Coding and Scoring Operational Code Beliefs by Hand: The Verbs in Context System (VICS) Manual* Copyright 2006 ALL RIGHTS RESERVED Revised August, 2007 Stephen G. Walker Department of Political Science Arizona State University Tempe, AZ 85287-3209 *The author would like to acknowledge Mark Schafer and Michael Young as co-authors of one part of an earlier version of this manual, which was presented as Appendix 1. VICS Coding Protocol in a conference paper presented at the 1997 Annual Meeting of the International Studies Meeting. We later revised and published the ISA paper (without Appendix 1) as three chapters in Jerold Post, ed., The Psychological Assessment of Political Leaders (Ann Arbor: The University of Michigan Press, 2003). I have edited the original version of the appendix and expanded it by adding sections on calculating and interpreting the VICS indices. The methodological and theoretical foundations of operational code analysis are in Mark Schafer and Stephen G. Walker, eds., Beliefs and Leaders in World Politics: Methods and Applications of Operational Code Analysis (New York: Palgrave, 2006).
25
Embed
2008 KU Workshop VICS Hand Coding Revised Manual- Dr. Stephen Walker
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Coding and Scoring Operational Code Beliefs by Hand:
The Verbs in Context System (VICS) Manual*
Copyright 2006
ALL RIGHTS RESERVED
Revised August, 2007
Stephen G. WalkerDepartment of Political Science
Arizona State UniversityTempe, AZ 85287-3209
*The author would like to acknowledge Mark Schafer and MichaelYoung as co-authors of one part of an earlier version of thismanual, which was presented as Appendix 1. VICS Coding Protocolin a conference paper presented at the 1997 Annual Meeting of theInternational Studies Meeting. We later revised and published theISA paper (without Appendix 1) as three chapters in Jerold Post,ed., The Psychological Assessment of Political Leaders (AnnArbor: The University of Michigan Press, 2003). I have edited theoriginal version of the appendix and expanded it by addingsections on calculating and interpreting the VICS indices. Themethodological and theoretical foundations of operational codeanalysis are in Mark Schafer and Stephen G. Walker, eds., Beliefsand Leaders in World Politics: Methods and Applications ofOperational Code Analysis (New York: Palgrave, 2006).
1
Coding and Scoring Operational Code Beliefs by Hand:
The Verbs in Context System (VICS) Manual
Introduction
The Verbs In Context System (VICS) is a content analysis
coding procedure designed to retrieve information contained in
statements made by political decision makers. Certain properties
of each politically related, verb-based attribution are coded for
subject (actor) verb (action), and object (target). Each one of
these coded attributions is then combined in different ways to
calculate index scores that provide answers to the research
questions associated with a leader’s operational code. This
construct is an individual’s system of diagnostic beliefs
regarding the political universe plus strategic and tactical
beliefs regarding effective political action, which a leader may
also share with a group or larger entity such as the state.
Coding Verbs In Context
There are two general types of verbs. The first type is an
action verb, i.e., a verb that “does” something to an object.
These types of verbs are transitive verbs. A simple example is,
“The boy hit the ball,” in which “hit” is the transitive verb.This verb attributes an action to a subject (boy) who “doessomething” (hit) to an object (the ball).
The other type of verb is an intransitive verb, which does
2
not "do" anything. Instead of "doing," intransitive verbs
express "being," a form of existence that does not connote
action, and are usually represented by such words as "is" and
"are." A simple example is, "He is tall." Here the verb in the
sentence, "is," does not take any action toward an object, but
instead links a description, "tall," with a subject, "He."
The VICS scoring system only codes and indexes transitive
verbs. To begin coding a text, follow the numbered instructions
below. Before doing so, you may want to answer the questionnaire
in Appendix I, which will provide an anchor of your expert biases
against which to evaluate the VICS results for your leader.
1. Identify all transitive verbs. Read the decision maker's
comments, underlining consecutively each transitive verb in the
text as you read. It is important to understand the context of
the comments. Speakers may make only a brief reference to a
topic or an individual that was addressed in more depth in a
prior paragraph. For example, in one paragraph a speaker may use
a pronoun such as "he" or “them” without specifying the name of
the subject or object. In order to determine the proper name of
the subject or object, it may be necessary to refer back to a
previous paragraph or to an even earlier point in the speech.
