You are on page 1of 11

1

Modeling Theories and Modeling Phenomena: A Humanist’s Initiation


Ann Taves

ABSTRACT: My introduction to modeling has gone through various stages. The first
stage, in which I worked with the modeling team, was like an initiation process in which,
as the initiate, I trusted my guides without really knowing where we were going or what
the outcome would be. During this stage, I worked with the modeling team and a class
of ten students – three doctoral students and seven undergraduates – to model Rodney
Stark’s theory of why new religious movements succeed or fail. Based on our
collaborative discussion of Stark’s ten propositions, the modeling team developed an
agent-based model of Stark’s theory in NetLogo. In addition to analyzing the theory to
help generate the model, the students and I provided the modeling team with historical
data on ten NRMs in order to test the model against what actually happened. The model
is currently at the validation stage, so we do not yet have analytical results. The second
stage involved learning the basics of NetLogo and the discovery of generative social
science, which builds models from the ground up in order to develop theories. This
allowed me to distinguish between two different goals -- modeling an extant theory in
order to test it and modeling a phenomenon in order to generate a theory. This gave me
a better understanding of what the modelers did to model Stark’s theory and a sense of
choices I can make as I move forward. I hope to draw from both experiences to model a
phenomenon central to my own research.

The decision to model Stark’s theory


I received the letter inviting me to participate in the modeling project just days
before the first meeting of my Winter Quarter course on New Religious Movements
(NRMs). Although my first impulse was to say “ask me again later,” it occurred to me
that Rodney Stark’s (1996) ten propositions, which attempt to specify the conditions for
the success or failure of a new religious movement, offered a logical theory to test
against the historical data of actual NRM growth. The organizers liked this idea and,
over the course of the quarter, I worked with a class of ten students – three doctoral
students and seven undergraduates – to model Stark’s theory and derive data from the
history of actual NRMs to test the model.

Figure 1: Stark’s Ten Propositions (Stark 1996)

1. New religious movements are likely to succeed to the extent that they retain cultural
continuity with the conventional faiths of the societies within which they seek converts.

2. New religious movements are likely to succeed to the extent that their doctrines are non-
empirical. 

3. New religious movements are likely to succeed to the extent that they maintain a medium
level of tension with their surrounding environment—are strict, but not too strict. 

4. New religious movements are likely to succeed to the extent that they have legitimate leaders
with adequate authority to be effective. (4a) Adequate authority requires clear doctrinal
justifications for an effective and legitimate leadership.
 (4b) Authority is regarded as more
legitimate and gains in effectiveness to the degree that members perceive themselves as
2

participants in the system of authority. 



5. New religious movements are likely to succeed to the extent that they can generate a highly
motivated, volunteer, religious labor force, including many willing to proselytize.
6. New religious movements are likely to succeed to the extent that they maintain a level of
fertility sufficient to at least offset member mortality.
7. Other things being equal, new religious movements will prosper to the extent that they
compete against weak, local conventional religious organizations within a relatively
unregulated religious economy.
8. New religious movements are likely to succeed to the extent that they sustain strong internal
attachments, while remaining an open social network, able to maintain and form ties to
outsiders.
9. New religious movements are likely to succeed to the extent that they continue to maintain
sufficient tension with their environment—remain sufficiently strict.
10. New religious movements must socialise the young sufficiently well as to minimize both
defection and the appeal of reduced strictness.
At the outset of the process, I was like an initiate who trusted my guides, knowing
only, as the invitation letter stated, that our goal was to “generate insights into the
chosen theory while going through the experience of developing the model,
implementing it into a simulation, and analyzing results against suitable datasets.” In
light of choosing Stark’s theory to model, I assumed that we would test the simulation
against historical data. I didn’t register the distinction between a “suitable [pre-existing]
dataset” and “raw” historical data. Nor did I realize how few models have actually been
tested against historical data. Nonetheless, the challenge I set myself and the class was to
translate the language of Stark’s theory into the language and world of actual historical
NRMs and, at the same time, to supply the modelers with whatever they said they
needed to create the model.

