** Chapter 4 Quasi Experimental And Single case study **

Quasi experimental and single case designs arc parity-hut nut hills true experimental designs; they control sonic but not all of the some internal invalidity. Quasi-experimental designs exist for situations in which complete experimental control is difficult or impossible. For example in the real educational world ie, that world confronting the educational researcher, it is possible that the school systems may not allow inter classes to be disrupted or divided to provide for random or equivalent samples or may not allow for a "treatment" to be given to some and withheld from others. Quasi experimental exists for cases where the power to randomize is either seriously curtailed or altogether absent .In other cases where the researcher might have access to only single mateal group, such as a classroom of individuals with learning disabilities, or to only one or two participants , such as a student with school fobia, the only possible way to do the research will be quasi -experimentation, and single case research designs In this respect several specific research designs were developed under the general headings of quasi experimentation.

The following pages will deal with the most important quasi experimentation designs.

1. The Interrupted time Series Design: In the interrupted time series design, there is only a single group of participants. In this case, the group is pre tested a number of times during the baseline phase Baseline refers to the observation of a given behavior before the presentation of any treatment. After this, the group is exposed to a treatment condition and often post tested a number of times during the treatment phase, as depicted in the diagram below.

Multiple pretests Treatment Multiple posttests 0,0,0.0,0, X, 0,0,0,0,010 Where O, O; O, O. O, are the multiple pretests and O, O, O, 0, Or0 are the multiple posttests and x is the treatment.

To illustrate this design, let us consider the following example .Two researchers wanted to investigate how making the classroom environment more positive affected the percentage of students who were engaged in (heir assigned activities In this experimental classroom, emphasis was placed on the positive. For example, classroom rules were stated positively (eg, show courtesy and respect to others), and points and praise were given to the students when they followed the rules .The students who were engaged in the assigned activities were defined as "oil task". The percentage of students who were on task were repeatedly measured before and after the teachers focused on making the classroom more positive .The students were assessed during the first seven class sessions before the implementation of the positive classroom environment. After the experimentation of the positive classroom environment, the same students were assessed over the next six class sessions.

Even if observation shows i ncrease in the number of on task students, we as researchers have to answer two important questions .First, was the change that occurred after the treatment significant? Second, can the observed behavior be attributed to the treatment condition? The answer to both the above questions need a test of significance

To sum up, a researcher using the interrupted time design must t consider all other events taking place at the lime of implementation of the experimental treatment and determine whether there might be rival explanations.

**2. Regression -Discontinuity Design: **Regression discontinuity design is a design that asseses the effect of a treatment condition by looking at a discontinuity in regression lines between individuals who score lower and higher than sonic predetermined cutoff score.

To illustrate what we are talking about consider the next example.

A researcher wanted to find out whether a certain program given to the gifted students enhanced their achievement more than a regular program. The cutoff score to be admitted into the gifted program was set at two or more standard deviations above the mean on the Standard-Benet intelligence test .Students who received this cutoff point were admitted to the program few the gifted, and students who did not receive the cutoff score were considered the control group and admitted to the regular program .Three years after the initiation of the gifted program .California Achievement Test was administered to both groups. Statistical analysis of the outcome data .demonstrated that a significant discontinuity lines for the two groups of students did not exist, indicating that the gifted program did not enhance the bright students academic achievement The next figure illustrates the results.

The straight line represents the regression line, Since the regression line is continuous, it indicates that the individuals that received the special treatment made a score of higher than 50 on the pre-assignment variable. The continuous regression line indicates that there was no effect of the treatment because the scored of people above cutoff of 50 and receiving the treatment continued the pattern of scores of people below the cutoff of 50 who did not receive the treatment.

Another example that illustrates this design: A school decided to initiate a remedial program for students who have been found to have deficiencies in reading skills at grade 5 (primary). A reading test was administered to all students in the school at this grade level A certain score on the test was used as a cutoff, so all students scoring at or below the cutoff score arc placed in the remedial program while students who scored above the cutoff point were placed in regular classes of instruction (strict adherence to the cutoff score is required in a regression discontinuity design).

