Policy*
S Funds are collected at all levels of the party, but “The attempt to create and exploit issues of public
large amounts are transferred to the national orga policy,” writes Schattschneider (1942, p. 136), “in
nization for allocation, or the national organ col volves leadership, discipline, and centralization,” The
ects most but ocal organs collect a significant
a ito un t. ‘This section was drattad by Gitberi Rctkin.
 Centralization of Power

TABLE lD.4a TABLE 10AS

FIRST SUOFILE SECOND
WaFfLE
ALLSCATNC FUN3S 6v9jb ALLOCATING FUNOS
BVsUb

A OJ US TED RELATIVE ADJUSTED

RELATIVE A LuTE FREQUENCY FREQUENCY
ABsOLUTE FREOUENCY tO U C NC FREQuENCY (PERCENT C PERCE MT I
LASEL COOt FREQUENCY C t RC[ NT I IPERCENTI CATEGORY LABEL CODE
:ATEGORY
3 5.6 HO SPECIAL STRUCTURE 5 S 3.b 6.6
NO SPLCIAL STRUCTURE 3.0
1.

17.6 33.8 LOCAL • UPWARD 01ST 2 Zb 06.3 31.6

• upwIRo GIST 2 2’
Ta REG ORG 3 is 3.0 S.6 TRANSFER TO REG ORG 3 3 2.0 3-9
TRANSFER
is 5.6 REG COLLECTXON—DIST is S 3.b 6.6
REG COLLECTION—D1St 3.0
5 18 25.’. TRANSFER TO NAT ORG 5 Li ii • 6 22. is
TRAN3PER TO NAT ORG 53-3
NATL TASK 6 1.7 12’S 23.9 PRINARILY NAIL TASK 6 22 I.S.O 21.9
PRIMARILY
BLANK 6’. ‘7.’ HISSING BLANK 71 46.3 HISSING
—

TOTAL 11.7 100.0

TOTAL 035 100 O 0.0

MEAN Jfl’75 STO GEM 1.816 MEAN 3.866 STO 0EV I.. 3h1

vARIABLE 1C9b VARXAOLC tOSeis —

NEAW 5.563 MEAN 5.616

information and expertise requisite to cogent policy S Major policy positions are formulated at the national
formation are presumably available principally at the level, but they are submitted to lower levels of the
top levels of party hierarchy. Nevertheless, parties differ party (local or regional organizations) for approval.
considerably on the extent to which lower organs partic 4 Local party organizations often enact policy resolu
ipate in policy making. Although there is some doubt tions and submit them I or national consideration,
about their influence on ultimate party policy, constitu but open argument in behalf of the resolutions is not
ency associations within the British Labour Party, for a common practice, md decision on the resolutions
example, continually formulate and press resolutions is not recuired.
upon the party at its annual conferences. The program 5 Majcr policy positions are determined by a national
of the Indian National Congress, on the other hand, is party congress, conference, or cnnventlon com
determined by a Working Committee dominated by gov posed of delegates from local or regional organiza
tions; policy positions may be stated provisionally
ernment officials. Numerous personalistic parties con
by individual party leaders, but approval of the
stitute the extreme case for the centralization of power position by the party congress is required before the
in making policy as the party position is expressed—or policy statement is considered to be effectively
“revealed,” as it were—by the party leader. binding as party policy.
“Policy” usually suggests the party’s position on sub 6 Major policy positions are determined by the na
smntive issues of government, but “policy” can also re tional committee, party counc:l, or parliamentary
fer to the party’s handling of intraparty affairs, such as party organizations; these positions are regarded as
campaign strategy. Our main concern in this variable is party policy’ without need for further approval by
to pinpoint the locus of power in determining the par other party organs.
ty’s position on substantive issues of government. Our 7 Major policy positions are determined and an
) concern is limited to the determination of party policy nounced by the party leader or a small subgroup of
and not its implementation in government. the national committee, for example, an execuiive
Operational Definition. A high score on this varia committee or “politbureau.” These positions are ef
ble corresponds to the determination and promulgation fectively regarded as “party poli.2y” without need
of policy at elite levels of the organization and is indica for approval by other party organs. (Not6 that a dis
tive of centralization of power. The lowest applicable tinction must be drawn between the leader’s pol
icy— sometimes pronounced by a leader who is also
score was coded.
the nation’s chief executive—and the party’s policy.
with respect to the United States, for example, the
0 Responsibility for formulating policy is diffused
president is not empowered to formulate or cnote
throughout the party; little or no structure is im
party policy, although he may implement It. The
posed on this aspect of party activity.
distinction is whether the announced policy is wide
1 Mtsf or policy stances are commonly determined by ly regarded as party policy or the personal policy of
Polling party members. the officeholder, acting in his capacity as a govern
2 Local party organizations enact policy resolutions, mental official and policy maker.)
argue them at the national level (usually the party
conference or convention), and frequently won Coding Results. The literature reveals enough of the
changes in party policy. political process within panics to assess the locus of
 VARIABLES, CODES, AND SUWAARY STATISTICS

