This chapter discusses the effect of social categorization process within group from the perspective of social identity, which is the product of social categorization in my opinion.
  One of the most important functions of social categorization through social identity (depersonalization) is that members would like to conform to certain norms in one group, which provides theoretical basement for regulating individuals’ behaviors in communities. However, I am interested in the development of stereotype. It seems that authors think that stereotype behaviors exist in the beginning when a group develops. But how about the case that some stereotype is generated in the process of development?  For instance, my classmates in college were from different provinces in China and we didn’t share much similar behaviors in the beginning, however, when we live together as a group, we develop some similar behaviors/ attitudes. From this perspective, stressing the concept of social category (in this case, we are classmates) time by time, is helpful for developing identical behaviors.
   They also articulate the role of marginal members and leaders. Reflecting these two points on the online community design, I would like to ask, how do we treat marginal users? They do affect the conformity in one group, but they also bring new things for a group, which is good for communities from the dynamic sight.
    The effect of Subgroups is also discussed in this chapter. One problem is the confliction between subgroup and larger group which also exist in online communities. In my opinion, individuals’ interaction in subgroup is more frequent than in larger group, which makes me suspect the likelihood of study it based on bond theory instead of social identity theory.  

This chapter provides a really comprehensive literature reviews in social Psychology relevant to people's behaviors in groups/communities. I would say there are really a lot of things.
However, I feel like I had hard time to understand some of the rationales of the claims and concepts.
First and the most, I am not sure what is the exact difference between social-categorization and social-identity. Because literately for me, the difference would be that in terms of social-categorization, one/or a group of people can not only identify the self but also the others; while social-identity is more likely to deal with perceiving of the self. However, the literatures seem to differentiate these two concepts as "adopting social-identity perspective" to understand social-categorization process. It is really hard for me, because in my view, categorizing is the same as identifying: using attributes in different dimensions to describe things (either people or groups), in order to reduce information uncertainty.
second, I could not understand the saying "the process of categorizing people exaggerates perceived similarities among people in the same group and difference between people in different groups." I suspect that the difference can be already there, rather than being intentionally exaggerated. Just using the statistic tool to test whether two groups are identical or not, we just see the ratio of in-group difference versus the between-group difference in a particular dimension, if the ratio is beyond some benchmark, then we consider they are different. So it can be that when one is identifying self whether belonging to a group, she might subjectively (or some other ways) to evaluate this ratio and categorize herself either inside or outside. So that is my thought, I could not imagine how people exaggerate the similarities and differences. People when identifying/categorizing, always in particular context, thus the importance of some dimensions can be exaggerated actually (not real...but maybe like fish eye view: ) and the decisions will largely based on those dimensions.

This article talks about "Social Categorization, Depersonalization, and Group Behavior" from a social psychology perspective. We are supposed to use the theories in the article to online community design. Then what are the theories we can learn from this article? I'd like to list a few here.

• Because of "self-categorization" and "depersonalization", people usually conform to the relevant group prototype.
• Polarization is conformity to a polarized ingroup prototype or norm.
• Attitude-behavior relationship is stronger when people self-categorize in terms of a salient group membership for which the attitude is normative/prototypical.
• Group cohesion or solidarity is not oly attraction among group members, but also attitudinal and behavioral consensus, ethnocentrism, ingroup favoritism and intergroup differentiateion, and so forth.
• Prototypical leasders do not need to exercise power in order to have influence; they are influential by virtue of their postion and the depersonalization process that assimilates members' behavior to the prototype.

This has a very large literature review and is very useful. It is also relevant to my recent turn towards the idea of "differencing," i.e. that act of making distinctions that creates categories and remainders.

"Social categorization" claims that groups are more than simply aggregates of member bonds (p65).

Of key relevance to the immanent plurality of panarchy:
"Social identity theory, and more general social categorization perspectives, make predictions about the nature of relations between subgroups as a function of the nature of their relationship to the superordinate group" (p72).

Some thoughts:
It always bugs me when i see statements like "people have a tendency to...."
where what is meant is that "people STUDIED have a tendency to...." Generalizing from a sample to a population is fine as long as we admit that the only people studied have lived within a relatively recent historical period. Those individuals' attitudes, beliefs, and behaviors should not be generalized across long historical epochs. We just cannot sample from the middle ages or ancient times.

Statements like "the collective self, or rather collective selves" are replete with atomistic Descarte-ian individualism, which is not surprising given the author's rejection of "relational self" theorizing at the beginning.

Furthermore, this paper makes an interesting contrast with other definitions/approaches:

--Markus and Kitayama 1991 defines the "interdependent self" as bonds with others
--Brewer & Gardner 1996 define the "relational self" as deriving from identification with a particular social category or group (Brewer & Gardner, 1996. "Who is this 'we'? Levels of collective identity and self-presentation." Journal of Personality and Social Psychology, 71:83-93.)

--------------------------------------------------------
PHartzog@umich.edu
--------------------------------------------------------
The Universe is made up of stories, not atoms.
--Muriel Rukeyser