How we helped generate the model and the data to test it


In the few days between receiving the invitation letter and the start of the
quarter, I revised the syllabus to reflect the new agenda. I explained the course’s
research focus during the first class session, indicating that each student would be
responsible for reconstructing the history of a new religious movement and ultimately
for drawing data from that history that we could use to test the model of Stark’s theory.
About half the students found this agenda rather daunting and dropped the course. For
those who stayed, however, it proved a transformative learning experience. For most of
the undergraduates, the course provided their first direct experience of historical
research and in-depth analysis of a theory. For the graduate students, who had
considerably more experience doing research, it provided experience analyzing a theory
in depth and an entirely new method for testing it.
To guide the class, I created a chart detailing the historical and modeling tasks for
each week. I envisioned that we would pursue each set of tasks independently through
the first half of the course, then begin integrating them in the latter half by drawing from
the students’ historical reconstructions to supply data that we could use to test the
model of Stark’s theory. The modelers arranged for me to visit them in Boston to work
3

on the model prior to our 7th class.



Week Historical Tasks Modeling Tasks
1 • Introduce overall historical task of • Explain what computer modeling and
Jan12 reconstruction based on real-time simulation and the steps involved
sources • Introduce Stark’s theory
• Introduce tools – American • Instructions for reading
Religion Collection (ARC) & Cult
Awareness Network (CAN) Guide
• Instructions for initial investigation
of a group
2 • Primary sources and real-time • Discuss Stark’s 10 propositions
Jan19 reconstructions • Translating from the language of
• Decide on group to research in sociology to the language and world of
depth our sources.
• Constructing a timeline of events • What is a model? How do we make
and sources one?
3 • Evaluating primary (and • Continued work on translating Stark’s
Jan26 secondary) sources propositions
• Scientology, CAN, and the cult • Continue developing a model
controversies
4 • Working with primary and • Preliminary coding of your group’s
Feb02 secondary sources. development
• Evaluation of CAN materials • Development rationale for coding
related to your group. decisions
5 • Continued work with ARC • Refine coding of your group’s
Feb09 materials development over time
• Refine historical reconstruction • Continued refinement of rationale
and interactions over time
6 Comparison of reconstructions and proposed coding of propositions
Feb16 Standardization of criteria and rationales for decisions
Development of common model of process
7 Creation of working simulation
Feb23 Report on Boston meeting
8 Running the Simulation and Testing the Theory
Mar02 Discussion of the results with modeling team?
What could we do to make it more precise?
9-10 Case Study Presentations

Figure 2: The chart detailing the historical and modeling tasks for each week (excerpted
from the syllabus for RS101A: New Religious Movements, Winter 2017)

The roadmap for the course worked reasonably well considering how little I
knew about modeling, because I knew a lot about reconstructing the history of NRMs
and could draw on my historical knowledge to translate Stark’s propositions into the
4

language of our sources. Prior to my trip to Boston, most of our modeling related work
involved clarifying Stark’s propositions in light of our work with primary sources. In
Boston, we worked with the document the students and I had been creating and further
refined it. The modelers began work on the model while I was there, which gave me a
much better idea of what a model involved. I returned from Boston able to give the
students an overview of where we were going. This included introducing them to
NetLogo, the platform on which on model was going to run; specifying the input data we
needed for each of their groups and the steps we needed to take to get it ready; the basic
idea of how the model itself would work; and how we would test and verify it.

