1
CULTURAL CONSENSUS ON CHARACTERISTICS OF GOOD LEADERSHIP IN AFGHANISTAN:
Policy Implications for the Development of Future Generations of Leaders
February 27th 2012
Ahmad Idrees Rahmani
Doctoral Fellow
Pardeed RAND Graduate School
2
TABLE OF CONTENT
1. Objective a. Policy Issues b. Policy Question c. Research Question
2. Background
a. History b. Leadership Literature c. Preliminary Study
3. Methods
a. State – 1: i. Sample
ii. Data Collection iii. Analysis
b. Stage – 2: i. Sample
ii. Data Collection iii. Data Analysis
4. Policy Conclusions
5. Timeline
6. References
7. Dissertation Committee
3
OBJECTIVE
Policy Issue:
As the Afghan and international communities cast their gaze toward the threshold
year of 2014, they must begin to lay the foundations for what will emerge as
effective leadership in a post-transition world (Wilson 2011). Hamish Wilson, a
senior stabilization advisor for the Foreign Office of British Government based in
Northern Helmand, believes two crucial issues will threaten the prospects of
stability over that time if they are overlooked – the strength of Afghan institutions
such as it’s security forces and government administration, and the effectiveness of
their leaders. Wilson goes further and makes it clear that there are simply not
enough resources to adequately accelerate the daunting task of developing local
leadership in time for the deadline of transition. “There are only six in my team, yet
we are working with the Police, Prosecutor, Governor’s team, Line Ministry officials,
local leaders, Council members... there is a real risk that it is too little too late...”
Finally, Wilson concludes that, the sacrifices of the past decade and the intensive
efforts of the next few years will only create the conditions for a durable political
settlement, for a lasting peace, if Afghan leadership succeeds1.
Policy Question:
Leaders will not succeed unless they posses the characteristics that are perceived to
be good, particularly, those that for the followers, are associated with success. This
is more of an issue in traditional societies where definitions of goodness and success
are the product of many cultural and societal norms and standards. While every
1 http://www.peopleinaid.org/pool/files/pubs/building-leadership-for-transition-afghanistan-(2011).pdf
The most dangerous leadership myth is that leaders are born – that there is a genetic factor to leadership. This myth asserts that people simply either have certain charismatic qualities or not. That's nonsense; in fact, the opposite is true. Leaders are made rather than born… - Warren G. Bennis
4
expert of Afghanistan agrees with Wilson’s conclusion, the policy questions that
need to be answered are:
1. What does good leadership mean for different groups of the Afghan
population? Is there any consensus on the concept of good leadership
amongst the Afghan population at the national level? If not, what variations
across different groups of population and different types of leadership exist?
2. To what extent the perceived characteristics of good leadership viewed as
achievable (versus being granted by nature)? What if Afghans associate
goodness with a leader’s identity? What if different groups of the Afghan
population associate goodness with different types of identity?
Research questions:
The aim of my research is to examine variation in cultural consensus among
different groups of the Afghan population over the most important characteristics of
good leadership. I am particularly exploring the degree of variation in key
characteristics of a President that are associated more with success and popularity.
The research questions that I am going to examine in this study are the following:
1. What are the key characteristics that Afghan people recognize as being most
important for a good leader?
To what degree do the characteristics that Afghans want to see in their
leaders vary across different types of leadership positions?
To what degree do these characteristics vary across different groups of
the Afghan population (ethnic, gender, generation, rural/urban, social
class, etc.)?
5
2. To what degree the perceived good characteristics of a President are
achievable (i.e. education, clean records, etc.) versus unachievable (i.e. ethnic
affiliation, religious affiliation, tribal affiliation, etc.)?
To what degree do these perceptions vary across ethnic, gender, age,
social class, and rural/urban population lines?
To what degree do these perceptions vary across different types of
leadership positions?
The study will use the answer to these questions to develop policy
recommendations to the country’s educational institutions, as well as the electoral
and the Parliamentarian institutions, to shape a platform through which future
generations of leaders are equipped with appropriate characteristics that are
perceived to be associated with good leadership. Understanding the degree of
variation of answers to these questions might also have significant implications for
broader Afghan and international policies that are beyond the scope of the research.
