OVERVIEW READINGS

GUIDE 1:  INTRODUCTION GUIDE 2: METHODS FOR STUDYING GROUPS GUIDE 3: GROUP STRUCTURE GUIDE 4: ASPECTS OF GROUP STRUCTURE II GUIDE 5: ATTRACTION TO GROUPS GUIDE 6: COHESIVENESS II GUIDE 7: INFLUENCE PROCESSES GUIDE 8: PERFORMANCE & DECISION-MAKING GUIDE 9: LEADERSHIP GUIDE 10: GROUP COOPERATION & CONFLICT

COURSE PROJECT PRESENTATION

The "good stuff" is coming soon: cohesion, Groupthink, leadership, cooperation...But first, the basics: study methods and group structure. These help expand your professional vocabulary and make you a more critical reader and thinker.

Many methods are available to study groups. Each has advantages and disadvantages. Which you choose depends on what you want to know. And the wise scholar will select several to study a particular group.

What all "true" experiments have in common is the random assignment of subjects, or, in our case, presumably, groups, to different study treatments. Some are considered "real experiments" and others "quasi experiments". A lot depends on how long the group has been interacting, why and how it was formed.

Experiments may take place in the laboratory or in the "field." Field experiments involving groups often take place in the "natural setting" of the group. The late, great Alvin Zander, for example, conducted many field experiments using corporations, charitable groups such as the United Way, and basketball teams. The late, great Kurt Lewin, often dubbed "the father" of group dynamics research, also ran many field experiments. Unlike Zander, who typically used intact groups, Lewin often created groups (married women who bought groceries or schoolboys) explicitly for experimental purposes. After all, most organizations want to complete their own tasks, not assist researchers, so all I can say is that both Lewin and Zander must have had charismatic persuasion down to a fine art! Their studies help the generalization of results, or one aspect of external validity.

Lab experiments may use groups that actually interact face-to-face. The studies on polarized decision-making typically have done so. Confederates may or may not appear within interacting experimental groups. For example, in Stanley Schacter's research on "group deviants," the "deviant" was typically a confederate so that the effects of continued nonconformity over time or of "sliding" toward conformity could be assessed in a standardized manner. The "learner" in the Milgram studies was also a confederate. Sometimes "group members" only interact via computer terminal (see below). It's cheaper (you don't have to hire confederates) and more standardized (you can use the same literal script over and over), and can be informative. In the latter cases, critics worry about "mundane reality," that is,how much the situation resembles the "real world."

Occasionally lab experiments use intact and pre-existing groups...but very rarely. It's true pre-existing groups add more complications but their use also adds important dimensions to the study of groups.

Laboratory experiments also allow "contrived groups." In these groups. no "real" interaction occurs and all group members save the naive subject are fictitious. For example, the subject may believe that s/he is interacting with group members online, or the subject listens to a standardized tape recording of a group s/he believe s/he will shortly join. Subjects receive bogus feedback from "group members" in a variety of guises. That's why I noted in Guide One that interacting with groups via computer is not a new situation. Scholars have used this teechnique to standardize conditions in experiments for decades.

Contrived groups (in which all members except the naive participant are fictitious) differ from ad hoc groups. In ad hoc groups, members are "real" (although at least one member might be a confederate) and may engage in face-to-face interaction. However, ad hoc members are typically strangers, are present solely to participate in an artificially created "group," and only attend because it is a course or other requirement. Under these circumstances, although we may gain information about how different social forces influence the group or individual, the group dynamics may differ from "real world" groups which typically have a purpose, members who at least partially agree with group goals etc. Ad hoc groups may bear certain similarities with mandatory groups, such as mandated counselling for offenders--alcohol, domestic violence, school truents--or prisoners or soldiers, i.e., "captive audiences".

Both contrived and ad hoc groups differ from "minimal groups". These are called minimal because only the barest interaction occurs and only the most minimal information about other members is provided. Minimal groups are sometimes used in studies of social identities (my favorites are the ones on pain tolerance.) You may be told other group members are similar to you (e.g., all new graduate students) or different from you (e.g., students at all class levels.)

Because of random assignment, the typical experiment can make stronger statements than most other types of research about causal relationships between independent and dependent variables (internal validity). Lab experiments, in particular, are high on causal control. For example, experiments on "brainstorming" indicate that it is not a particularly effective way to generate new ideas (over and above the total number contributed by independent individuals) although it has become very popular in many settings.

