WHERE WE'VE BEEN

We have introduced the topic of research methods as a set of tools that enable educators to think better, develop and understand educational principles, and make increasingly broad and accurate generalizations.

WHERE WE'RE GOING NOW

In this chapter we'll describe the process of identifying research variables and clarifying the research problem. We will discuss several types of variables that can become part of a research study and ways to integrate these into the statement of a research question or hypothesis in order to lay the foundation for a well-conducted, well-interpreted, and well-reported research study.

CHAPTER PREVIEW

This chapter will introduce the concept of research variables¹. No educational research-whether practical or theoretical-can be conducted without identifying, conceptually developing, and operationally defining or describing in detail the research variables. Much educational research consists of the conceptualization, description, or careful operational definition of educational variables. These descriptions enable educators to develop realistic depictions of persons, programs, and contexts related to education and to analyze and understand more accurately what is happening in various educational settings. While they are useful in their own right, these descriptions also lay a foundation for research that looks at more complex relationships among variables.

¹ We are going to use the term variable several times in this chapter before we eventually define it. If this does not bother you (that is, if the meaning is sufficiently obvious from the context), skip the rest of this footnote.

A variable is a conceptual entity, an invention based on reality that exists in the minds of people. Variables are so called because they vary (that is, they may take on any of several values; thev are not constants).

The present chapter introduces variables studied in several types of educational research. The concepts introduced here will provide a useful foundation for understanding the ideas discussed in subsequent chapters. In addition, since many readers will be conducting research or writing research proposals while reading this book, it is important to enable them to begin identifying and examining research variables and investigating reference materials related to these variables as early as possible in their study of research methodology. All of the concepts introduced in this chapter will be discussed in greater detail and integrated with specific research methodologies in subsequent chapters.

To conduct research yourself at any level or to interpret the research of others, it is Important to understand the nature of the variables in research studies. After reading this chapter, you should be able to

1. Identify examples of educational problems and specify the variables that constitute the statement of these problems.
   
   
2. Define and give examples of research questions and hypotheses.
   
   
3. Define and give examples of (a) quantitative descriptive, (b) qualitative, (c) experimental, (d) correlational, and (e) meta-analytic research studies.
   
   
4. Define and give examples of (a) dependent (or outcome or criterion), (b) independent (or treatment or predictor), (c) moderator, (d) control, (e) intervening, and (f) extraneous variables.
   
   
5. Combine a set of given variables into a correctly stated hypothesis.
   
   

EDUCATIONAL RESEARCH: A USEFUL TOOL

As we stated in chapter 1, educational research is a useful tool for both practical and theoretical problem solving. It helps us to identify concepts, contexts, treatments, and outcomes; to make predictions; and to determine the nature of relationships among variables in education. It helps us make generalizations about various aspects of education and to determine the degree of confidence we can have regarding these generalizations.

The educational research process begins with the realization that a problem exists. The following are examples of problems related to education:

1. Some of the members of a school board want to eliminate corporal punishment from the school system, but they are concerned that its removal may result in rampant anarchy. Members of a child advocacy group argue that the elimination of corporal punishment will actually improve self-discipline among the students.
   
   
2. The parents in a school system are aware that AIDS is a serious health risk for their children. They want their children to avoid this disease, but they are also concerned that introducing children to the issues could lead to sexual promiscuity.
   
   
3. A teacher needs to know whether his method of discipline, which definitely seems to enhance academic learning, might also be having the unpleasant side effect of promoting negative self-concepts among his students.

Educational research does not in itself solve these problems. It provides a tool that gives hard-working, conscientious educators some of the ammunition they need to successfully attack them.

The first step in the educational research process involves identifying and clarifying the problem. Only after we start to know what the problem is can we take steps to solve it. We can identify and clarify research problems by specifying the variables involved in those problems, stating these variables in research questions or hypotheses, and operationally defining these variables or describing them in detail. Often the variables are not easy to identify, and often they resist easy definition. However, only after we have identified and operationally defined the variables or described them in detail can we sensibly collect data, relate them to one another, and attempt to make generalizations that will help us solve our educational problems.