After the establishment of the groups remedial and regular Appropriate v instruction was dispensed to each group .At the conclusion of the treatment, all students in both the regular and remedial groups were again administered a reading test. Analysis of data showed a significant discontinuity of regression lines for the two groups.

This discontinuity indicated that the treatment had an effect .see the next graph:

To understand the difference between an effective and a non- effective treatment look at the above figure. The discontinuous line indicated that the treatment had and effect.

The regression discontinuity design is an excellent design that can be used when researchers want to investigate the efficiency of some programs or treatment but cannot randomly assign participants to comparison groups h is in fact more powerful than other quasi -experimental designs if the following criteria are met .

**Single case experimental Designs:**

Single case uses a single participant in the experimentation design to investigate the efficacy of an experimental treatment condition. A study that investigates an individual can be used anytime you want to investigate some phenomenon but have access to only one or two individuals who demonstrate this phenomenon. For example, assume you have an unusually bright student in your class and you want to study this person's learning strategies Because only one student with this ability level is in your class, you have to use a single case design.

Single- case always requires repeated measurement on the dependent variable prior to and following implementation of the experimental treatment condition .Treatment responses are used as comparison responses for assessing the effect of the independent variable.

Single- case and experimental designs, are frequently used with an intact group of participants as well as with single case participants. There are times when you cannot break a group of participants as a class into an experimental and control groups but still want to investigate the efficacy of an independent variable. In these instances you can treat the class as a single- case experimental design.

**A-B-A-B designs: **A-B-A-B is a design that is extended to include the reintroduction of the treatment condition. Suppose we have a student called Michael, who was identified as having a high rate of disruptive behavior during academic. instruction He used to talk without permission .walk away from the instructional area without permission and make non-directed negative verbalization. This was recorded for ten class periods .Baseline recording began from the moment the teacher gave him math assignment and ended with the completion of the math activity or the expiration of thirty minutes of observation.

The next phase was the introduction of the treatment condition .The teacher provided Michael with the information that would insure his getting the correct response before presenting him with the task he was to perform. For example during intervention the teacher would say" Michael 6x4 = 24 What is 6x4?" if he had miscalculated the problem.

After Michael completed seventeen periods with intervention, the teacher reverted to her baseline behavior of not providing information that would ensure a correct response. The teacher then went back to introduce the treatment condition for a second time. If Michael ends the experiment with the positive effect aspired, then the treatment condition would be considered effective, Any reversal in behavior means that the change in behavior was not caused by the treatment condition but by an extraneous factor, Once the influence of the experimental treatment has been demonstrated, attention could be placed on its persistence.

Multiple -Baseline Design: This is a single case experimental design in which the treatment condition is successively given to different participants or same participant in several settings after baseline behavior? has been recorded for different periods of time.

The multiple-baseline design, focuses on two or more different behaviors in the same individual or on the same behavior exhibited by one individual but in different settings.

Let us focus on the same behavior exhibited by four different individuals. With this focus, the first phase of the design involves collecting baseline behavior on all four individuals. During the second. phase, the treatment condition is administered to the first individual, and baseline behavior is maintained on the other three individuals. For each subsequent phase, the treatment condition is successively administered to the next two individuals, if the individual who is exposed to the treatment condition shows change in behavior and no behavioral change exists for those who continue the baseline behavior, evidence exists supporting the efficacy of the treatment condition. See the next figure that illustrate this design .