2. Edit the identified verbs. Assign a number only to those
underlined verbs that are transitive and have some political
content, i.e., address issues or policies. Do not number
intransitive verbs or any verbs contained in routine introductory
3
comments, acknowledgments, jokes, or other non-political matters.
Some sentences contain multiple verbs. In such cases, each verb
is considered a separate “utterance” and should have a unique
number. Be aware, however, that some verbs will be in the
infinitive form (preceded by the word "to") and are thus
necessarily linked with the preceding non-infinitive verb to make
up one (and only one) “utterance” as a verb construction.
For example, a speaker might say: "We intend to send troops
to the region." Here there are two verbs in the sentence intend
and send. Because the second is an infinitive form (see the word
“to” before “send”), the sentence actually contains only one verbconstruction (utterance), namely, intend to send, and would thus
be coded only once.
After identifying and numbering the transitive verbs with
political content, edit each politically-related attribution
mentally (i.e., in your head) to form a sentence in the active
voice, which specifies the properties of interest. The form of
the reconstructed sentence should be as follows: subject—editedverb construction--target. The subject is the person or entity
"doing" the verb. The target is the entity on the receiving end
of the verb (to whom the verb is directed or targeted). Here is
an example of a sentence with only one utterance, underlined and
coded “#1” by the coder. Then the coder has coded the verb as a
Self (S) utterance, i.e., the speaker is the subject “doing” thesending.
4
#1SExample 1. "U.S. troops were sent to Bosnia for peacemaking
purposes."
However, the sentence in this example is written in passive
voice and needs to be reconstructed mentally as follows in order
to make the coding decision of attributing the verb’s subject asSelf (S) or Other (O). First, identify the verb construction,
which in this case is "were sent." Second, identify the subject,
or the person "doing" the verb. Inspection reveals that the
subject is not specified because the sentence is written in
passive voice. Ask yourself who "did" the verb, i.e., who did
the sending?
Since it is not specified in the original construction, the
task at this point is to reconstruct the sentence in active form
and make the specification. If President Clinton is the speaker
of this text, it can be inferred from this information that the
U.S. sent the troops--or perhaps the President sent the troops--
in the end both of these constructed subjects will be coded the
same, as "Self"(S).
Once the subject and the verb are specified in active form
as "U.S. sent," ask yourself toward whom the action of the verb
is directed. In this case, the target of sending troops is
inferred from the speech to be the Bosnian Serbs. The mentally
edited sentence now looks like this:
5
Subject--Active Voice--Target Object
"U.S.--sent troops--Bosnian Serbs."
In some cases the subject of the sentence, as written in its
original form, is not a political actor but an inanimate object
or force. Wherever possible, these subjects should be converted
to political actors, e.g., the President, NATO, Germany, Iraqi
people. To illustrate, "These reform efforts never got to this
point," said President Clinton referring to the efforts of former
presidents regarding NATO reforms. The syntactic subject of the
sentence is "reform efforts," but the sentence would be rewritten
from the context as the following attributions in the active
voice: "Previous U.S. presidents--never got their NATO allies--
to this point with reform efforts."
3. Code the edited transitive verbs. The next step is to
actually code the edited verb construction, and this step usually
follows quite easily once the verb has been edited mentally or
rewritten into the proper format. There are five coding
decisions to make for each transitive attribution in the active
voice.
First, specify the subject as either Self or Other, as in
Example 1 above. Simply analyze the edited sentence that already
specifies the subject and ask: Is the speaker referring directly
to Self, or to Self's country, government, cabinet, etc., with
whom the speaker identifies? Or is the speaker referring to
Other i.e., to other leaders, countries, political opponents,
6
people from another land, etc., with whom the speaker does not
identify? So, for instance, in Example 1 above, assuming that
Bill Clinton is the speaker, the coder has coded the attribution
as "Self," because he clearly is speaking either about himself or
his own country.
Second, specify whether it is a word or a deed. It is a
deed if the speaker is talking about a specific action that
someone has taken (past tense) or is taking at the time (present
tense). The verb in Example 1 is an example of a deed: U.S.--
sent--troops--for Bosnian peacekeeping. Words are not specific
actions, but rather are verbal intentions and expressions.
Examples of these include protests, demands, threats, promises,
or appeals, where the speaker is not actually taking an action,
but is talking about taking action. For example, "The U.S.--
promises to send--foreign aid--to the region" is an attribution
coded as a word because, even though the verb "send" is an action
verb, it is qualified by the verb "promise."