BACKGROUND & MOTIVATION
History
Regardless of what characteristics are most appreciated for good leadership, one of
the central elements of having effective influence among the followers is to possess
the characteristics that are culturally admired in a particular society. The future of a
country, to a great extent, is determined by the level of influence of its political
leaders, as much as by the vision they possess. From the founding fathers of the
United States to the great leaders of Asia, it has been the effective influence of
leaders that has allowed them to realize their visions.
Some characteristics of leadership have been discussed by different scholars as
being important in the context of Afghanistan’s political life. One of them that is
6
often emphasized more than others is that a good political leader2 has to belong to a
Pashtun ethnic group, or even more specifically to the Durani tribe. Pashtuns have
been the dominant ethnic group in modern Afghanistan since the mid-eighteenth
century and currently comprise about 40 percent of the country’s total population
(Barfield 2010: 24). “The share of power in Afghanistan [has] mostly been [based]
on Pashtun domination throughout Afghan history” (Jose Oberson 2002). Shahrani
(2000) believes that “the cornerstone of kin-based mode of Pashtun tribal, social
and political organization…has been the defining attribute of Afghan politics since
the creation of Pashtun-dominated centralized polity”. Some scholars have gone
further and have concluded that for the Pashtuns good leadership is not even
transferable across different generations. “Leadership cannot be transmitted and is
inherent in the individual…the good and great qualities of a leader are buried with
him in his grave” (Ahmed 1980: 158)
While it is true that for the most of the last two centuries3 Afghanistan was
politically led by a series of Pashtun leaders mainly coming from Durani tribes, and
it is also true that in 1921, when a Tajik leader rose to power, considerable political
instability was triggered in the southern parts of the country, there is no academic
agreement whether the revolt was more driven by the rise of a non-Pashtun leader,
or by other phenomena. There are, in fact, two major competing theories of group
dynamics that are often used to explain violence. In addition to the already noted
ethnic theory, there is a theory that attributes conflict to a disconnect between
Afghanistan’s traditional rural population and progress-oriented urban elites. “The
urbanized Pashtun in Qandahar or a Tajik in Kabul experiences a political,
occupational, and cultural milieu far removed from their fellow Pashtuns or Tajiks
inhabiting remote mountainous of Uruzghan or Badakhshan” (Barfield 2010: 65).
The rebellion of the rural population, mostly mobilized by traditional tribal and
religious values, “[has] less to do with [ethnicity or] Islam than it [does] with the
2 In this case, particularly, they meant “legitimate” or “acceptable” leader rather than good leader. 3 Except for the two decades of war and instability (1979 – 1991)
7
long-standing clash of values between luxury-loving urbanites and the puritanical
rural villagers who had come to wield power over them” (Barfield 2010: 65).
After President Dawood’s death in 1978 many different Pashtun tribes and non-
Pashtun leaders rose to power, yet all of them faced political instability. However,
only the non-Pashtun leaders were perceived to have been generating a Pashtun
nationalistic sentiment that contributed to the instability. “The resistance that
followed the Soviet invasion of 1979 as well as the subsequent civil war allowed
non-Pashtun ethnic groups to assert political and economic autonomy both from the
state and from Pashtun dominance. From 1992 to 1996, the mainly Tajik Jamiat-i
Islami party under President Burhanuddin Rabbani controlled the central
government. Pashtun opposition to a Tajik- dominated political order, and support
from Pakistan and Saudi Arabia, paved the way for the Taliban, a largely Pashtun
fundamentalist movement that ruled most of the country from 1996 to 2001”4 (ICG
Report No. 62 2003: 2).
However, post Taliban
political development
of Afghanistan, once
again, produced
conflicting results that
could not be explained
by either theory. First,
a Pashtun leadership
from the Durani
dynasty could not
bring political stability.
Second, both
4 Afghanistan: The Problem of Pashtun Alienation ICG Asia Report N°62, 5 August 2003, pp2
Figure 1 shows distribution of provinces to leading presidential candidates in 2004 election
8
presidential elections produced highly polarized results along ethnic lines that
included the votes from both urban and rural population.