TABLE lO.Sa TABLE lO.5b

S US F IL E FIRST SUIFILE SECOND
SVga! FORHULATINC POLICY eV9fl FORMULATING PCL!CT

RELATIVE AOJUSTCO RELATIVE AOJUSTEO

ASSOLUTC FREQUENCY FREDUENCY ABSOLUTE FREGUENCY FREQUENCY
CATEGORY LAREL CODE FREQUENCY (PERCENT) (PERCENT) CATEGORY LASEL CODE FREQUENCY (PERCENT) (PERCENT)
NO SPECIAL STRUCTURE * .7 .9 LOCAL. PUT TO NATL 2 3 2.6 2.4
LXAL. PUT TO PaATL 1 3 2.2 2,7 NATL, LOCAL APPROVAL 1 2 1.’. 1.6
NATL, LOCAL APPROvAL 3 i.5 1,8 LXAL, NO PRESSURE 1 .7 .6
LAL. NO PRESSURE a 1 ‘7 NATL PARTY CONGRESS 5 25 17.1 20.2
HAIL PARTY COKGRZSS 5 25 ia.5 22.1 NATIONAL COMMITTEE 6 55 37.’. ‘.4.4
NATIONAL COMMITTEE 6 ‘.9 36.3 N’.! LEADER OR SUOGROUP 7 38 25.9 10.6
LEADER OR SUBGROUP 7 29 21.5 26.’. BLANK 23 15.6 HISSiNG
ELAM Z5 16.5 HISSING TOTAL 1’.? 180.6
TOTAL t3 1;;:i’ STO 0EV 1.036
MEAN 5.9’.b
MEAN 5.00) SID 0EV 1.195

VARIASLE *09)5
WARIASLE 0C955 MEAN 633L
NUN 6.h36

power in “formulating policy” for 80 percent or more tains its position as the main vehicle for party communi
of our patties with relatively high confidence (see Tables cation. Therefore in considering the control of party
l0.Sa and i0.Sb). In nearly half these parties, policy for communications we look first and foremost at the party
mulation is centralized in the national committee, in an press.
other quarter, this function is entrusted to the party From the standpoint of centralization of power, our
leader himself or a small group of leaders. Of the re concern with controlling communications is the level of
maining quarter of the parties, almost all reserve the the party that controls the content and distribution of
major role of policy formulation for a national congress party newspapers (or other form of communication, if
or convention. Very few parties occupy the more decen relevant). Apart from the most decentralized situation
tralized positions on the scale, suggesting that there are which obtains when the party has no mass communica
definite empirical Umitations to decentralization in pol tions media of any kind, we regard the control of party
icy formulation regardless of the conceptual possibil press by local organizations (defined as contituency/
ities municipal/county/commune or lower) as a main condi
tion for decentralization of power. Power is centralized,
Ba.ic ValoN. LOG however, if the national press is considered to be
Controlling Communkallons* “influential” either in the sense of being widely read
Control of communications or information flow is across the country by party members and nonmembers
often cited as a key factor, if not the key factor, in con alike or intensively and closely read only by party mem
trolling an organization. Undoubtedly, informal con bers.
versation is one of the main forms of communication in Operational Definition. These criteria of (I) exis
virtually all organizations, and conversation among par tence of facilities for mass communication, (2) the locus
ty members is difficult to control directly. By skillful use of control within the party over publication of infonta
of more formal types of communication, however, a tion, and (3) the “influence” of the communication are
party can often structure the content of conversation in incorporated into the following scale. The higher the
directly. This leads us to consider the party’s access to score, the greater the degree of centralization.
and control of mass communications media as the ma
0 The party controls no important communications
jor indicator of controlling communications and there
media at any level.
by centralizing power within the organization.
Literature distribution (newspapers, magazines, party I Local organizations control the media without cen
sorship from the national organization: the media
documents) and electronic broadcasts (radio, television)
are regarded as “inlluential.”
constitute the two major types of mass communication,
with the latter being somewhat more “mass” than the 2 Regional organizations control the media without
censorship from the national organization; the me’
former. Despite the increasing use of electronic broad
dia are regarded as “influential.”
casts as a mass communications technique, especially in
countries with low literacy rates, the newspaper main- 3 Either local or regional organizations control the
media without national censorship, but the media
‘Donald Sylvan assisted in writing thi5 SEction. are not regarded as influential.
 Centralization of Power