Input Data:
what you need to provide
✤  Group Matrix (discussed on next slide)
✤  Excel Sheet (tracks changes over time)*
✤  Were they proselytizing? (#5)
✤  Were there falsifiable events? (#2)
✤  Data on actual group size over time**
✤  Practices (indicate key changes over time on excel sheet) - #3/#9, #6/#10, #8
✤  Mandatory practices
✤  Frequency expected/mandated
✤  Harshness of sanctions for not performing
✤  “Suggested Practices”
✤  Must be central to group ID
✤  Frequency expected
✤  Authority and leadership roles (changes over time on excel sheet)
✤  Authority structure of group and key figures (#4)
✤  Personal Matrix for key figures (discussed in “Creating Agents”)



Figure 3: Slide from PowerPoint summarizing what I had learned at the Boston meeting,
presented in class, 23 February 2017

At the Boston meeting, the modelers introduced the idea of individual, group, and
environmental belief matrices. The group matrix was to identify the core beliefs that
distinguished a given NRM from all the others and the environmental matrix was to
identify the prevalent beliefs in the environment, which we agreed had to be period
specific. Individual belief matrices, I learned, would be generated by the computer
program. As instructed by the modelers, the students and I came up with the minimal
set of beliefs that allowed us to differentiate our ten NRMs. Working together and using
Heaven’s Gate as an example, we came up with the following Belief Matrix headings.
5

Group Matrix – Heaven’s Gate



Heaven Jesus Immanent Non-Biblical Self Psychic Aliens
Other World Eschatology Revelation Transformation Powers
Heaven [Heaven]
Other World
Jesus Inhabits NL [Jesus]

Immanent Immanent Exit [Eschatology]
Eschatology to NL
Non-Biblical Heaven = NL Jesus’ alien End immanent [Non-Biblical
Revelation nature Demonstration Revelation]
revealed & Exit near
Self Preparation for Preparation for True Self = [Self
Transformation NL Exit to NL alien Transformation]
Psychic Powers Mind-to-Mind NL informs Aliens Prayer (M2M) as [N-O Powers]
communication when to exit contact M2M means of
with NL transformation
Aliens Inhabit NL Jesus was Exit to NL Aliens give Overcoming = Aliens use to [Aliens]
an alien Acquire alien continuing realizing true communicate
form guidance alien self

NL = Next Level
M2M = mind to mind

Table 1: Group Matrix for Heaven’s Gate, developed in class, 23 February 2017

Having developed the matrix, we spent the next class locating their NRMs on the
matrix. This turned out to be a very interactive class session and one of their favorites.
We drew the matrix on the board so that each student could place the initials of their
group in the boxes they thought were appropriate. Then they explained why they placed
their group as they did. This precipitated a lively comparative discussion of the groups.
Discussions of non-biblical revelation and psychic powers were particularly interesting,
as students researching Jehovah’s Witnesses and the Church of the Nazarene initially
wanted to include their groups under these headings. In our discussion, we clarified
that the belief matrix had to reflect the group’s beliefs as insiders understood them.
Whatever outsiders might think, both Witnesses and Nazarenes claimed that their
beliefs were based solely on biblical revelation. We extended the same principle to
psychic powers; insiders had to formally embrace a belief in psychic powers to be
included. Based on the discussion, students researching the more Christian groups
deleted them from many of the matrix boxes.
After a long process of clarifying how Stark’s propositions translated into the
actual language of our groups and the development of a ten-point scoring scheme
running from low to high, we went through a similar comparative process to arrive at
ratings for each of the groups on each of Stark’s propositions. With some propositions,
the translation process was fairly simple. So, for example, with Proposition #8, which
referred to the strength of internal versus external attachments, we were able to
translate it easily into a continuum with low and high scores.


8. New religious movements are likely to succeed to the extent that they sustain
strong internal attachments, while remaining an open social network, able to 6
maintain and form ties to outsiders.