STATING RESEARCH QUESTIONS AND RESEARCH HYPOTHESES

Research questions ask about the nature of a variable or concept or about the relationship among two or more variables. Research hypotheses state the expected answers to these research questions. Research questions and hypotheses give focus, structure, and organization to the collection, analysis, and interpretation of the data collected by the person conducting the research.

A good researcher (or for that matter, a good thinker) would say that, stated in their present format, all three of the problems cited earlier (page 18) are too vaguely expressed to generate clear solutions. For example, what 'is the real problem in the first description?

1. Is the problem that somebody wants to know whether the removal of corporal punishment could objectively cause an increase in "rampant anarchy" (whatever that means)?
   
   
2. Is the problem that somebody doubts that the removal of corporal punishment could objectively cause an increase in "self-discipline" (whatever that means)?
   
   
3. Is the problem that somebody wants to know what it will take to placate the child advocacy group? If board members retain corporal punishment, can they withstand the hostile publicity that will be evoked from this group?
   
   
4. Or is the problem that the board members simply don't have enough information to make a sensible decision? Maybe they need answers to specific questions, like the following:
   
   
   1. How often is corporal punishment actually administered in the school system? How does the frequency of corporal punishment in this school system compare with that in the rest of the country?
      
      
   2. How does the frequency of corporal punishment compare with other forms of punishment?
      
      
   3. Do students really experience negative effects from corporal punishment? Assuming that there are negative effects, is corporal punishment more likely to cause negative effects than other strategies, such as detention or school suspension)
      
      
   4. What do teachers do when they give up corporal punishment, Do they become permissive and let students get by with anything) Do they resort to other strategies, which may be more harmful in the long run? Or do they begin to reason together with the students to arrive at solutions to disciplinary problems?

The preceding questions are more specific than the original problem. An educational researcher can now begin to offer information that will help solve the problem, because it is now more clear what the problem is. The difference between the original question and these refined questions Is that the latter focus more on specific variables and the relationships among them.

RESEARCH VARIABLES

A variable is a concept that can assume any one of a range of values. Factor and feature are common English synonyms for variable in the following sentences (outcome is another word that often means variable, but note that not all variables are outcomes):

1. The parents in the school system are aware that AIDS is a factor (variable) that is a serious health risk for their children.
   
   
2. The teacher needs to know what features (variables) of his method of discipline might promote negative self-concepts among his students.
   
   
3. The teacher needs to know whether his method of discipline might promote negative outcomes (variables) among his students.
   

A variable varies in the sense that it can take on different values or conditions. It is a characteristic that can be the focus of a research study.

The definition of the term variable makes the concept seem more difficult than it really is. This is because it is a definition of an abstraction; and definitions of abstractions sometimes become -very abstract (for example, it's easy to exist, but it's difficult to define existence). Therefore, it may be useful to look at some examples.

• In the set of examples we started examining on page 18, corporal punishment is a variable. More specifically, whether corporal punishment is permitted is a variable. It can vary by either being permitted or not being permitted.
  
  
• The frequency of corporal punishment could also be a variable. It can vary from, say, not happening at all to happening once a day to happening several times a day.
  
  
• If a researcher wondered whether corporal punishment might cause more anxiety than suspension from school, then corporal punishment, anxiety, and suspension from school could all become variables for this researcher to study.
  
  
• If the researcher focused attention on any other factor, such as child abuse, self-discipline, religious beliefs, gender of the students or teachers, or ethnic background, these could become variables in a research study, or even merely in a thought process.
  

Variables are developed theoretically and abstractly in the mind of the researcher. The researcher draws on previous experience and theoretical knowledge to identify and specify variables, which serve as unifying factors to help organize the research process. All human thinking employs concepts, which are called variables when we focus attention on them, measure them, or deal with them in some other way as part of the research process. (The ways researchers define and measure variables will be the subject of later chapters.)

The next section of this chapter discusses the basic categories of studies that examine variables of interest to educators.

TYPES OF RESEARCH STUDIES

The following sections describe several basic categories of research studies that serve useful purposes in education. These descriptions should enable you to see what kinds of studies are possible and begin to identify research variables. These categories are not mutually exclusive. For example, the first category is called descriptive research, but actually all research is descriptive. At this point in the book, we are not going to worry about the degree of overlap among these categories. As Table 2.1 shows, strategies for conducting each type of research are covered in later chapters. We will wait until the distinctions among the types of research have practical significance before making those distinctions.