4 Control of media is mixed: the national and local! Basic Variable 9.07
regional organizations share about evenly in the Administering Discipline
control of media, regardless of whether they are in
iiuentiai or not :nfluen:iai. The term “discipline” carries negative connotations
in the sense that it suggests punishment rather than re
5 National organization conzro;5 media of its own, but ward. If we take rewards to mean pleasures, satisfac
the media are nct regarded as tniluectiaL
tions, or gratifica’.ions that an individual person enjoys,
6 Naiionai orgarnzation does not have any direct con- and punishment to refer to other experiences that a per
trot of media of its own, but through censorship and son does not enjoy, then the denial of rewards when
distribution of information it controls the content of they are expected can be interpreted as a form of pun
local or regional media. ishment. Our concept of discipline thus includes both
7 National organization controls media of Its own, and rewards and punishments used as inducements to moti
the media are regarded as influential. vate individuals to conform to group behavior.
Before considering some typical techniques of disci
Coding Results. Tables lO.óa and lO.6b establish pline. we must identify the class of individuals who are
our experience in coding about three quarters of the par the targets of discipline. Obviously, any party member
ties for the locus of power in “controlling communica might be the target of party discipLine, but we are inter
tions,” The quality of the information in the literature ested only in the class of party members who are govern
is good, producing relatively high AC codes, but the sig mental officials or candidates for governmental office,
nificant correlation of .38 between BV906 and AC906 with particular emphasis on parliamentary or Legislative
reflects a tendency to conclude that the party does not office. We want to determine how the party disciplines
control a newspaper if the Literature fails to mention the behavior of these members in conformity with group
one. Although we have elaborated an eight-point scale behavior or party principles.
for BV906, ranging from 0 to 7, just three of the scale In reviewing specific techniques of discipline that are
positions account for about 85 percent of the parties. In associated with inducing conforming behavior from
nearly half the cases, the national committee controls governmental officials or candidates for governmental
influential media (code 7). Around 20 percent of the office, we restrict ourselves primarily to those that are
parties stand at the other extreme of the scale, publish purely within the party’s capability for delivery rather
ing no party newspaper at all. The significant fraction than the government’s. For example, rewarding con
of the remainder have national organizations in control forming behavior with a government contract may in-
of unimportant media (code 5). From the standpoint of valve the party directly, but this is not purely a party
interval scaling, one might question whether this latter act. On the other hand, expulsion from the party is a
scale position ought not be rescored toward the lower disciplinary action that is purely internal to the party.
end of the scale under the argument that control of “un One main reason for drawing this distinction is to con
influential” media contributes little to centralization of troL for the additional techniques of discipline in the
power. This possibility of rescoñng needs to be consid form of patronage that are available to governing par-
ered when analyzing the data. tie5 but not nongoverning parties,

TABLE 1C.6a TABLE 1a.Sb

SUBTILE FIRST SUBTILE SECONO
8V946 CONTROLLING COMIWMICATIONS 0V916 CONTROLLING CONNUNICATIONS

RELATIVE AOJUS9O AISOLUTE FA?

•BSOLUTE FREQUENCY FREQuEN V (PERCE)J)
CATEGORT LABEL CODE FREQUENCY IPERCENY) PtRcENfl cATEGORY LABEL COPE FREQUENCY (PERCENTI
NO IMPORTANT NEDtA I 21 15.6 25.2 NO IMPORTANT MEOIA 0 22 18.8
PEG ORG. INFLUENTIAL 2 1 .7 L.a kEG ORG. INFLUENT tAt 2 2 1.5 1.7
LOCAL, NOT IHFLUENTL 3 4 3.0 3.8 LOCAL. NOT IMFLUENTL 3 2.7 3.5
cm.TROL MIXED BY ORG i. 1.0 iS. 9.6 CONTROL MIXED ST ORG • 1.0 6.4 4.5
NAIL. NOT INFLuCUIL 5 1.5 10.’. 13.5 WATt, NOT INFLUENIL S 21 13 17.9
WRIt USES CENSORSNtP 6 5 3.0 3.8 WAIL USES cENSORSHIP 6 5 3... 5.3
NAIL. INFLUDdIZAL 7 58 37.0 58.1. WATt, INFLUENTIAL 7 53 38.1 45,3
BLANN 31. 23.0 MISSING BLANk 3 20.5 MISSING
TOTAL L35 sai.o ioi.D TOTAL 1.47 1.01.0 111.0