On the other hand, in laboratory experiments:

The setting may be so artificial that external validity is low, i.e., it is unclear how, if at all, the experimental results may be generalized or which groups the results may be generalized to. Many simulation studies suffer from this problem. The Milgram study and the Haney, Banks and Zimbardo "prison"  studies also have problems with external validity because the circumstances were so unusual.

However, there is such a thing as "experimental reality". Zimbardo's participants (the "prisoners", anyway) were so traumatized, the study was ended after three days (it had been planned for two weeks). Milgram's participants were in agony over shocking a "learner". I've seen social psychology students practically murder each other to win a prisoner's dilemma game class  "kitty" of 40 cents. Even relatively trivial experiments can be engrossing.

When "group members" interact, most laboratory "groups" have members who are strangers to one another. This is a very critical problem. As we have since learned from helping behavior research, "groups" of strangers behave very differently than groups of acquaintances or friends.

We don't know whether the effects of "contrived groups" or "minimal groups" on the individual are the same as the effects of real and intact groups, even ad hoc groups. Studies tend to use either contrived groups OR created ad hoc groups OR intact groups, but rarely even two out of these three. As a result, it is difficult to make comparisons.

Most laboratory experimental groups meet for a limited time. Group development is truncated. There may be no time for "storming" or "norming" and the experimenter decides when the groups adjourn. Under these circumstances, I would be extremely cautious about making statements about stages of development in groups, or even generalizations about "emergent leadership." For example, Freed Bales (who followed his groups for a semester or longer), found that in early group sessions, the most talkative person tended to be considered the leader--but this often did not hold as the group continued to meet.

The big problem in field experiments is distance from the laboratory, usually accompanied by loss of causal control.

In lab experiments, one often can "hold everything else equal" except the independent variable(s). In field studies, "ceterus paribus" is frequently impossible. Stage of group development may be confounded with the manipulated independent variable or some aspect of group composition may inadvertently be varied at the same time.

Field and lab experiments are both subject to reactivity (sometimes called the "Hawthorne effect"), i.e., people change their behavior because they know they are being studied. Group members may feel more obliged to honor idealized group norms. Other social desirability processes may occur. Members may self-monitor their own behavior more closely. Thus, the act of studying the group may actually cause changes in the group.

It's not always possible to manipulate the independent variables (e.g., group size or the number of members endorsing a group course of action) and it's not always possible to randomly assign participants to different treatments. When different conditions are assigned on a non-random basis. the research design might be what is often called quasi-experimental. Some methodologists even claim that it is impossible to do a "true experiment" using intact groups (e.g., classrooms, work teams) and that a quasi experiment is the best controlled design that is possible under such circumstances.
 

Observational research on groups typically occurs in the field. Thus observational research is highly contextual and may take many forms. In virtually all instances, however, the research is not experimental (although again, see Alvin Zander's scholarship). Causal evidence may be correlational rather than experimental; thus causal statements are weaker in observational studies than they are for experiments. Or evidence may lie in ethnographic detail or even logical assertions.

In participant observation research, the researcher participates in the daily life of the group. S/he usually keeps a journal and scribbles copious field notes. The researcher may also interview group members (interviews are usually less structured and more in-depth) and use structured coding instruments. Group literature, group symbols, and group history are often included in the data, lending the research ethnographic authenticity. Participant observation research often starts with working hypotheses, rather than structured deductive hypotheses because very often the researcher doesn't know what s/he will find. Or what is found is very different from what was expected and working hypotheses are revised.

When I began studying local churches and synagogues, I was interested in how pastors and congregants constructed epistomologies about reality. I expected that some congregations would emphasize faith (yep) and be very hostile toward science (nope.) I discovered that science was conspicuous by its absence--but wow! was there ever a mountain of material on group cohesion, a lot of it unexpected to me. Since issues in group cohesion were apparent at the very beginning of the study, I was able to incorporate measures very early--but it wasn't where I started at all.

This isn't very unusual in more qualitative research.

Group observers are not always participants. It may be clear that the observer is a visitor (as it is in studies of hospitals or medical schools). The observer may be relatively aloof from the group, tape-recording, taking notes, or even sitting behind the glass of a one-way mirror.