Type of Study	Description	Chapter in Text
Quantitative descriptive research - status study	The researcher uses quantitative strategies (such as questionnaires or observational techniques) to collect information about the characteristics of a person, group, program, or other educational entity.	4 through 8
Qualitative research	The researcher uses qualitative strategies (such as ethnographies or detailed case studies) to collect detailed information about the characteristics of a person, group, program, or other educational entity.	9
Experimental/ quasi-experimental research	The researcher assigns participants to treatments or otherwise structures conditions to determine whether a treatment has an effect on a specified outcome variable.	11-12
Correlational research	The researcher uses statistical techniques such as correlation coefficients to examine the relationships among two or more variables, without making generalizations about the causal nature of these relationships.	13
Criterion group research	The researcher examines the characteristics of existing groups (that is, groups not experimentally assigned to particular treatments) to determine the degree to which these groups differ with regard to specified characteristics.	13
Meta-analysis	The researcher combines and analyzes the results of several studies to examine the degree to which these combined results lead to generalizations regarding the variables in the studies.	16

Descriptive Studies - Status Research

In almost every educational research study, the researcher is interested in measuring an outcome. (The researcher may also be interested in other variables, but these will be discussed later in this chapter.) Outcome variables are concepts or characteristics of end results or products that can occur in educational settings. They are the results - the characteristics or events that teachers and other educators may wish to encourage or discourage in educational settings. In the first revised hypothesis of the corporal punishment problem, anarchy is the out come upon which the question focuses. In the second example, the outcome variable is self-discipline. In the third example, the outcome is the degree to which the board can withstand the hostility of the advocacy group. Try it yourself: What are the outcome variables in statements (a) through (d) of the fourth example on page 19?

Answers:

a. frequency of corporal punishment

b. frequency of corporal punishment and also frequency of other forms of punishment

c. negative side effects

d. types of punishment employed by teachers; permissiveness exhibited by teachers; other strategies employed by teachers; reasoning to arrive at solutions

In actual practice, the school board members would usually be interested in more than just one of these outcome variables. By refining the questions through identification of the outcome variables, they would be able to focus more accurately on a solution to their problems.

Let's assume that one of the questions the school board actually wants answered is how often corporal punishment really occurs in the school system. How would a researcher collect data? Table 2.2 shows the steps that would be involved in answering this question (the next few chapters describe in detail how to conduct a research study of this kind). The second question asks about the frequency of corporal punishment in this school system compared with that in the rest of the country. The school board members could simply look that answer up in a professional publication; Table 2.3 shows the steps that would be undertaken by a researcher commissioned to find the answer to this second question.

Descriptive research studies often focus exclusively on the measurement and description of outcome variables (in chapter 1 we referred to this as Level I research). The strategies for conducting descriptive studies are the focus of chapters 4 through 8. When these studies collect numerical data - for example, by counting how often corporal punishment occurs, by using a questionnaire to measure people's attitudes toward corporal punishment, by relating these variables to other variables, or by making predictions-these studies are referred to as quantitative descriptive studies. Qualitative studies (chapter 9), on the other hand, often study variables and outcomes in a more global sense; for example, a researcher may observe a classroom over a prolonged period of time and record anecdotal information regarding what happens when disruptions occur. Both of these are valid and important ways to describe educational contexts and outcomes.

Experimental Research

Experimental studies do more than describe contexts and outcomes-they examine cause-and-effect relationships. For example, if our school board members want to know whether the removal of corporal punishment may cause an increase in anarchy, they are asking about the causal nature of the relationship between the removal of corporal punishment and anarchy. To answer this question, they need more than descriptive research-they need to examine the results of experimental research!.

An experimental study focuses on a treatment and an outcome. (The study may also include additional variables, such as the context in which the experiment is conducted.) The outcome that is expected to result from the treatment is referred to as the dependent variable, or criterion variable. Such outcome variables are dependent in the sense that they depend on the treatment. They offer a criterion that can become the basis for making a judgment about the results of the experiment. If the treatment has one effect, a person's score on the dependent variable may be low; whereas if the treatment had a different impact, that person's score on the dependent variable would be high.