8. New religious movements are likely to succeed to the extent that they sustain strong
internal attachments, while remaining an open social network, able to maintain and
form ties to outsiders. Comment: Only under some conditions will agents create link
with other agents. Measure of Relational Mobility. Measured by clique density.
Low High
Associate with everyone, no sense of Association only with in-group members
group • No contact with family members
HIGH mobility, free associations • Sever ties with previous religion
• Sever ties with previous way of life.
• No outside information
Group provides all needs – social,
medical, financial
LOW mobility, no out-group links,
sanctions against out-group links

Figure 4: Operationalization of Proposition 8 (excerpted from a class document titled

“Stark’s Criteria v. 5”)

Proposition #4, on leadership, was much more difficult. Stark himself broke it down
into two sub-propositions, one having to do with the leader’s authority and the other
with the member’s sense of participating in the system of authority. Once we were able
to specify “adequate authority” in terms of teachings that justify leadership roles, we
realized that “effective authority” also involved the power to impose sanctions. This led
us to divide 4a into two aspects.
3 7


4. New religious movements are likely to succeed to the extent that they have
legitimate leaders with adequate authority to be effective.
(4a) Adequate authority requires clear doctrinal justifications for an effective and
legitimate leadership.
4.a.1. Adequate authority & justification for leadership
Low High
The group does not have Some roles are justified The group has teachings
teachings that justify their and others are not. that justify their form of
form of leadership. leadership.
Comment: Doctrines must
define the basis for leader-
ship. Who may lead, how is
leadership is obtained, &
what powers are granted to
leaders.
4a2. Adequate authority to assert it = What sanctions may leaders impose?
Leaders cannot or do not Leaders impose sanctions
impose sanctions for for violation of norms
violation of norms (=consequences for not
(=consequences for not adhering to practices)
adhering to practices)
(4b) Authority is regarded as more legitimate and gains in effectiveness to the degree
that members perceive themselves as participants in the system of authority.
Low perception of Middle, e.g. megachurch High perception of
participation in system of with lots of small participation in system of
authority, e.g., optional groups authority, e.g. Communal
passive participation groups high

5. New religious movements are likely to succeed to the extent that they can
Figure 5: Operationalization of Proposition 4 (excerpted from a class document titled
generate a highly motivated, volunteer, religious labor force, including many willing
“Stark’s Criteria, v. 5”)
to proselytize.

Actions of #5 are based on actions in #3. We expect that the number of converts
Once we had each of the propositions clarified, we were able to have a group
will vary depending on the nature of the connections that group members are
conversation in which we discussed and collectively agreed on where each group fell on
allowed to maintain with out-group members. OK
the continua and inserted them in a common spreadsheet (see Table 2). In this context,
High Low
Scientology provided a particularly interesting case as the answers differed depending
Large number of contacts with out-group Few contacts with out-group members
on the level of the members’ involvement. Thus, Scientology is specified in the table in
members with aim of getting them to join. with aim of getting them to join.
relation to three different levels of engagement: SCIPUB (public/auditors), SCI-OT

(Operating Thetan), and SCI-SEA (SeaOrg).
6. New religious movements are likely to succeed to the extent that they maintain a
level of fertility sufficient to at least offset member mortality.

General: Biological entry. Fertility is measured in terms of whether or not they have
kids. This is connected to #3 in terms of beliefs in celibacy or pre-existent entities
that need bodies, which relate to sexual practice. Quality of life of children is
8

Stark Proposition #2 [LOW] [HIGH]


No signs & wonders SCIPUB NAZ CAW SCIOT Continuous signs & wonders
Non-empirical

1 2 3 4 5 6 7 9 10
Stark Proposition #3 [LOW] NAZ 1st gen. NAZ 2nd-3rd gen [HIGH]
Strictness of Practices Practices are optional CSL BNL AS Practices are mandatory
Mandatory Rating Practices are legal SCIPUB SCIOT Practices may be illegal
Skepticism is allowed ON Skepticism is not allowed
CAW PT SCISEA JW
HG PT