Also note that anytime an action verb is stated in the
future tense, it is a word and not a deed. Unless it has been
done or is being done, then it is only a statement of intent
(word) to take the action. For example, "The U.S.--will defeat--
Iraqi troops" is a future tense attribution ("will") with an
action verb ("defeat") and thus should be coded as a word.
Third, specify the valence of the attribution either as (+)
or (-), depending upon whether it is positive or negative,
7
conflict or cooperation, friendly or hostile. If the speaker
said, "We--have attacked--Germany--over support for racism," then
code the valence as a negative (-) deed because it is a conflict
action (past tense). On the other hand, some attributions are
clearly cooperative in their expression. For example, "We--will
work--with the United Nations--for peace" would be coded as a
positive (+) word (future tense).
In some cases the valence may not be apparent from the
attribution. Sometimes looking at the location of the statement
in a paragraph will be helpful. In some instances, there is no
logical valence associated with the attribution and, therefore,
do not code it (mark it "NC" for not coded). An example of such a
“Not Coded” verb is the example below in the utterance attributed
earlier to Bill Clinton about NATO and previously reconstructed
from passive to active voice as follows:
Example 2. “These reform efforts never got to this point.”#2S(NC)
“Previous U.S. presidents--never got their NATOAllies--to this point with reform efforts.”
Fourth, use the Word/Deed and + or - valence to locate each
transitive attribution in one of the six verb categories in
Figure 1. The verb categories are: reward, promise,
appeal/support, oppose/resist, threaten, and punish. These
categories are either positive (+) or negative (-) and either
words or deeds. Some of these can be derived simply from your
A quote taken from President Carter's January 4, 1980 address to thenation: "Massive Soviet military forces have invaded the small, non-aligned, sovereign nation of Afghanistan..."
1. Subject. The subject is "Massive Soviet military forces" which iscoded as other, that is, the speaker is not referring to his or her selfor his or her state.
2. Tense and Category. The verb phrase "have invaded" is in the pasttense and is a negative deed coded, therefore, as punish.
3. Domain. The action involves an actor (Soviet military forces)external to the speaker's state (the United States); therefore, thedomain is foreign.
4. Target and Context. The action is directed toward Afghanistan;therefore, the target is coded as Afghanistan. In addition, wedesignate a context: Soviet-Afghanistan-conflict-1979-88.
The complete data line for this statement is: other -3 foreign pastafghanistan soviet-afghanistan-conflict-1979-88.
Figure 1. Steps in the Verbs in Context System for Coding Verbs
9
previous coding decisions. Some require one more coding
decision. In Example 1 above, we have already specified the
statement as a conflict (-) deed, because we earlier inferred
from the speech that “Bosnian Serbs” were the target of the U.S.troops. In Figure 1 there is only one Negative Deed category
(Punish) with a value of (-3). Therefore, the verb in Example 1
is coded (-3) as follows:
#1S(-3)"U.S. troops were sent to Bosnia for peacemaking purposes."
(U.S.--sent troops--Bosnian Serbs.)
The Word categories in Figure 1 require more judgments by
the coder than the Deed categories, as Reward ((+3) is the only
possible VICS category for Positive Deeds. If an attribution is
coded previously as a Positive Word, then it is necessary to
decide whether it is an Appeal(+1) or a Promise (+2). The same
type of judgment needs to be made if the attribution is already
coded as a Negative Word--it is necessary to choose between
Oppose (-1) and Threaten (-2). In some cases this coding
decision will not be difficult, and it will be apparent from the
attribution itself. In other cases, the decision will be harder.
Some general indicators and rules-of-thumb for making these
decisions include the following:
In general, Promises are pledges to take some
positive action in the future. On the other hand,
Appeal/Support attributions are generally
rhetorical expressions about things that should
10
occur, or an activity the leader would like to
encourage.
A similar distinction can be made on the negative
side of the word category. A Threat is
essentially a negative pledge about taking action
in the future, while the Oppose/Resist category is
a negative rhetorical statement about things the
leader thinks should not occur or an activity that
s/he does not want to encourage or see happen.
Anytime the words should, ought and must appear in
the original sentence, two inferences follow:
first, a future tense is implied (indicating a
Word category) and second, that it belongs in one
of the two rhetorical categories, either
Appeal/Support or Oppose/Resist. The valence (+)
or (-) then dictates which of these two categories
is correct.