The treatment, context, or condition that is expected to produce an outcome is referred to as the independent variable (the terms independent variable and treatment or condition are used synonymously in this text). The independent variable is independent in the sense that it does not depend on the outcome variable. Another way to say this is that in experimental research the independent variable is the cause and the dependent variable is the effect. The independent variable is the treatment; the dependent variable is the outcome.

REVIEW QUIZ 2.1

Examine each of the following hypotheses or questions and identify the dependent and independent variables (all answers to Review Quizzes in this text are at the end of the respective chapter):

1. Studying Shakespeare leads to a greater appreciation of Western culture.
   
   
2. Behavior modification leads to noncreative behavior among elementary school children.
   
   
3. Does programming in the Logo language cause improvement in higher-order thinking skills?
   
   
4. Delaying reading instruction until the sixth grade will have no adverse impact on reading ability by the time the child reaches adolescence.
   
   
5. Do classrooms that emphasize a language-rich environment have students who are more eager to share their opinions with their peers?
   
   Click here for answers.

In each of the cases in the preceding quiz, the independent variable produces an outcome, and this outcome is referred to as the dependent variable.

You may have noticed that some of the variables (such as noncreative behavior, studying Shakespeare, and appreciation of Western culture) are vaguely defined. Operational definitions (discussed in chapter 4) will help remove this ambiguity.

Let's assume that one of the questions the school board actually wants answered is whether the removal of corporal punishment would increase disruptive behavior in the school system. How would a researcher collect data? Table 2.4 shows the steps that would be involved in answering this question (chapters 10 through 14 describe in detail how to conduct a research study of this kind). Note that while Table 2.4 recommends an experimental strategy to answer this question, it would also be possible to deal with this question through a qualitative research strategy (chapter 9) or a correlational strategy (chapter 13) or by conducting a meta-analysis (chapter 16).

Correlational Studies

Correlational studies are descriptive studies in which the researcher not only describes variables but also examines the nature of the quantitative relationships among them. For example, a researcher might suspect that children who often receive corporal punishment are more likely to hit other children during recess. This researcher is interested not only in two variables (receiving corporal punishment and hitting other children during recess) but also in the relationship between these two variables (and perhaps others). The researcher could investigate this possibility by giving a questionnaire to a group of children to find out how often each child has received corporal punishment and by observing the same group of children on the playground to see how often (if ever) each child hit another child. If there was a tendency for those who hit other children more often to be the same children who received corporal punishment more often, this evidence would support the belief that there is a relationship. This kind of study would enable the researcher to determine whether there is a relationship between corporal punishment and hitting other children in this group of children.

Each of the following three hypotheses could be investigated by correlational methods:

• The students who do well in algebra will also do well in gym.
  
  
• People who arrive early for their therapy sessions usually arrive early for athletic events.
  
  
• There is a strong positive relationship between GRE scores and success in this course.

The steps in a correlational research study are the same as those described in Table 2.4, except that the experimental design is replaced by a correlational method, which is discussed in chapter 13.

Note that the existence of a relationship does not necessarily mean that one of the variables is the cause of the other. For example, if there is a relationship between the incidence of corporal punishment and hitting other students, this could be because corporal punishment caused hitting, because hitting caused corporal punishment, or because some other prior factor (such as a negative self-concept) led to both increased corporal punishment and increased hitting.

MODERATOR VARIABLES

Moderator variables are characteristics that influence (moderate) the impact of the independent or treatment variable upon the dependent or criterion variable. Moderator variables appear only in studies that include independent and dependent variables - usually experimental studies. In the following hypothesis, for example, the sex of the child is a moderator variable:

• The use of popular music will increase the appreciation of poetry among elementary school children. This impact will be greater among girls than among boys.
  

In this example, the use of popular music is the independent variable and appreciation of poetry is the dependent variable. The sex of the child is a moderator variable because the statement indicates that this factor influences the degree to which the independent variable will influence the dependent variable: if the child is a girl, popular music will tend to have one impact with regard to the appreciation of poetry; whereas if the child is a boy, the same independent variable will tend to have a different impact.