The mapped data published by the Independent Electoral Commission (IEC) of
Afghanistan shows distribution of provinces to leading candidates driven mostly by
majority ethnic group in each province (Hamed Karzai Pashtun, Younus Qanooni
Tajik, Mahqiq Hazara, and Dostum Uzbek)5. With a few exceptions (Herat, Balkh,
Takhar, Badghis), every province with a majority ethnic group voted for the
presidential candidate belonging to that ethnic group. This ethnically polarized
result was worse in the second presidential election, in 2009. While these elections
were not perfectly done, and many additional factors influenced the voting behavior
of the Afghan population, one can still notice the relevance of the policy question for
the future leaders of the country. What is it that different groups of Afghan people
want to see in their leaders before they perceive them as good leaders? How do
they vary across different types of leadership positions? How could future
generations of Afghan leaders become more legitimate, more successful and more
influential? If no consensus nationwide, what composition of leaders from different
groups is viewed just and fair?
There is very little – almost no – academic literature that examines these important
policy questions in the context of Afghanistan. The few scholars who have studied
Afghan leadership culture and the dynamics of political leadership in the context of
tribal communities of Pashtuns tend to disagree with each other, and offer
conflicting theories.
With respect to the variations of characteristics of good leadership perceived by
different groups of the Afghan population, my literature search has found no
academic studies so far.
5 http://www.iec.org.af/Public_html/Election%20Results%20Website/english/english.htm
9
Leadership Literature:
Despite the fact that there is no academic literature on the variation of
characteristics of leadership in the context of Afghanistan, there is a vast literature
that discusses the topic of leadership in the context of societies other than
Afghanistan. While these studies are important and generally applicable to any
context, they are not adequate, and in some cases not relevant to the cultural mosaic
of Afghanistan.
A general assertion is made that leadership is primarily determined by the character
and personality of the leader as well as the environment in which certain events and
decisions engrave [a] leader’s image in the eyes of their followers (Chemers 2000).
Leadership research can be reduced to focusing on the functions that leaders must
perform to be successful. Leaders must produce an image that arouses trust in
followers, develops relationships with subordinates that enable subordinates to
move forward individual and collective goal attainment (Barlow et al 2003).
Kirkpatick and Locke (1991) argue for the importance of character as a leadership
trait. They conducted a qualitative synthesis of earlier research postulating that
leaders differ from followers on six traits: drive, desire to lead, honesty and
integrity, self-confidence, cognitive ability, and knowledge of the business in which
they are supposed to lead. Some researchers such as Barlow et al. believe that these
six traits define a leader’s character for his/her followers. Some other scholars
defined character as doing the right things despite outside pressure to the contrary
(Lickona, 1991). The root of the word “character” is the Greek word for engraving,
but according to Barlow, applied to humans, it refers to the enduring marks left by
life that sets one apart from others. Typically, enduring marks are set early in life by
our religious beliefs, parental influences, education, and early interactions as a child
(Barlow et al, 2003).
Other authors suggest traits we should look for in leaders: character, vision, and
political capacity (Gergen 2001). Clowney (2001) suggests that “ethic” comprising of
10
sincerity, honesty, and humbleness are the main traits of leadership in our era.
Josephson (1991) asserts that leader’s achievements are shaped by the collection of
dispositions, habits, attitudes, education, and skills. Barlow et al (2003) summarizes
the 12 dimensions of leadership traits in a more systematic way:
1. “Integrity. Consistently adhering to a moral or ethical code or standard. A
person who consistently chooses to do the “right things” when faced with
alternate choices.
2. Honesty. Consistently being truthful with others.
3. Loyalty. Being devoted and committed to one’s organization, supervisors,
coworkers, and subordinates.
4. Selflessness. Genuinely concerned about the welfare of others and willing to
sacrifice one’s personal interest for others and their organization.
5. Compassion. Concern for the suffering or welfare of others and provides aid
or shows mercy for others.
6. Competency. Capable of performing tasks assigned in a superior fashion and
excels in all task assignments. Is effective and efficient.
7. Respectfulness. Shows esteem for, and consideration and appreciation of
other people.
8. Fairness. Treats people in an equitable, impartial, and just manner.
9. Responsibility and Self-Discipline. Can be depended upon to make rational
and logical decisions and to do tasks assigned. Can perform tasks assigned
without supervision.
10. Decisiveness. Capable of making logical and effective decisions in a timely
manner. Does not “Shoot From the Hip,” but does promptly make a good
decision after considering data appropriate to the decision.
