|
|
|
|
|
Experimental Designs (Part II) Key Terms or Concepts to Understand
Module Objectives: When you have completed this module, you should be able to: 1. Interpret the SPSS Outputs from the MANOVA/MANCOVA procedure. (Case Assignment) 2. Know how to report and discuss MANOVA/MANCOVA (Case Assignment) 3. Compare and discriminate between the application of true experimental research designs to the health sciences. 4. Explain the theory supporting clinical trials and the importance of randomization and blinding in these studies. Experimental Designs (Part II) In Module 3, we discussed the different experimental designs and the elimination of bias and the "Concept of Control." In this module we will examine closely some important issues with the experiment and the design. We will also critically review clinical trials. The classic experimental design is usually called a Pretest-Post-test Control Group Design. It consists of randomly selecting subjects from a defined population and assigning them to either the experimental or control group. The researcher cannot influence the assignment of individuals to either group. This design is typically described as follows: Experimental R O1 X O2 Control R O3 O4 The difference between the two groups can be described as follows: Experimental R O2 - O1 = de Control R O4 - O3 = dc This design allows for tests of causality. In fact, for an experiment to demonstrate causality three distinct operations must occur. They are covariation, nonspuriousness, and temporal or time ordered. Covariation means that phenomena vary together. Thus, if you theorize that one phenomenon causes another they must covary. if they don't covary then one phenomenon cannot cause the other phenomenon. The researcher must also show that the relationship between the phenomena of interest is nonspurious. This means that the relationship between the phenomena cannot be explained away by a third variable. Therefore, the study must be designed appropriately to CONTROL all extraneous variables. The third condition that the researcher must demonstrate is the the cause precedes the effect. This is called a temporal condition or relationship since it is also a function of time. The three conditions described above are also part of the "logical model of proof" in true experimental studies. The classic experimental research design consists of four steps or processes. These processes are comparison, manipulation, control, and generalizability. Comparison is related to covariation; manipulation is the exposure or the treatment and is related to temporal factors; and control is related to nonspuriousness and covariation. Generalizability is related to sampling and the representativeness of the sample population. CLINICAL TRIALS Clinical trials are very important to the field of medicine since it allows a researcher to conduct an experiment intended to demonstrate cause and effect while avoiding some serious ethical issues found in the classical experimental design. Clinical trials "work" in the real world because they operate in an environment where "variability is the law of life." People vary greatly and no two individuals will behave in exactly the same way. Clinical trials are also invaluable to the researcher since it allows the control group to be treated for the disease of interest in most cases. Thus, some significant ethical issues are avoided. A clinical trial can be described as a randomized controlled clinical trial comparing one therapy or treatment versus another therapy or treatment. The control therapy or treatment should be the accepted standard treatment by which the effectiveness of the experimental treatment or therapy is compared. In some clinical trials the accepted standard treatment is actually the "gold standard" which is the best treatment available for use as a comparison standard. Steps in a Clinical Trial The first step in conducting a clinical trial is to develop the research question or hypothesis. This is very important since the the research question determines the type of treatment/therapy to be compared, and the nature of the outcomes to be assessed by the researcher. Probably the most important question, at this stage in the development of the clinical trial, to be answered is the parameter that will answer the research question. This parameter is called the primary end point. Examples of primary end points are quality of life, survival, complications, and immediate measures. (See SLP Module 6) Sample Size Determination The sample size of the study is an important consideration. Many clinical trials use a small sample size; however, these studies usually have to be repeated at a later date with a stronger sample size. This generally occurs when the results of a clinical trial utilizing a small sample size are considered to be clinically significant but lacks the statistical power to be useful to the profession. Of consideration to the researcher in determining sample size is whether or not a significant finding would be important to providing care for the type of patient studied in the clinical trial. Randomization Random assignment is essential for any clinical trial. Without proper randomization of the study's patients internal validity factors cannot be controlled and the results of the study will not be creditable. Placebos and Blinding The use of placebos and blinding is frequently used in clinical trials to control the influence of the treatment on the patients' perception of their disease and treatment. A single blinded study is used to control patient bias and a double blinded study is used to eliminate observer bias. |
|
|