In the preceding example, the sex of the child was merely one of many variables that could have been chosen as moderator variables. The following are two other variations of the same hypothesis, with a different moderator variable inserted in each case:

The use of popular music will increase the appreciation of poetry among elementary school children. This impact will be greater among older children than among younger children.

The use of popular music will increase the appreciation of poetry among elementary school children. This impact will be greater when the teacher is of the same sex as the child than when the teacher is of the opposite sex.

You can easily see the advantage of using such moderator variables. The researcher (or reader) can discover not only whether the treatment has had the desired impact but also under what circumstances or contexts this impact is likely to vary. This is extremely useful information for helping us generalize the results to other situations, as well as for helping us selectively apply results to specific types of learners within a given educational setting.

The following are restatements of the first two hypotheses in Review Quiz 2.1. Moderator variables have been added to each hypothesis. To avoid stereotyped language and to show a fuller range of how moderator variables can be introduced into research studies, the wording has been varied in each hypothesis.

1. Studying Shakespeare leads to an increased appreciation of Western culture among collegebound students but not among non-collegebound students.
   
   
2. Behavior modification will reduce creative behavior among elementary school children in middle-class schools but will increase creative behavior in lower-middle-class schools.
   
   Now you try to identify the moderator variables in the following hypotheses, again adapted from Review Quiz 2.1:
   
   
3. Programming in Logo will cause substantial improvements in higher-order thinking skills among slow readers who are not classified as having learning disabilities. However, no such improvements will occur among slow readers who are classified as having learning disabilities.
   
   
4. Delaying reading instruction until the sixth grade will have no adverse impact on reading ability by the time the child reaches adolescence. This absence of impact will be equally true among low-IQ, medium-IQ, and high-IQ students.
   
   
5. Elementary school classrooms that emphasize a language-rich environment will have students who are more eager to share their opinions with their peers. However, in middle school classrooms, language-rich environments will have no such impact.
   
   

The moderator variables in the preceding hypotheses are as follows:

1. collegebound vs. non-collegebound students
2. social class of the schools
3. being classified as having learning disabilities versus not being classified as having learning disabilities
4. IQ of students
5. grade level of classroom (elementary versus middle school)

In each of these examples the original hypothesis has been enriched by the addition of the moderator variable. In each case the moderator variable is a factor that could influence the impact of the independent variable upon the dependent variable. For each example, other factors could have been selected as moderator variables; the actual moderator variables selected in any study depend on what the researcher is interested in learning about the relationship between the independent and dependent variables.

CONTROL VARIABLES

Control variables are characteristics that are controlled by the experimenter to reduce any impact they might otherwise have on other variables or on the interpretation of the results of a study. They control extraneous variables (which are discussed later in this chapter). This control can be attained through any of the following methods:

1. Isolation and elimination. For example, a researcher can control the extraneous effect of intelligence by including only students of average intelligence in the study-excluding low-performing or gifted students. This is an effective strategy described further in the present chapter.
   
   
2. Precise description of the context, treatment, or subjects. A detailed, accurate, and unambiguous description enables persons applying the research findings to make a valid judgment regarding the degree to which the results of a particular study can be generalized to other settings. This is an effective strategy covered in chapters 9 and 15.
   
   
3. Equating across groups. This method is often used in experimental studies. For example, a researcher can control the extraneous effect of intelligence by randomly assigning subjects to the experimental and control groups, thereby making them similar with regard to intelligence (and many other variables). This is an effective strategy described in chapter 11.
   

In any research study, the researcher can select and describe control variables to help define the limitations on how far the results of the study can be generalized. Furthermore, in experimental studies, the researcher may select control variables because these factors may otherwise influence the impact of the independent variable on the dependent variable. such extraneous influences would make it difficult to determine the precise nature of the relationship between the independent and dependent variables. Therefore, the researcher controls these extraneous factors to balance or reduce their influence in a particular study.