11. Spiritual Appreciation. Values the spiritual diversity among individuals with
different backgrounds and cultures and respects all individuals rights to
differ from others in their beliefs.
12. Cooperativeness. Willingness to work or act together with others in
accomplishing a task or some common end or purpose.”
11
While all of these characteristics of leadership are important, some characteristics
that leaders are born with, such as charisma, good looks, belonging to a noble family,
or belonging to a specific ethnic group are not discussed in the Western literature.
However, they might be relevant in some Afghan communities.
Preliminary Study:
Given the shortage of academic work on this subject in the context of Afghanistan,
and also, given that applicability of a cultural consensus model in the context of
Afghanistan needed some testing, I decided to conduct some preliminary data
collection and analysis to see whether there are variations across a small sample of
respondents6, and whether such variations are detectable by the methodology
chosen. To do this, I interviewed 60 average Afghan citizens (20 Hazaras, 20
Pashtuns and 20 Tajiks) and gave them the question of; could you please list for me
the main characteristics of; 1) a good human being, 2) a good leader, and 3) a good
president. The respondents were allowed to freely list the characteristics for each of
these personalities on a piece of paper based on their personal preferences.
The responses were than categorized based on repetition of key words, and turned
into two mode matrices of respondent by items. If an item was mentioned by a
respondent, the cell corresponding to the item (in the column) and the respondent
(in the row) was scored 1, otherwise 0. Even though free listing does not produce
comparable data across different respondents7, the result do suggest a general
cultural domain for key characteristics of leadership as different groups of people
included similar items in their free lists. The data also detected some variations
across different ethnic groups and considerable variations across different types of
leadership. Thus, seems to confirm the relevance of this thesis and applicability of
the methodology chosen for the study.
6 Samples were mostly selected in the urban area of Kabul City. 7 Because informants responded based on different assumption, and based on how their memory worked at the time of interview
12
Furthermore, UCINET social network software was used to compute the consensus
amongst respondents, and visualize the variations across different groups of people.
In the graph below a red dot represents a respondent and every blue cube
represents an item (characteristics of a President).
Respondents 1 to 20 are Pashtuns, 21 to 40 are Hazaras, and 41 to 60 Tajiks. The
more lines going through a blue cube suggests more popularity of that characteristic
among different groups of respondents. A blue cube surrounded by red dots
suggests more consensus over that characteristic than a cube appearing on the
margins of the chart. In the chart above, “education” and “ethnically not biased”
seem to be more important characteristics of leadership for these respondents than
being an Afghan or being a Muslim8. Even with a very small sample and very basic
analysis9, variation across different groups of people and different characteristics of
leadership appears to be detectable.
8 When we asked them what do you mean by “An Afghan” the response was not having two passports which means not
having double citizenship 9 The actual cultural consensus analysis is not yet applied on this data. Please refer to the methodology section for more detail in this regard.
Figure 2 Consensus on characteristics of a successful and popular president based on 60 responses in Kabul City
13
Characteristics more associated with a good human being (our control question in
this preliminary study) are completely different from those of the President and the
political leader. This is a good indication that even with such a small number of
samples, variation of characteristics across different types of leadership position is
detectable. The similarity of characteristics between a President and a political
leader is most probably due to respondents’ assumption that the position of a
political leader is as important as that of a President, or possibly because of having
very similar traits. In fact, those two terms are used interchangeably in Afghanistan
except that the term political leader is also used for a religious leader (Jehadi
leader), a political party leader, or even a highly powerful tribal leader, who are
political active, but the term President is used only for one position—the head of the
state.
Figure 3 Consensus on characteristics of a successful and popular political leader
14
Figure 4 Consensus on characteristics of successful and popular human being (control question)
METHODOLOGY
Since my research questions are designed to detect and describe variations in
Afghan’s perception of good leadership characteristics, and there is no quantitative
data available in this regard, I will need to collect my own primary data. Collecting
primary data often entails survey research. Because we don’t know the right answer
to what constitutes key characteristics of good leadership for the Afghans, we
cannot ask questions for which the right answers are a priori known. In the absence
of right answers, a good methodology for producing good results is the Cultural
Consensus Analysis Model (Bernard 2006). Cultural consensus analysis uses a
mathematical model to determine the degree of shared knowledge within groups
and estimates the “culturally correct” answers where an answer was previously
unknown…the analysis initially solves for individual estimates of competency by
factoring an agreement (correlation) matrix among informants10. The ratio between