MEAN 5.788 STO 0EV 2.715 MEAN 5.803 SIb 0Ev 2,637

VARIABLE AC906 VARIABLE RCMB

NUN 6.365 N OW
 VARIABLES, CODES, AND Su?II{ABY STATISTICS

In addition to expulsion from the party (the most se lO.7a and lO.7b. Despite the relatively large proportion
vere form of party discipline—excepting bodily harm), of parties that were coded on 8V907 and the neat ap
techniques of discipline can include the denial of rights pearance of the data by categories in the table, the oper
normally accorded other members of equal status, such ationalization of this variable presented severe problems
as participating in party caucuses, receipt of party com for coding. First, it proved difficult to focus coding on
munications, access to party resources, and removal discipline of legislative members or candidates, as em
from party office. Although control over nomination phasized in the conceptual discussion, and there was lit-
for party candidacy constitutes an important technique tie consistency among the targets of party discipline.
of party discipline, it is not included within this variable Second. the literature seldom treats party discipline in a
because of its previous inclusion in BV903. All other comprehensive manner, requiring judgments to be made
techniques of discipline available within a party to in from sketchy accounts of what happened to a few lead
duce conformity on the party of governmental officials ers here and some party member there. Finally, the top
and candidates for government office are included in of the scale, code 4, should have been elaborated into
the concept. additional coding categories to allow further distinc
tions, for fully half of our parties gravitated to this sin
Operational Definition. The higher the code, the gle point- Some discriminatory power no doubt remains
more centralized the administration of discipline. in the scale, but it is not one of the better items in the
study. There was no significant correlation between
0 Either there are no discernible techniques ot discv BV907 and AC907.
pine to be administered tc party merncers or re
sponsibility for administer:ng dscipiine is diffused
throughout the party. with little or r.o structure im Basic VariabI. 9.08
posed on this aspect of party activity. Leadership Concentration
i Local organizations—defined as constituency/mu Duverger notes two facts that “have dominated the
nicipal/commune/county organs or lower—admin evolution of political parties since the beginning of the
iner majordascipiinarv echnoues.
century: the increase in the authority of the leaders and
2 Regional organizat:ons administer ma:or discipli the tendency toward personal forms of authority”
nary techn.ques: an :ricepenceni discipltnarv tnbu (1963, p. 168). In the most extreme situation, “The
nal or organ exists for disciplinary purposes, and it leader is omniscent, omnipotent, infallible, and infinite
acts independently of the national organ.
ly good and wise: every word that falls from his mouth
3 The national porhamentary pofly organization ad is true; every wish emanating from him is party law”
ministers the major disciplinary techniques.
(1963, p. 182). Although it does not necessarily follow
4 The executive ccmmi:tee. party counc:i, or party that power ts more centralized in a party with few lead
leader administers the major disciplinary tech ers rather than many, the fewer the number of individ
niq ues
uals involved in party decision making, the more condu
-

cive the conditions for centralization of power.

Coding Results. Approximately 75 percent of the “Leadership concentration” refers to the number of
parties were coded for “administering discipline,” and individuals who constitute the top party hierarchy and
they are distributed along the five-point scale in Tables who are regarded as key decision makers within the par-

TABLE l0.la TABLE 11Th

SUBFILE FIRST SUBFXLE SECOND

B V 93 7 AOMZNISTERINC OISCtPLINE BV937 AOMINtSIEflkC DISCIPLINE

EtA EYE ‘04(15 lEO RELATIVE ICJL’$ r€o

flECUENCY €0 U E N ABSOLUTE FREQUENCY ‘IiQUENCY
CATEGORY LABEL :oor PERCENYi iPERCENT) CAtEGORY LABEL CODE FRCQUEhCT I PERCENT’ IPEP C NT