Phase 1 Phase 2 Phase 3 Phase 4 Phase 5 Different A Baseline Treatment Treatment Treatment Treatment people Different B Baseline Baseline Treatment Treatment Treatment Behaviors, C Baseline Baseline Baseline Treatment Treatment or different D Baseline Baseline Baseline Baseline Treatment settings

Changing-Criterion Design: The changing criterion design is a single case design in which a participant's behavior is gradually altered by changing the eriterion for success during successive treatment periods. Thus we see that this design is used whenever we require shaping behavior over a period of time or in case where a step by step increase in accuracy, frequency or amount are the goals of the research.

The changing-criterion design, requires an initial baseline measure on a single target behavior.

A treatment condition is thus implemented and continued across a series of intervention-phases. During the first intervention or treatment phase, an initial criterion of successful performance is established. When the participant achieves the established level of performance, the experiment moves to the second phase, where a new and more difficult criterion level is established while the treatment condition is continued. When behavior reaches This new criterion level and is maintained, the next phase with more difficult criterion level is introduced, in this manner, each successive phase of the experiment requires a step by step increase in the criterion measure.

Next is an example to illustrate this design, Let us take the case of a child called Ahmad who refused to complete arithmetic problems, To overcome this resistant behavior, the teacher first obtained a baseline measure of the average number of assigned arithmetic problems that Ahmad would complete during a forty five minute session. Then Ahmad was told that a specified number of problems had to be completed correctly during the subsequent session. If he completed them successfully, he could take recess and a star; if he does not\he would have miss recess and stay in the room until they were correctly completed, During the first treatment

phase, the criterion number of problems to be solved was set at five, which was one more than the mean number completed during the baseline phase. After successfully achieving this criterion performance on three consecutive days .Ahmad had to finish an additional problem to take recess and a star. The result of the experiment showed that Ahmad»s performance increased as the acceptable criterion level was increased. This demonstrated the effects of the treatment, Ideally, the treatment phase should be of different lengths or if they are of a constant length, the baseline phase should be longer than the treatment phase. This is necessary to insure that the step-by-step changes in behavior are caused by the experimental treatment and not by history or maturation variable that occuis simultaneously with criterion change. Each treatment phase should be long enough to allow the behavior to change to its new criterion level and then stabilize. The criterion change should be large enough to detect a behavioral change but small enough that it can be achieved.

The discussion of the single- case experimental designs mentioned above shows only the most basic and commonly used designs. Regardless of the design used, there are several methodological issues that must be considered in attempting to conduct a single-case studs. These are summarized in the following table:

Methodological issues in single-case studies 1. Baseline -The behavior of the participant before the administration of the experimental treatment condition. Baseline serves as the standard for assessing change induced by the experimental treatment condition, and to serve as this standard, it must be stable. A stable baseline is characterized by (a) an absence of trend or baseline data that does not increase or decrease over time and (b) little variability which is defined by Sidman (1960) as only five percent variability.

2. Changing one Variable at a Time-In single case studies, only one variable can be changed from one phase of the experiment to the next. This is necessary to be able to isolate the effect produced by that variable,

3. Length of phases-There is no agreement among researchers with regard to the length of phases. Some researchers state that all phases should be of equal length and others state that each phase should be continued until some degree of stability has been achieved.

4. Assessment of Treatment Effect; There are two approaches to assessing the effect of an experimental treatment:

**(a)Visual inspection:** Looking at the data generated at the various phases is sufficient to identify a treatment effect if the baseline and intervention data do not overlap there is a change in the level of the baseline and intervention phase)or if the trend of the data in the baseline phase is different from that which exists in the intervention phase.

**(b)Statistical Analysis: **A statistical analysis such as a time series analysis is necessary if there is a great deal of variability in the data. In general, the statistical analysis is not needed if there is little variability in the data and the baseline data is very stable, When those two conditions do not exist, a statistical analysis should be used.

To sum up the changing -criterion design can be used when the goal is to increase the amount, the accuracy, or frequency of some behavior. Alter a baseline is established, a treatment condition is administered following attainment of specific criterion of initial successful performance. When this performance has been achieved, the criterion is progressively increased until the desired behavior is attained.