10 Romney et al. (1986) is a very good resource for understanding how the mathematical framework of the model works.
15
the first and second eigenvalues determines whether a single factor solution exists,
which would indicate a single, shared cultural belief system (Chavez et al 1995). As
Russell Barnard describes in his book, the theory behind the model has three
assumptions:
1. Informants share a common culture and there is a culturally correct answer
to any question you ask them. The culturally correct answer might be
incorrect from an outsider’s perspective (as often happens when we compare
folk knowledge about illness to biomedical knowledge). Any variation you
find among informants is the result of individual differences in their
knowledge, not the result of being [a] member of [a] culture.
2. Informants give their answers to your test questions independently of one
another.
3. All the questions in your test come from the same cultural domain – that is,
things that can be listed, like kinds of animals or hand tools, or things you can
do on a weekend. People can be competent in one domain but incompetent in
another. Cultural consensus method must be used for people who are
knowledgeable about a particular domain.
To use the consensus technique, simply give a sample of informants a test that asks
them to make some judgments about a list of items in a cultural domain. You can use
true-false questions, Likert scale questions, multiple-choice questions, or fill-in-the-
blank questions. The cultural consensus model is an important contribution to the
social science method (Bernard 2006)11.
In this research I will employ a cultural consensus model in the second stage, but
before that, I will use some other other methods, such as free listing and pile sorting,
to determine the cultural domain of good leadership and key items of the domain:
11 For more technical details about the cultural consensus model please refer to Romney et al. 198611, Caulkins 2001,
de Munck et al. 2002, Furlow 2003, Swora 2003, Harvey and Bird 2004, Jaskyte and Dressler 2004, and Miller et al. 2004.
16
1. Stage One: Identifying the range of characteristics that Afghans associate
them with good leadership. Pile sorting of those characteristics to produce
the key bins of characteristics, and defining each bin through a series of
analysis.
2. Stage Two: Using the answers from stage one to develop a structured
questionnaire that can be administered on different groups of the Afghan
population to determine the cultural consensus among and between different
groups. I will also use, additional methods such as factor analysis to cross
check my analysis, and finally apply those characteristics to some key
political leaders so the readers understand what each characteristics mean to
different groups of the Afghan population.
Stage – 1:
The set of open-ended questions that will be employed for the first phase of the
study are intended to produce Free Lists, which are very useful in determining the
overall cultural domain of good leadership characteristics. Free listing is a common
elicitation technique in the social sciences (Weller and Romney 1988) 12 .
Researchers use free lists to identify items in a cultural domain and to calculate each
item’s relative psychological or cultural saliences (i.e., prominence, importance,
familiarities or representativeness)13. Other researchers use free lists to measure
cognitive characteristics of informants, including their knowledge of a domain and
their categorization patterns14. I will also employ pile sorting and triad methods.
1. Sample: for the purpose of generating a master list of items included in the
cultural domain of good leadership, we do not need a probability sampling. This
12 Bernard 1994 and Borgatti 1999 also provide very good overview of elicitation techniques. 13 Successive Free Listing: Using Multiple Free Listing to Generate Explanatory Models. Gery Ryan and Justin Nolan,
2000. 14 For more technical details about the use of Free Listing please see Chapter One of Stephan Borgatti’s 1998 book,
“Elicitation Techniques for Cultural Domain Analysis”, and also Greenwood 1983; Brewer 1995; Robbins and Nolan 1997.
17
is because the essence of a cultural domain is existence of common shared
knowledge amongst a group of informants who come from the same domain and
are competent about their culture. For example if we ask 10 doctors about the
main causes of pneumonia we most probably get the same sets of answers, and it
might not vary that much if we ask the same question from 100 doctors. Gery
Ryan suggests collecting 20 samples and checking for the variation of items, and
then collecting another 10 to 15 samples. If the most frequently mentioned
items did not vary more than 10%, that means sampling maturity has occurred
and we do not need to sample further. For the purpose of this study I will collect
30 to 50 samples from all strata of population (ethnic groups, age, gender,
rural/urban settlers, social class, etc.).