Fifth, identify whether the location of the utterance
(Subject—Verb—Object) is in the domestic policy or foreign policy
domain. If both Subject and Object are in the speaker’s domesticdomain, then code the utterance as domestic (D). If either
Subject or Object is in the speaker’s foreign domain, then codethe utterance as foreign (F). This category allows the coder to
combine or separate the leader’s foreign and domestic operationalcodes, which may vary significantly across the two domains.
11
Because “Bosnia” Serbs as Object is a target located outside the
speaker’s (Self) domain, the completely coded utterance in our
example is as follows over the underlined verb:
#1S(-3)F"U.S. troops were sent to Bosnia for peacemaking purposes."
(U.S.--sent troops--Bosnian Serbs.)
The domestic/foreign distinction does not strictly apply if
the speaker is not the leader of a nation-state. Leaders of non-
state actors do not have a territorial base that distinguishes
domestic from foreign territorial locations. However, it may be
useful to use a parallel distinction between “intra- and inter-”domains for groups, organizations, or blocs if the leader is a
member of a non-state entity. Depending on the explicitness of
the contextual content, it is possible as well for the coder to
code topical context such as trade, human rights, security, or
cultural issues, and actor-target context, such as transitive
verbs with particular bilateral combinations, e.g., Sino-American
utterances or Israeli-Iranian utterances, as shown in the data
-3 Punish:** -3 Punish:***_________________________________________________________________* Use f’s and d’s instead of asterisks to tally foreign anddomestic attributions and calculate the domestic and foreignoperational codes of the leader.
14
Let’s stipulate that a hypothetical text has twentytransitive verbs (N=20) with political content, coded and
distributed into the Self and Other columns as shown in Table 1.
Fifteen to twenty coded verbs by convention is the minimum number
of codable verbs necessary in order to include a text in a sample
of texts by the speaker. The Frequency (Freq.) columns show the
distributions for each type of transitive verb attributed to Self
and Other. The Weighted (Wtd.) columns show the frequency of
each attribution weighted by the scale value (+3 to -3) for each
type of transitive verb. The Percent columns calculate the
relative frequency of each attribution by dividing the frequency
of each attribution by the total number for Self or Other.
The Totals in Table 1 refer to the sums of the numbers in
each column. The Measures refer to the Mean (Mn.)or the Mode
(Md.) for a column, in which the Mean is the Weighted column’stotal divided by the number of attributions and the Mode is the
highest percent in the percent column’s distribution. Plug the
relevant information from Table 1 into the formulas for each
index, as shown in Table 2, to calculate operational code scores.
For example, the index for Philosophical belief (P-1) Nature
of the Political Universe is the sum of the %Positive minus the
%Negative attributions for Other where %Positive is the sum of
%Reward plus %Promise plus % Appeal/Support and %Negative is the
sum of %Oppose/Resist plus %Threaten plus %Punish. The index for
15
Table 2. A Hypothetical Leader’s Operational Code Profile________
Philosophical Beliefs
P-1. Nature of Political Universe (friendly +1.0/hostile -1.0)Index =(%Positive Other) minus (%Negative Other)
Passive Conflict (I1/P1) Active ConflictTYPE DEF QUADRANT TYPE B QUADRANT
Figure 2. VICS* Prediction Typology for Tactics and Strategies**
*I-1 and P-1 Indices are scaled along the vertical axis, and P-4 Indices arescaled along the horizontal axis.
**Reward, Deter, Punish and Compel tactics around the midpoint of the verticaland horizontal axes are variants of reciprocity tactics in which an actorinitiates either an escalatory (E) move or de-escalatory (D) move and thenresponds in kind when the target escalates (E) or de-escalates (D) in responseto the actor’s initial move. Appease, Bluff, Exploit and Bully tactics at theextremes of the vertical and horizontal axes are variants of unconditionalconflict or cooperation tactics in which an actor initiates either anescalatory (E) move or de-escalatory (D) move and then does not reciprocateafter the target escalates (E) or de-escalates (D) in response to the actor’sinitial move.
Historical Control scores for Self (P-4a) and Other (P-4b)
adjusted to P4d difference scores (Self P4d = P4a – P4b; Other
19
P4d = P4b – P4a) so that a unit change in P4 as a vector is
weighted equivalent to a unit change in I1 or P1, making the P4d
indices range between -1.0 and +1.0 just like I1 and P1. Using
these scores as coordinates, it is possible to locate a leader’simages of Self (I1, P4a) and Other (P1, P4b) in the four
quadrants of the typology, as shown in Figure 2.