1 2 3 4 5 6 7 8 9 10
Stark Proposition #4 [LOW] [HIGH]
4A1: Adequate authority No teachings that justify leadership 3HO CSL NAZ AS SCI DM JW Has teachings that justify leadership
& justification for leadership CAW ON PT
BNL HG
SCI LRH
1 2 3 4 5 6 7 8 9 10
4A2: Sanctions [LOW] [HIGH]
No Sanctions HG1 AS SCIPUB HG2 ON SCIOT Sanctions are severe
CSL NAZ BNL JW,
CAW PT
SCISEA
1 2 3 4 5 6 7 8 9 10
4B: Members perceive that [LOW] JW NAZ PT [HIGH]
they are particpants in the Only the leader has impact on 3HO ON HG Everyone has impact on decisions
system of authority decisions BNL CAW CSL and feels like they have a job
AS

1 2 3 4 5 6 7 8 9 10
Stark Proposition #6 [LOW] SCISEA, HG ON 3HO JW SCIOT PT [HIGH]
Level of fertility offsets Post-reproductive age CAW SCIPUB NAZ Reproductive age
members mortality Anti-fertility oriented beliefs BNL CSL Fertility oriented beliefs
Mortality promoting beliefs AS Health promoting beliefs

1 2 3 4 5 6 7 8 9 10
Starks Proposition #8 [LOW] AS CAW JW PT [HIGH]
Internal/External bonds and Associate with everyone, no sense BNL ON Association with only in group members
relational mobility of group. CSL SCIPUB NAZ SCIOT SCISEA No contact with family members
High mobility, free associations Sever ties with previous religion


1 2 3 4 5 6 7 8 9 10
Starks Proposition #8A [LOW] HG1 SCIPUB SCIOT AS JW Strong ties within group
In-Group Ties Weak ties with individuals inside group CAW HG2 PT

Table 2: Ten NRMs located on Low-to-High continua for Propositions 2-4, 6, & 8
CSL SCISEA
NAZ
1 2 3 4 5 6 7 8 9 10

(excerpted from a class document titled, Stark Proposition Spreadsheet)


Starks Proposition #8B
Out-Group Ties
[LOW]
Few ties outside of the group
HG
SCISEA
JW AS
SCIOT
NAZ CSL
SCIPUB
PT
Many ties outside the group
CAW

Key: AS = Aetherius, 3HO = Healthy, Happy, Holy Organization; NAZ = Church of the
Stark Proposition #9 [LOW]
1 2 3 4 5 6
SCIPUB
7 8
JW
9
[HIGH]
10

Nazarene; PT = People’s Temple; SCI = Scientology; OC = Oneida Community; CAW =


Maintain strictness of
practices
Free riders (doubters, uncommited,
non practioners) accommodated
CAW
CSL AS SCIOT NAZ ON
SCISEA
Doubt cannot be accomodated
i.e. expressed
Practices mandatory not optional

Church of all Worlds; BNL = Brotherhood of New Life; CSL = United Centers of Spiritual Sanctions for violations of norms
HG
PT

Living; JW = Jehovah’s Witnesses


Stark Proposition #10 [LOW]
1 2 3
JW
4
SCIPUB
5 6 7
SCIOT
8
PT
9
[HIGH]
10

Socialization Lose hatchlings ON CAW NAZ SCISEA Retain hatchlings

HG
AS
CSL

Although we had the input data on the NRMs ready to test in the model by the
end of the course, ten weeks wasn’t enough time to both develop the model and test it.
The class reconvened after the quarter was over to see the model in action, but it took a
while longer before the specific groups were added to the model and it was not until
some months later that we finally created an environmental matrix that approximated
beliefs in the general society over time.