2. Data collection: each respondent will be given the same set of open-ended
questions such as the following examples:
I. Could you please list the characteristics of a good human being?
a. Ask probe questions to get more items on the list.
b. For each item listed ask the follow up question of how would you
verify/measure presence or absence of those characteristics.
c. If items included general terms, ask for specific definition that the
respondents have in mind.
This question will serve as the baseline or control question to detect
whether cultural domain of leadership is defined differently from that of an
average human being.
II. Could you please list the characteristics of a good leader?
a. Ask probe questions to get more items on the list.
b. For each item listed ask the follow up question of how would you
verify/measure presence or absence of those characteristics.
18
c. If items included general terms, ask for specific definition that the
respondents have in mind.
This question will help us produce a master list of the characteristics that
Afghans associate with good leadership.
III. Who are considered good leaders in our society, could you please list?
a. Ask probe questions to get more items on the list.
b. For each item listed ask the follow up question of how would you
verify/measure presence or absence of those characteristics.
c. If items included general terms, ask for specific definition that the
respondents have in mind.
IV. Where do you think good leaders come from, how they are brought up?
a. Ask probe questions to get more items on the list.
b. For each item listed ask the follow up question of how would you
verify/measure presence or absence of those characteristics.
c. If items included general terms, ask for specific definition that the
respondents have in mind.
V. How did they become good leaders, can you list the key factors?
a. Ask probe questions to get more items on the list.
b. For each item listed ask the follow up question of how would you
verify/measure presence or absence of those characteristics.
c. If items included general terms, ask for specific definition that the
respondents have in mind.
VI. What do you expect a good leader to do, could you please list?
a. Ask probe questions to get more items on the list.
b. For each item listed ask the follow up question of how would you
verify/measure presence or absence of those characteristics.
19
c. If items included general terms, ask for specific definition that the
respondents have in mind.
The last three questions intend to explore the institutions that are perceived to
be developing/generating new generation of leaders in Afghanistan. It is
important to know what institutions are given the credit for bring up new
leaders of the country.
The response from all informants will be entered into an excel sheet and will be
translated into local languages and back to English with rigorous checking for
consistency and standardization in the use of synonyms.
3. Data analysis: the data entered into excel sheets will be categorized for each
question to list specific words repeated in each respondent’s answers to
different questions. A collection of all the characteristics that different
respondents use will give us a master list, which will be used as the head row
defining all the
columns. The
names or IDs for
each respondent
are used as the
first column
defining each row.
An item by
respondent matrix
is produced by
scoring 1 in each
cell if the
corresponding
respondent has Figure 5: Example of Scree Plot from preliminary data analysis
20
mentioned the item in his/her responses, and 0 otherwise. The two-mode
matrix produced in this manner is then used to generate a scree plot15 by adding
across different columns which gives us the frequency of items mentioned by the
informants. A visible knee of the scree plot (defined by an arrow in the graph
above) will be used to decide on the number of most important characteristics to
be further studied in the second stage of research.
Alternatively we could use pile sorting and/or triad methods to group a number
of characteristics that belong to the same sub-domain.
Stage – 2:
The second phase of the study is essentially composed of a representative survey of
many different groups of informants who vary on their demographic patterns and
are expected to have varying views about characteristics of leadership in
Afghanistan. The exact survey research questions for this phase will be produced
based on the results of the free listing and pile sorting exercise in the first stage.
However, the form of survey questions in the second stage will be mostly Likert
scale, multiple choice, agree/disagree, and fill-in-the-blank questions. We need
these types of questions in order to use the cultural consensus model for the
analysis of the data16.
The analytical methods used for this study are going to be a hybrid of qualitative
and quantitative analysis. For example; Social Network Analysis software, UCINET
will be used to analyze and visualize the data, which is essentially based on
quantitative methods. Additional methods such as factor analysis, cluster analysis,
multi dimensional scaling, focus group and structured interviews might also be used
to generate and analyze the data.
15 Scree Plot is a term used by STATA software for depiction of frequency (figure 5 shows an example) 16 Please see Romney 1986 for more technical details.
