A TEMPLATE TO FOLLOW WHEN DOING A GROUP DESIGN EXPERIMENT

1. Describing and Measuring the Problem (the dependent variable)

>Identify a behavior, attitude, thought/belief, condition, or situation that at least 12 clients of yours would like to change. Perhaps 12 female teen moms would like to learn about ways to physically and intellectually stimulate their babies and toddlers, and how to provide support to one another.

Describe the problem and/or needs that these clients share and have targeted for change. Define the problem/need clearly and describe how you will measure fluctuations (up and down movement) in the problem. Will you be counting the frequency of the problem? Intensity? Duration? What is/are your data source(s)-- will you rely on clients' self-report? Will you or some other reliable person be doing observations? If so, what will you observe? Will you use an observation or coding sheet? When will observations take place?

2. The Clients/Subjects

Briefly describe who your clients/consumers/subjects are, including: The age range and average age; the gender breakdown; their educational levels; racial makeup; religious makeup; socioeconomic levels.

Rationale for the Intervention(s) you have chosen

Provide evidence from a journal article, book, or direct experience that you have that shows that the intervention has been effective in situations or with persons having similar difficulties.

Intervals in which you'll measure the dependent variable

Pre-intervention phase: When and how will you measure the clients on the dependent variable?

Post-intervention phase: When and how will you measure the clients on the dependent variable?

3. Describing the Intervention

Explain what intervention each client will receive. Indicate the number of sessions that will be provided; the length of each session; the time between each session; and the entire duration of the intervention(s).

Describe the format of each session, i.e., sessions can be one-on-one, group, couple, a group of couples. The formats can vary, sometimes individual, sometimes group.

Include written material that explains the content being presented to each client. For example, if you are working with male batterers and you plan to teach them how to use problem solving instead of manipulation and domination strategies, you would include a sheet that outlines problem-solving and explains how each client will be taught this strategy.

4. Data Analysis

Most people who don't work regularly with statistics can feel intimidated by them. Don't hesitate to ask for help from your research instructor and/or consult a beginning text on statistics, e.g. R.W. Weinbach ; R. M. Grinnell's Statistics for Social Workers or D.M. Pilcher's Data Analysis for the Helping Professions. You may need to do a t-test, within-subject design or possibly a chi-square.

5. Your Findings, Conclusions and Implications

Was the intervention shown to have a positive effect beyond what would be expected by chance? This means that the critical value you obtained is greater than what would be expected in the probability tables.

If not, why do you suppose your intervention was less effective than expected?

If yes, what can you do to advocate for this intervention among co-workers and in the social work field?

Refresher Notes on Using a Group Design to Evaluate the Effectiveness of one or more Interventions on several Clients

There are many types of group designs or experiments. The basic idea behind a group experiment is to measure clients who have similar problems or goals before the worker provides an intervention and then measure these same clients after the intervention to determine whether their situation, behavior, attitude and/or feelings (whatever has been targeted for change) have improved. The researcher uses some test of statistical significance to see whether the amount of change is more than what would be expected by chance.

Group designs are supposed to be able to control for extraneous factors that could bring about change, so that if the clients do change, the worker/researcher can have confidence that it was exposure to the intervention and not some other factor(s) that brought about the desired change. The clients are measured on at least one dependent variable (the targeted behavior, feeling, condition, and usually the clients are measured twice: once pre-intervention and again post-intervention.

First, the researcher needs to have at least 12 clients with a similar condition or behavior that is being targeted for change. You can measure the frequency, intensity, duration or a behavior; or you can measure attitudes or beliefs/thoughts. Again it is possible to measure frequency, intensity, or duration. Some variables can be measured through observation; other variables e.g. feelings or thoughts, you would need to rely on self-report. You can have clients fill in paper and pencil devices, self-reports, or surveys/scales. All clients should be measured using the same procedures. This is what is meant by being systematic.

Once the dependent variable has been operationally defined and measured prior to intervention, you can use material from internet, library searches, books, journals and experience to <B>locate interventions that have been successfully employed given the type of problem that your clients are experiencing. Next you select empirically proven interventions; minimally, you should choose interventions that have a logical basis given the reasons behind your clients' problems. For example, social skills training would logically be helpful for shy clients who lack the know-how and experience of initiating conversation.

You need to <B>explain the intervention that you will provide for each client</B>. What format does the intervention take? Is it couple counseling? Does it involve group work? How many sessions will be given? How often is each session? How long is each session? Over what period of time will the intervention take place? Who attends each intervention? What is structured to happen each session? You should describe the intervention specifically so that another person could replicate what you have done with different clients if s/he wanted to test your treatment again. Again you are being systematic.

By offering the intervention to some clients-- those in the experimental condition-- and not providing intervention for clients in the control condition, you have a basis of comparison. You have reason to believe, if you used randomization (or chance to arrive at 2 groups) that the only difference between the clients/subjects in the experimental and control conditions is that experimental clients have received the intervention, and control clients have not. Having a randomized control group allows you to rule out the effects of outside influences (testing, maturation, selection and statistical regression to the mean, to name a few of the extraneous sources of variation) on the dependent variable. If there is change in the dependent variable shown in a difference between pre- and post-intervention measurements of the dependent variable, then that change can be attributed to the presence of the intervention.

A) The classic experiment uses at least 2 groups or conditions; the experimental condition and the control condition. The two groups are arrived at by using randomization. Both groups are measured prior to intervention and these measurements are known as the pre-test measurements of the dependent variable. The experimental condition clients are exposed to the intervention and the control condition clients are not. Then both groups of clients are measured a second time; after the experimental clients have completed their exposure to the intervention. This measurement interval is known as the post-test even though the control group clients have not been exposed to the intervention.

The symbolic diagram for this group design is--

R-O-X-O; this first group is the experimental condition

R-O----O; this second group is the control condition

R stands for randomization or using chance to determine which clients are in the control and experimental groups.

O stands for observations or measurements of the clients' behavior, feelings, thoughts. The dependent variable is being measured.

X stands for being exposed to the intervention; a - means the clients have not received any intervention

O- again stands for observations or measurements of the clients' behavior, feelings, thoughts; but this is a post-intervention measurement for the experimental group clients.

It is possible to do a simple pre-test, post-test one group design. It would be diagrammed as--

O-X-O.

C) You can also do a static group comparison design in which two groups are being compared but the clients in both groups are only measured once. It would be diagrammed as--

X-O; the first group of experimental subjects receive an intervention &amp; are then measured

----O; the second group of comparison subjects do not receive an intervention but they are measured

You then compare the average score of the experimental subjects with the average score of the comparison group subjects. It's called a comparison group because the two groups were not arrived at using randomization but were probably pre-existing.

For other types of group designs, you can review a research text and/or speak with a research professor.

NOTE! Just because these are known as group designs, does not mean that the intervention must be presented in a group. Clients can still receive one-on-one intervention.