Study of Complexity Profiles

From NECSIWiki

Study of Complexity Profiles of Individuals in Research Groups

by Joao Castro, Tina Champagne, Nassim Abdi Dezfooli, Heike Sichtig, John Voiklis

New England Complex Systems Institute: One-Week Intensive Course MIT, Cambridge, MA January 9-13, 2006

Disclaimer

The research reported here was developed, executed, and analyzed within a four-day period during the NECSI Intensive Course at MIT in 2006. Analysis methods were meant to explore complex systems and to make preliminary conclusions. Further research is needed to validate both the instruments, methods,and our results.

- Small Group Dynamics Team

Complexity Theory

Complexity theory emerged from a variety of approaches in the physical sciences, such as cybernetics, general systems theory, and catastrophe theory (Ashby, 1966; Bertalanffy, 1968; Thom, 1975).

The theory became more widely known after the publication of Gleick’s (1987) and Waldrop’s (1993) descriptions of the social dynamics among the scientists who developed respectively, chaos and complexity theory.

Complexity Methods

Complex Systems Science provides a framework for studying how relationships (behaviors and interactions) among the parts of any given complex system culminate in the behaviors of the system as a whole (Bar-Yam, 1997, 2004).

Research Topic

The exploration of complexity in group pattern formation using complexity research methods.

Application of Complexity Methods

“Organization is an emergent property of the replication of a discourse (i.e., communication), complexity arises from and is enabled by that replication” (Price, 2004).

Although each individual’s perception about their research group’s pattern formation is very “subjective” (Freeman,1995); complexity methods afford the ability to research the complex nature of human systems (group pattern formation) without delving into the nonlinear dynamics of the emergent patterns of behavior (Eoyang, 2004).

Complexity research methods were employed to track and visually demonstrate the degrees of divergence and convergence of participant’s perceptions of the group formation process.

Hence, this study explored the complexity profiles of group pattern formation across diverse research project groups.

Participants

In complex systems research: Participants are composites (parts) of the system (the class/group)

Participants: 40 students attending the NECSI 5-Day Intensive Course (Jan 2006) The class of forty participants subdivided into nine different work groups in order to complete nine different research projects (self-determined topics) over a 5-day timeframe as part of the NECSI intensive one-week course on complexity research methods.

Using a survey, each individual’s perceptions about their research group’s pattern formation process was gathered on days 2, 4, & 5 of the course.

SURVEY All information on this survey will be kept confidential. Please answer individually.

Group Topic:     _____________________________
Participant ID:  _____________________________
Please circle your most appropriate response 

Scale: 1 2 3 4 5 (1 - weakest, 5 -strongest)

1.  How interested were you in your group's topic before any discussion?
2.  To what degree did your topic change after last night’s work session? 
3.  How interested are you in the group’s topic after last night’s work session?                     
4.  Did you have any previous knowledge of your group’s topic?
5.  How much has your group project increased your understanding of the topic until now?
6.  To what degree do you think your group is committed to the group project?
7.  Rate the amount of responsibility that you have in your group project?
8.  Do you feel your group was receptive to your ideas/suggestions?
9.  How would you rate your group’s current performance?
10. How much influence does the current leader(s) have on the group's project? 

Data Analysis

Clustering Identify non-gaussian distributed groups in our study

Dendrogram Binary tree with a distinguished root Has all the data items at its leaves Tree clustering method uses correlation to measure the distances between objects when forming our clusters Click here to view a simple construction Clustering Results

Ordering the Results

We can then illustrate the ordering process Each line represents a participant and each column his answer to a particular question. The ordered image should have little variations of color when traversing it from the top to bottom. Two consecutive rows show similar answers by two different participants.

>> Add images here

Correlation Coefficient MATLAB Definitions Linkage = computes a hierarchical cluster tree using the algorithm specified by correlation Pdist = computes the distance between objects a data matrix using correlation Correlation coefficient compares the distance information in the cluster, generated by linkage, and the distance information in the data matrix, generated by pdist. Magnitude of this value should be very close to 1 for a high-quality solution.

       C (Tuesday)          C (Thursday)         C (Friday) 
       ----------------------------------------------------
       0.6216        ->        0.7070         ->     0.7340

>> add images here

Group Span

By identifying the positions of the members of each group we can then observe The group span of opinions inside the group relative to the entire class span of answers

Group Span Movements

>> Add images here

Movements Within a Group

We can also track movements of participants within a group. We have several surveys through time to track changes of opinion Members can shift their positioning inside a group radically.

>> Add images here

Cohesiveness

Through time we can also observe that most groups start to exhibit a more cohesive answers relative to the class scope. This is shown in the graph by a smaller group scope and even elimination of elements from other group. This might be an observation of the acquisition of “company culture” or values.

>> Add images here

Movements Within a Group

>> Add images here

Future Directions

Extend study across disciplines to explore universality:

--> Biology
--> Engineering
--> Psychology
--> Education
--> Physics

Implement “on the fly” data collection

--> PDA technology for dynamic data entry

References

Ashby, W. R. (1966). An introduction to cybernetics. New York: John Wiley.
Bar Yam, Y. (2004). Making Things Work. Boston: Knowledge Press.
Bar Yam, Y. (1997). Dynamics of Complex Systems. Boston: Addison-Wesly. 
Bertalanffy, L. v. (1968). General system theory : foundations, development, applications.  New York: G. Graziller.
Eoyang, G. (2004, Fall). The practitioner’s landscape E:CO Special Double Issue, 6 (1-2), 55-60.
Freeman, W. J. (1995). Societies of Brains. New Jersey: Lawrence Erlbaum.
Gleick, J. (1987). Chaos: Making a New Science. New York: Penguin.
Price, I. (2004, Fall). Complexity, complicatedness and complexity; A new science behind organizational intervention? E:CO Special Double Issue, 6 (1-2), 40-48.
StatSoft, Inc., (1984-2004). Retrieved January 12, 2006 from http://www.statsoft.com/textbook/stcluan.html
Thom, R. (1975). Structural stability and morphogenesis. New York: Benjamin-AddisonWesley.
Waldrop, M. M. (1993). Complexity: The emerging science at the edge of order and  chaos. New York: Simon and Schuster.

Thank You

http://www.necsi.org/community/wiki/ - Small Group Dynamics -

For comments & critique, please email:

Joao Castro: joaoc@mit.edu
Tina Champagne: tina@ot-innovations.com
Nassim Abdi Dezfooli: dezfooli@usmd.edu
Heike Sichtig: hsichtig@binghamton.edu
John Voiklis: jv37@columbia.edu