21
1. Sample: for the purpose of the survey in the second stage, all respondents are
going to be stratified across ethnic groups, gender, age, social class, and
rural/urban settlements. One of the key conditions for this model to produce
robust results is that the respondents are selected such that they share cultural
knowledge, and are their cultural competency is above average (0.5)17. Most
analysis and cross comparison will be conducted at the marginal (column) level
unless funding is available to collect large number of samples to maintain
acceptable levels of significance at joint (cell) level. If average cultural
competency amongst respondents is between 0.7 and 0.8, the model produces
robust results at 95% confidence interval with only 4 samples per strata. If
increased to 27 samples 99.9% confidence interval is maintained (Romney
1986). For the test to reliably distinguish cultural competence among
informants, it is best to have about 40 questions in the questionnaire (Russell
Bernard 2006), but in this study I might need to include more questions in the
questionnaire because most of the variation comes from the number of
questions not the number of samples, which is another unique property of the
model that ensures cost effectiveness of the study.
For this research I am aiming at 4 samples per strata to allow comparisons
across cells (at 95% Confidence Interval), and across columns (at 99%
Confidence Interval). That means 64 X 4 = 256 responses are needed to examine
cultural consensus on good leadership across different groups of population in
Afghanistan18.
2. Data collection: A survey composed of about 40 to 50 questions will be
developed based on data collected in the first stage of the study. Survey
questionnaire will be translated in both Farsi and Pashtu languages and
crosschecked for clarity through back translation and reliability check using item
17 Please see Romney 1986 for more details. 18 As discussed I will have about 64 strata; rural/urban, men/women, old/new generations, elite/non-elite informants,
and four ethnic groups composing of Pashtuns, Tajiks, Hazaras, and Uzbeks. A complete list of different sample options is calculated by Romney 1986 which could be used as the basis of sample calculation.
22
response theory. The data from the survey will be entered into an Excel sheet
and checked for systematic measurement error and consistency in spelling. As I
described for stage one, the data will be turned into a two-mode matrix by
scoring each respondent’s measure of agreement or disagreement with each
question. Therefore, cells corresponding to each characteristics of leadership
and each respondent will contain figures ranging from 1 to 10 (or any other
range given the type of question).
3. Data Analysis: While two-dimensional matrices are the direct input for UCINET
software for some analysis one-dimensional matrices will also be produced.
One-dimensional matrices are, particularly used for multi dimensional scaling
and cluster analysis where variation and similarity of different groups of
Afghans over each characteristics of leadership are calculated and presented in
visual graphics19.
POLICY CONCLUSIONS
While the exact policy conclusions of the study will obviously depend on the data
that will be collected and analyzed, there are a few possible conclusions that I expect
will result from such an analytical exercise, and they will have important policy
implications for the people of Afghanistan, as well as the international community.
The most extreme scenarios that may have critical policy conclusions would be:
The data suggesting lack of consensus both across and within groups of the
population over key characteristics of good leadership. This would imply a
number of conflict management policy recommendations in the near term, and
large-scale education in the long run to education people on the extreme
diversity of views in the society.
19 For more technical details about matrices please refer to Bernard 2006.
23
The data suggesting lack of consensus across different groups of population, but
presence of consensus within each group. This would imply the need for policies
that ensure access to resources, as well as equal opportunity for all groups to
train and raise their own leaders into the country’s leadership system.
The data suggesting consensus over the type of characteristics that is not
necessarily achievable by all Afghans equally. For example, if there is consensus
that good leadership is determined by blood relationship to a specific ethnic
group or tribe. This would imply need for policies towards more public
education campaigns to raise public awareness about the challenges that such
views might entail and public education to build consensus.
The data suggest that there is national consensus over achievable characteristics
such as education, representativeness, accountability, etc. This would be the best
case scenario and would imply specific policy recommendations to different
national institutions to prepare the next generations of Afghan leaders with the
characteristics that Afghans want to see in their leaders.
The policy recommendations could also entail specific measures for revisiting the
form of governance, and structure of political institutions to accommodate diverse
expectations of different groups of the population. Or it could also imply adjustment
of international assistance, which will be discussed if the data suggested so.
24
TENTATIVE TIMELINE
27th of February 2012: dissertation proposal defense
1st of March through 1st of August: fund raising and literature review
2nd of August through end of November: first round of Data Collection
1st of October through end of October: analysis of data and questionnaire
development.