Also represented in the typology are different strategic and
tactical orientations associated with each type of leader. The
tactics bisect each quadrant at a forty-five degree angle between
a symmetrical zone of likely reciprocity tactics with settlement
and deadlock as likely strategic outcomes versus an asymmetrical
zone in which reciprocity tactics are less likely and strategic
outcomes of domination and submission are more likely. Mapping a
leader’s location for Self and Other within these quadrantsspecifies predictions about what moves and tactics the leader is
likely to use in bargaining with others and what moves and
tactics the leader expects others to use.
In the case of our hypothetical leader in Figure 2, Self’sscores from Table 2 map a location within the Type C Quadrant of
active cooperation that predicts Self has a slightly stronger
choice propensity for a cooperative initiative in the form of
asymmetrical tactics of Exploitation (DDE), in which Self De-
escalates, but when Other De-escalates, then Self Escalates.
Self’s location close to the horizontal axis also reflects apropensity to shift between De-escalate and Escalate as tactical
20
initiatives while preferring settlement as a strategic outcome.
These predictions reflect the tension within the leader’sbelief system between the relative strength of Self’s belief incooperation and Self’s belief in historical control, which is
resolved as a vector location within this typology. The
historical control difference score for Self (P4d = .60 -.40) is
.20 to the right of the midpoint of +.00 on the horizontal axis
while the strategic orientation score (I-1 = +.07) is only .07
above the midpoint of +.00 on the vertical axis. This combination
of coordinates places Self to the right and below the forty-five
degree angle bisecting Quadrant C, in which the historical
control score is further away (+.20 v. +.07) from the origin
(center) of the graph than the cooperation score.
Self’s location of Other in the passive conflict Quadrant
DEF and within the zone of reciprocity reflects the image of a
political universe in which the greater distance from the origin
of Other’s P-1 score (-.25) v. P4d score (P4d = .40 - .60 = -.20)
leads to the prediction of a conflict initiative in the form of
symmetrical Compel tactics (EDD) of reciprocity by Other rather
than the asymmetrical prediction that Other will engage in Bluff
(EED) tactics. Other’s location near the vertical axis also
reflects a shift propensity to use Punishment (EEE) as a tactic
if Self Escalates in response with deadlock as a likely strategic
outcome. Self’s likely optimum response, therefore, is De-
escalate in order to encourage Other to De-escalate (EDD) toward
21
settlement rather than deadlock as a likely strategic outcome.
Conclusion
These examples of plotting the locations of Self and Other
within the typology of operational codes illustrate how
operational code analysis can profile a leader and predict both
initiatives and responses as well as tactical sequences and
strategic directions. It is also possible to: (a) plot multiple
coordinates of successive texts uttered by the same leader to
baseline a leader and detect changes in beliefs over time; (b)
map the coordinates of more than one leader within a state in
order to compare different leaders and identify areas of
convergence and divergence in their belief systems, in order to
forecast consensus or conflict between them and continuity or
change in their decision-making patterns should one leader
replace another.
Coding a leader’s public statements by hand has both
advantages and disadvantages. Coding a single speech by hand
gives a fast, detailed account of the leader’s diagnostic and
choice propensities at a single point in time. Automated coding
gives a leader’s average diagnostic and choice propensities overtime by coding a larger number of speeches more quickly at the
cost of a smaller number of verbs per speech. Both approaches use
the key VICS indices (P1,I1,P4) to plot diagnostic propensities
attributed to others and choice propensities attributed to self
in the domains of domestic and foreign policy or between dyads.
22
APPENDIX I
OPERATIONAL CODE QUESTIONNAIRE
Before you begin to code a leader’s public statements,please answer the following questions about the leader tothe best of your ability. For each question circle theanswer that best describes the leader’s general beliefsabout the nature of the political universe and generalbeliefs about the most effective exercise of politicalpower.Your responses can provide an anchor for you tocompare your beliefs about the leader with what you havelearned from the VICS content analysis.
23
OPERATIONAL CODE QUESTIONNAIRE
LEADER______________ Page 1 of 2
Directions: Circle the word that best describes your rating of theleader for each of the following items.
PART I. BELIEFS ABOUT THE POLITICAL UNIVERSE
P-1. NATURE OF THE POLITICAL UNIVERSE (Hostile to Friendly)