Methods of coding group behavior and member interaction may be highly structured. For example, R. Freed Bales' Interaction Process Analysis directs the observer to watch for highly specific behaviors. SYMLOG (also Bales) is another example of using precoded categories in group observation.

Observational research is high on mundane reality, i.e., the everyday realism of the setting. Groups are often studied for several months or even years, so that group development can be observed. For better or for worse, the groups studied are typically intact and are not ad hoc.

Here are some problems with field research studies:

The case base is often small; sometimes only one case study or one location is involved.

Groups are often selected using judgmental methods on who is "typical." The selection of the group could easily influence the findings and we often don't know just how "typical" the participants are. Groups that are intriguing or "interesting" may be selected--and be typical of nothing.

Many groups refuse access. It isn't clear how groups who agree to be studied differ in other ways from those who refuse. Permeability may vary by type of institution: for example, religious congregations welcome visitors, even if their goal is research, not spirituality, while most medical schools or hospitals refuse many scholars in-depth access.

All of these issues influence external validity, or how and to whom findings may be generalized.

In participant observation research especially, the researcher may "go native" and lose objectivity, developing so much empathy with the group under study that the data become suspect.

Observational studies are also subject to reactivity. If investigators are covert, or "under cover," results may be more valid but other ethical problems arise. Most researchers are very careful to protect the anonymity of the group and individuals within it and to respect group dignity. However, "under cover" researchers are often asked to assume larger roles in the group; if they accept, they can easily influence the very phenomena they wish to study. This is especially a danger in smaller groups which typically need volunteers and active members.
 

In survey research, the researcher asks individuals questions and records the answers.

Individuals may respond as either individuals or as group representatives. Because the unit of study is the individual, survey research may seem a strange choice to study groups. However, consider the following:

Individuals, acting as group representatives, may give considerable information about group customs and history. A survey may be the only way--or the fastest way--to gather such information. A principal may know the history of the school or of teachers within it.

Ethnographies often use in-depth interviews with "key informants." Key informants are often selected because they occupy a key role in the group (e.g., principal, CEO) or because sociometric diagrams indicate that they are important to group functioning.

Individuals give relatively direct (although not always accurate) information reflecting their feelings toward the group. It may be very difficult to discover this kind of information through direct observation because people often hide their "true feelings" (or try to) toward individuals during interactions. Sociograms, in which members express their attraction to or esteem for each other, are important in studying group structure.

The degree of similarity or diversity of opinions about a topic across group members may indicate group cohesion. Indeed, Ken Wald and his colleagues at University of Florida used the coefficient of variation on a series of political attitude questions as a measure of cohesion in different religious congregations.

One can compare the survey statements of different group members to see whether they have comparable perceptions of group goals or how members perform tasks. For example, a very typical finding in the area of family research is that husbands and wives in the same family may report very different divisions of labor on household tasks.

Surveys of individuals can assess the impact of groups on the lives of individuals. For example, one can study whether those who belong to larger numbers of professional, hobby, or neighborhood groups in fact are more likely to vote in elections (Robert Putnam in Bowling Alone). Or, we can study whether those more enmeshed in group life suffer less from depression.

Survey research results have lower internal validity, or causal certainty, than experiments do.  In addition, when applied to groups, be very careful not to confuse individual perceptions or attitudes (separately or aggregated) with group reality. An individual, even as a group representative, can only reflect their own reality. And, of course, a respondent may dissemble.

Surveys of the general public are not surveys of groups! They cannot be used in place of surveys of groups.

Nevertheless, we can see that surveys, whether unstructured interviews with group informants, or more standardized polls of group members, can be useful to study and evaluate groups.

Researchers such as Irving Janis, who created the concept "Groupthink," (and Phillip Tetlock who has extensively studied Groupthink) have used various archival and historical data sources to study group phenomena. Archives may include the journals of group members that are later published or appear in collections; articles written by or about the group; census or demographic information available about a particular group; newspaper articles, videotapes, or other media; income and expenditures; group symbols; and other paraphenalia relating to the group studied.

Archival and historical data may provide the only way to study some groups, particularly if the group is secluded or no longer exists.

Obviously these kinds of data can allow the scholar far more latitude in interpretation than other, more standardized forms of data collection. S/he must often rely on other people's interpretations of the group and may never be able to check what "the facts" are by group observations or surveys of the membership. If the group wants to preserve its "place in history" or has an unflattering history, it may try to hide these flaws.