Our Model and its Limitations
In developing an agent-based model, the modelers had to fill in lacunae in Stark’s
theory. The most obvious problem was that Stark’s model referred to NRMs rather than
agents. The modelers had to assume that NRMs were groups with members (agents),
that groups had specific beliefs and practices, and that Stark’s theory predicted whether
agents would leave or join groups, thus causing them to gain or lose members. Creating
an agent-based model, in other words, required translating Stark’s theory from the
passive voice (NRMs gain and lose members) to the active voice (agents chose to leave
or join groups). Giving agents “agency” meant giving them the characteristics that
Stark’s model seemed to presuppose but didn’t actually state. The modelers specified
these characteristics in light of assumptions implicit in the theory. These characteristics
included the capacity to hold beliefs, connect with others, assume roles as leaders or
followers, leave and join groups, convert others and be converted, reproduce, and
socialize offspring.
9

The idea that “religious movements are likely to succeed to the extent that they
maintain a medium level of tension with their surrounding environment – are strict, but
not too strict” -- is central to Stark’s theory. Again, this is specified passively in terms of
NRMs rather than actively in terms of agents. To create a dynamic, agent-based model,
the modelers had to create a metric to determine whether agents were inclined to leave
their current group and search for another. They specified this in terms of an agent-
based calculation of their “happiness” with their current situation. To program this, the
modelers specified Stark’s “tensions” as tension between agent and leader, tension
between the agent’s beliefs and those around them, and tension between the agent’s
actual and desired level of connection to others. Agents were programmed with several
personality traits, e.g. skepticism and credulity, that we thought particularly relevant.
In retrospect, it is obvious that Stark’s theory had two key limitations: it
presupposed the existence of NRMs and attempted to account for their growth or
decline in passive terms. Upal (2005a,b) identified these limitations and offered an
alternative socio-cognitive model to account for the emergence as well as the growth of
NRMs. When we decided on Stark’s theory, however, I was unaware of Upal’s (2005b)
computational model and, in any case, didn’t have the ability at that point to assess it. In
retrospect, I think his was more theoretically sound. Where we had to supply agent
characteristics and motives in order to model Stark’s theory, Upal began with an agent
based theory built (ironically) on Stark and Bainbridge’s (1987) entrepreneurship
theory of religion. Upal’s model presupposed that humans seek information to solve
problems, that founders of new movements are information entrepreneurs, and the
people join movements if and when they think the entrepreneur’s new ideas offer
solutions to their perceived problems.
With the creation of our model of Stark’s theory, it – like Upal’s model – is ready
to be tested against actual historical data. Given the ways that we had to flesh out
Stark’s theory in order to model it, the computer model version of Stark’s theory is now
somewhat closer to Upal’s model than Stark’s printed theory was. The belief matrices
we generated for each of the NRMs are analogous to the new information generated by
problem-solving agents in Upal’s model. Testing Upal’s model against actual NRMs
would likely be as complicated as testing the model of Stark’s theory, however. Just as
we had to operationalize Stark’s propositions in order to test the model of his theory, so
too we would need to operationalize NRMs founders’ problem-solving ability based on
historical data to test Upal’s model. This would require identifying problems in the
environment, the solutions proposed by founders, and assessing the relative strengths of
the solutions. Ultimately it might make sense to run the same groups through both
models and see how close each came to reproducing the actual history of the group.
It’s important to note that I could not have written this description of how the
modelers created the model of Stark’s theory when I returned from Boston or even at
the end of the NRM course. My ability to explain what they did rests in large part on a
later conversation with Justin Lane, which took place after I spent some time learning to
understand the code hidden behind the NetLogo user interface.