NO 5TRUCTURE.’NEtHDOS a zt.s 28.7 NO STRUCTURE—METHODS 33 22.4 29,5

REGIONAL OR TRI8UnAL 2 5.? 7.9 REGIONAL CR rRtaup,a 2 ii a’s 8.4

PARL PARTY ORGAN 3 8.1 to 9 • PARL PARTY ORGAN 3 Li, 7-5 ¾ 8

NATIONAL LEVEL TASK 1. 39,3 52.5 NATIONAL LEVEL TASK So 39.5 St. I

BLANK 3’. 25.2 MISStNC GLAN’C 35 23.5 MISSING

ToTAL 135 13..: ‘CTAL 11.7 •0 t03.i

MEAN 2.58k 510 0EV - 7’.5 MEAN 2.545 510 DEW ‘755

VARIABLE AC4OT VARIABLE tC937

MEAN 6.000 MEAN 6,3,87
 Centralization of Power

ty. Commonly, these individuals constitute a subgroup 4 Leadership is collectively centralized into a group
ci three to five party ieaoers; the decisions oi this
of the party’s “executive committee” (see variable
grcup are regarded as cir.d:na on the party.
8.01), which often serves as the effective seat of power
-

in the organization. But sometimes such committees are S Leadership is sr.ereo by two individusis; their oint
only organizational showcases, which merely legitimate dec:s:or.s are regarded as o:nd:ng on she party.
decisions made by a dominant figure inside or even out 6 Leadership is exercised by one individual who can
side the committee. This variable aims at effective lead personally comm:s the party to binding courses of
ership concentration, whether formal or informal. action -

Operational Definition. The following scale incor

porates a range of alternative situations in the concen For this variable, the exercise of leadership is not to be
tration of leadership, from low to high. limited to the electoral stage, which often generates an
illusion of leadership concentration as party activists de
C Leadership IS SO dispersed that only local or re fer to the candidates in the conduct of campaigns. This
gional leaders can be identified; no one presents a
kind of leader-follower relationship may be short-lived
serious claim to the position of national party
with the resumption of normal poLitical life.
spokesman, much less central decision maker.
I Leadership is clearly decentralized: there are more Coding Results. The data obtained after coding al
than jive :eaders who Irequen:y make p;ononce most 90 percent of our parties on BV908 are presented
ments in behaif of the nat:onei party, but they are
not reqarded as authori:sitveiy binding spokesmen.
in Tables l0.Sa and 10.8b. What amounts to one-man
rule holds in almost 40 percent of our parties, with the
2 Leadership is decentr&zzeo: trom one to live per exact percentage somewhat higher in the later l9SOs.
Sons frequently speak in behalf of the party, but they
The remaining 60 percent of the parties are well distrib
are not regarded as authoritatively binding spokes
men by themselves for they do often disagree. uted along the scale, with collective leadership emanat
ing from a group larger than five (code 3) embracing the
3 Leadership Is collectively centralized into a group
of more than five party leaders; the decisions of th:s next largest segment of parties. There was a t:ndency
group are regarded ns authoritatively binding an for situations of one-man rule to be well described in the
the party: there may be a pany leader, but he alone literature, resulting in part for the significant correla
is not poweHul enough to control party pcltcy. tion between BV9OS and AC908 of 33.

TABLE LO.Ba TAELE 1O.eb

SUIFILE FIRST SUOFILE SECOND
BV9iI Lt*DERSNZF CONCENTRATED’. 5V908 LEADERSHIP CONCENTRATION

lOSOLUtE
RELATIVE &ojUSflO
CCDC
ABSoLuTE
FREQUENCY
‘htU
ItRCENTt
cfl
PERCENT)
CATEGORY LIBCL CODE FREQUENCY IP6CENTI IPRCEN1I CATEGORY LABEL
OECENTR*LIZEO. 5 37 ‘..Z DECCNTRALIZEO. 5 5 S 3.. 3.7
1. 5 -

DECENTRALIZED, 5—5 55 .S-i 12.5 OECENTRALZZEO. i5 2 1’. 3.5 SC.’.

cENTRALIZED, 9 3 52 23.7 26.7 cENTRALIZED, 5 35 25.9

CENTRALIZED, 3—5 ., i.z, t.z CENTRALIZED, 3—5 b 56 12.2 53.3

JOINT LEADERSHIp 9 6.7 7.5 JOINT LEADERSHIP 5 ii 6.6 7.’,

PERSONAl, CONTROL ‘2 31d. 35.0 PERSONAL CONTROL 53 36.1 35.3

BLAN( 55 i.t.t HI5SINS SLANt 52 6.2 HISSING

TOTAL 535 iG.i 1s:.1 TOTAL Sb? £00.5 £00.5

MEAN ‘.133 510 0EV 5.619 MEAN ‘.285 STO DCV S • 610

tIllABLE *0908 VARiABLE *0908

MEAN MEAN 6.519