Time continuity may be excellent, again allowing study of group development.

Biases about race, gender, religion, social class, or other factors among observers or achivists and in archival materials may be present. These individuals may believe that African slaves could not have had a nuclear family system or that women's consciousness raising groups had to be "gab fests." Again, there may be no way to correct any misperceptions or stereotypes, although oral histories conducted among group members may provide counter-balance.

Historical and achival data may, of course, be used to supplement other methods, enhancing the ethnography and providing group history.

In simulation studies, the investigator creates a hypothetical world or situation. Participants play the roles of group members. A major purpose of simulations is to study group processes in a more "pure" or abstract sense. Scholars often assert that by isolating and studying key elements of groups, the results, such as how groups make decisions or maintain productivity, can be generalized to several settings.

For example, the group task may be to unanimously decide on the degree of risk the group would allow an individual to take before undergoing surgery or changing jobs. Group members may role play heads of state, students and teacher, or spouses in a "dysfunctional family."

One famous example is the classic Haney, Banks, and Zimbardo "prison study" in which male Stanford undergraduates were randomly assigned to play the role of prisoner or guard in a simulated prison built in the basement of the Psychology building.

CLICK HERE for more information on the Haney, Banks and Zimbardo study.

As the Haney et al. study illustrates:

• Simulation studies may be experimental, thus have relatively high internal validity.
• Participants may become extremely engrossed in the study setting, even if it is artificial, that is, psychological realism may be very high. I have seen high psychological realism even during small in-class simulated group sessions.

• The hypothetical quality of the situation, or the degree to which it is contrived, may make generalizations difficult.
• Subjects are typically college undergraduates, also making generalizations to other individuals or groups difficult.
• Subjects in the study may have no experience in the study setting, or even in similar settings. Relatively few undergraduates are married parents; even fewer are city commissioners or heads of state! (Or prison guards.) Thus, the subjects may have virtually no relevant behavioral repertoires to draw upon. As a result, their behavior may reflect societal stereotypes or media portrayals of the roles in question. Families in television "sitcoms" bear little resemblance to "real" families.
• The time for the study, as in many lab studies, is limited. The group may have little chance to develop and, once again, is typically composed of initial strangers.

Focus groups are a more qualitative method that are often invoked to study individual attitudes in a group setting. The idea is to see how people holding diverse opinions about the same topic express these opinions and, at least sometimes, influence these opinions. Because group discussions and informal group leaders (for example, in neighborhoods, work groups, or religious congregations) often influence social change and public opinion, this method has regained popularity. Focus groups have been used for at least 50 years.

Focus groups comprise 6 to 12 members who meet for an hour or two (some meet far longer) to discuss a specific topic. The topic may include a plan of action or may be a "brainstorming" session, in which ideas are generated and ranked in a non-judgmental atmosphere. Members usually are interested in the focus group interview topic but ideally represent a range of opinions on the topic and often some diversity on other characteristics. A set of about a dozen unstructured questions is used to stimulate group discussion. A moderator influences the flow of the discussion and attempts to bring closure to it. A team of students from this class in 2016 created focus groups of USA and International students to see the type of interaction that occurred in college between USA and International students.

The purposes are usually to gain  insight, gathering a range of dimensions that MAY be used in future, more structured research, and to give more depth to opinion surveys.

We often use focus groups:

• for exploratory work
• to assess the many dimensions of a topic
• for in-depth knowledge
• and to study how opinions form and change via group discussion.

We VIRTUALLY NEVER use focus groups to generalize to [sub]populations because typically respondents do not come from a probability sample and studying the group is an unstandardized, less structured process.

Focus groups are usually audiotaped and an assistant moderator takes copious notes. The analysis and writeup are generally qualitative, relying on quotes and summarized assessments of typicality.

Focus groups are often used in applied settings and for program evaluation.

However, focus groups if observed carefully, offer insight about how groups influence individual opinion, how groups of initially disparate individuals may come to a unified conclusion (e.g., juries or admissions committees), and dynamics of persuasion in a group setting.

For a Focus Group site with methodological tips, click HERE.

OVERVIEW

READINGS

This page was built with Netscape Composer.
Susan Carol Losh January 2 2018

Click on the picture to have some fun with the Claritas/Nielsen link. Enter your zipcode under "Zipcode Lookup" and see what you find!