1st of December through end of March 2013: second round of data collection
1st of April 2013 through 1st of June: analysis and review of data from second
stage of study.
10th of June 2013 through 1st of August: conducting policy analysis and search
for viable and feasible policy recommendation
15th of September 2013: where I am stuck presentation
1st January 2014: preparation of first draft of dissertation
1st February 2014: review of draft dissertation by committee members
1st May 2014: incorporation of feedbacks from review.
1st April 2014: dissertation seminar and final submission.
25
REFERENCES
Afghanistan: The Problem of Pashtun Alienation (International Crisis Group
Report No. 62). Retrieved February 2, 2012 from
http://www.crisisgroup.org/en/regions/asia/south-asia/afghanistan/062-
afghanistan-the-problem-of-pashtun-alienation.aspx
Afghanistan Presidential Election Results – 2004 (Map). Retrieved Jan 5, 2012
from
http://www.iec.org.af/Public_html/Election%20Results%20Website/english/e
nglish.htm
Akbar, A.S. (1976). Millenium and Charisma among Pathans: A Critical Essay in
Social Anthropology. London: Routledge.
Barfield, T. (2010). Afghanistan: A Cultural and Political History (pp. 24–65). New
Jersey, NJ: Princeton University Press.
Barlow, C.B., and Jordan M. (2003). Character Assessment: An Examination of
Leadership Levels. Journal of Business and Psychology, 17, 563–584.
Bernard, H.R. (2006). Research Methods in Anthropology: Qualitative and
Quantitative Approaches (4th Edition) (pp. 202–209). Lanham, MD: AltaMira
Press.
Chavez, L.R., Hubbell, F.A., McMullin, J.M., Martinez, R.G., Mishra, S.I. (1995).
Structure and Meaning in Models of Breast and Cervical Cancer Risk Factors: A
Comparison of Perceptions among Latinas, Anglo Women, and Physicians.
Medical Anthropology Quarterly, New Series, 9, 40–47.
Chemers, M.M. (2000). Leadership Research and Theory A Functional
Integration. Group Dynamics: Theory and Research, 4, 27–43.
Clowney, K. (2001). New Definition of Leadership. Executive Excellence, 18, 8–10.
Gergen, D. (2001). Character of Leadership. Executive Excellence, 18, 5–7.
Josephson, M.S. (1991). Character: Linchpin of Leadership. Executive Excellence,
16, 2–4.
26
Kirkpatrick, S.A., and Loke, E.A. (1991). Leadership: Do traits matter? The
Executive, 5, 48–60.
Lickona, T. (1991). Educating for Character. New York, NY: Bantam Books.
Oberson, J. (2002). Political Alignment, Leadership and the State in Pashtun
Society. Retrieved Jan 19, 2012 from kit10.info Website:
http://www.kit10.info/Khans%20&%20Warlords%20in%20Pashtun%20Societ
y%20PART%201.pdf
Romney, A.K., Weller, S.C., Batchelder, W.H. (1986). Culture as Consensus: A
Theory of Culture and Informant Accuracy. American Anthropologist, New Series,
88, 313–338.
Shahrani, M.N. (2000). The Taliban Enigma. Person-Centered Politics & Extremism
in Afghanistan. Retrieved Feb12, 2012 from Leiden University, Repository
Website: https://openaccess.leidenuniv.nl/handle/1887/17446
Weller, S.C., Romney, A.K. (1988). Systematic Data Collection. Newbury Park, CA:
Sage Publications, Inc.
Wilson, H. (2011). Building Leadership for Transition – the challenge of
Afghanistan. Retrieved February 12, 2012, from Hamish Wilson’s Personal Blog:
http://wasafiriconsulting.com/blog/?p=207
27
DISSERTATION COMMITTEE
Terrance Kelly (Chair), Senior Operations Researcher, RAND Corporation
[email protected] ext. 4905
Signature Date
Thomas Szayna, Senior Political Scientist, RAND Corporation
[email protected] ext. 7758
Signature Date
Gery Ryan, Senior Behavioral and Social Science Researcher, RAND Corporation
[email protected] ext. 7925
Signature Date
Francis Fukuyama, Olivier Nomellini Senior Fellow, Stanford University
[email protected] Tel: 650-723-3214
Signature Date