Concluding Thoughts: Modeling Theories versus Modeling Phenomena
Much as I learned from going through the collaborative process with the
modelers, I came away frustrated by my limited understanding of the actual modeling
10

process. By the end of the quarter, I had a rough idea of what a model was, but no idea
how they wrote the code. Behind the NetLogo user interface, the actual model was a
“black box.” To learn more, two of the undergraduates (Emily Stolken and Garrison
Freeman) and I spent time during Spring Quarter learning the basics of NetLogo. Here
again we started with a theory we wanted to model – in this case a theory of how people
“cope,” that is, how they appraise situations and decide how to act. But in contrast to the
course, we didn’t try to model human coping, but looked into how other animals cope
and tried to imagine the simplest form of coping that we could. Because we had no
coding experience, we wanted to begin by creating the simplest possible version of the
process.
The challenge again was one of translation. As humanists, we had to figure out
how the language we were using – that of animals appraising situations and deciding
how to act – could be translated into a modeling language, such as NetLogo. As is the
case with any new language, learning the basics wasn’t the same as becoming fluent, but
the basics gave us an idea of what is possible. Thus, in NetLogo language we learned:
turtles are agents, turtles exist in worlds, situations are programed to occur, and turtles
are programmed to appraise them.
Although watching our first primitive program run was extremely satisfying, the
most significant thing I learned was that the approach we took– starting with the
simplest possible version of the process we wanted to model – is how generative social
science models are being created. Reading Epstein (1996, 2006, 2013) on generative
social science as we learned the basics of NetLogo (Wilensky & Rand, 2015), I learned
that generative modelers begin with a phenomenon they want to model and derive a
theory from their model, rather than modeling a full-blown theory they want to test.
Asking humanists to model a theory rather than modeling a phenomenon had
certain advantages: we began with something familiar (a theory and historical data), we
were forced to analyze the theory in great detail and figure out how to operationalize
historical data in the terms of the theory, and, if we had had more time, we could have
tested the theory in light of the simulation. In this collaborative approach, the
humanists do not actually construct the model; the modeling and simulation experts do
that. While this undoubtedly made the process less daunting, I couldn’t understand how
they translated the pencil and paper version of the theory into a computer model of the
theory until I tried to do a little modeling myself.
Moreover, inviting humanists to model a theory as if that were the only option
obscured an important alternative: modeling a phenomenon and deriving a theory from
a model that generated it. This approach requires humanists to learn something about
modeling – it requires us to look into the black box – but doing so introduced us to an
important development – the emergence of a generative social science with its
commitment to building phenomena from the ground up. In the end, neither approach
resulted in a finished product but devoting a quarter to each -- on a very part time basis
– did give me insights into the strengths and weaknesses of each and whetted my
appetite for more.

Thanks to my modeling team – LeRon Shults, Justin Lane and John Balch – and thanks to
the students who took up the modeling challenge: Lakshmi Alban (Aetherius), Shandeep
Ahdi (3HO), Julie Best (Church of the Nazarene), Chris Edge (People’s Temple), Jenn
11

Farinet (Scientology), Garrison Freeman (Oneida Community), Damian Lanahan-Kalish


(Church of All Worlds), Dell Rose (Brotherhood of New Life), Emily Stolken (United
Centers of Spiritual Living), and Jill Suarez (Jehovah’s Witnesses).

References

Epstein, Joshua M. 2006. Generative social science: Studies in agent-based computational
modeling. Princeton, NJ: Princeton University Press.

Epstein, Joshua M. 2013. Agent Zero: Toward neurocognitive foundations for generative
social science. Princeton, NJ: Princeton University Press.

Epstein, Joshua M. and Axtell, Robert. 1996. Growing artificial societies: Social science
from the bottom up. Cambridge, MA: MIT Press.

Stark, Rodney. 1996. Why religious movements succeed or fail: a revised general model.
Journal of Contemporary Religion 11, 133-146.

Upal, M. Afzal. 2005a. Towards a cognitive science of new religious movements. Journal
of Cognition and Culture 5(1), 214-239.

Upal, M. Afzal. 2005b. Simulating the emergence of new religious movements. Journal of
Artificial Societies and Social Simulation 8(1), 1-17.

Wilensky, Uri, and Rand, William. 2015. An introduction to agent-based modeling:
Modeling natural, social, and engineered complex systems with NetLogo. Cambridge,
MA: MIT Press.

You might also like