Let us examine two of the sample hypotheses that we have been following throughout this chapter, and this time we shall add control variables:

1. Among high school seniors, studying Shakespeare leads to a greater appreciation of Western culture.
   
   
2. Behavior modification leads to noncreative behavior among elementary school children in the public schools of Gotham City.
   

In the first hypothesis, "high school seniors" has been added as a control variable. It is possible that the effect of studying Shakespeare would be different for students of different ages; therefore, the grade level of the students might influence how they would respond to Shakespeare. Thus the researcher has decided to include only high school seniors in the research study. This constraint rules out the extraneous factor, but of course it also restricts the generalizations to those made about high school seniors only. The logic goes like this:

Students In the experimental group (who studied Shakespeare) were measured as displaying greater appreciation of Western culture than the students in the control group (who did not study Shakespeare).

Since there were only high school seniors in both groups, differences in appreciation of Shakespeare could not possibly have arisen from basic differences between juniors and seniors.

Therefore, the influence of grade level (seniors versus juniors) has been controlled; but generalizations from the experiment have also been restricted to those made about high school seniors only.

 

In the second hypothesis, there are three control variables: grade level of children (elementary rather than secondary or middle school), type of school (public rather than private), and location (Gotham City).

In the preceding examples we added control variables to the original hypotheses before they were enriched by the addition of moderator variables. It is possible (and often desirable) to have both control variables and moderator variables within the same research study. Each of the following examples contains independent, dependent, moderator, and control variables. The moderator variables are indicated by italics, the control variables by boldface.

 

3. Among children between the ages of 9 and 11, programming In Logo will cause substantial improvements in higher-order thinking skills among slow readers who are not classified as having learning disabilities. However, no such improvements will occur among 9- to 11-yearold slow readers who are classified as having learning disabilities.

 

4. Among children who have no external pressure from their parents to learn to read, delaying reading instruction until the sixth grade will have no adverse impact on reading ability by the time the children reach adolescence. This will be equally true among low-IQ, medium-IQ, and high-IQ students.

 

5. In middle-class schools, elementary school classrooms that emphasize a language-rich environment will have students who are more eager to share their opinions with their peers. However, in middle school -classrooms in similar schools, language-rich environments will have no such impact.

In the third hypothesis, there are two control variables: age of child (9 to 11 years old) and level of reading ability (slow readers are included, but average and fast readers are excluded). There are no 8 year-olds or above-average readers in the study that tests this research hypothesis. The factors of age and level of ability are ruled out as contaminating factors, but the researcher can no longer generalize to students who fall beyond the restrictions set by these control variables.

You have probably noticed that there is considerable similarity between moderator and control variables. This similarity is a result of both types taking into account factors that may influence the impact of the independent variable upon the dependent variable. The distinction is that moderator variables control this extraneous influence and examine it in such a way as to describe its precise impact. Control variables, on the other hand, merely reduce or describe this extraneous influence without providing any information about its relationship to the independent and dependent variables. Thus, while moderator variables have the potential to restrict, refine, and elaborate generalizations, the role of the control variable is usually confined to merely restricting generalizations.

Because of the similarity stated in the preceding paragraph, any variable that can be used as a moderator variable in an experimental study can also be used as a control variable, and vice versa. A sensible strategy is to identify as many extraneous variables as possible that might influence the impact of the treatment on the dependent variable; then, some of these can be made into moderator variables and some into control variables, and some can be left uncontrolled. At first thought, it might appear that since the moderator variable does everything the control variable does and more, then we should use the moderator variable as often as possible. This would be good advice, except for an important practical consideration. It is often difficult to obtain the number or variety of subjects that is needed for analyzing a factor as a moderator variable. In addition, having a large number of moderator variables can make a study cumbersome. Similarly, it might at first seem that no extraneous factor should be left uncontrolled. Again, however, there is a problem of feasibility. There are so many factors that could influence the impact of the treatment on the outcome that it is not possible to isolate and control all of them.

This discussion of control variables has focused on their use in experimental studies. This focus has been useful to point out their relationship to moderator variables. However, control variables are also important in descriptive research. For example, notice the impact of the control variables italicized in the following research questions and hypotheses:

 

1. Fifth graders in the public schools of Oregon do better in reading than in mathematics.
   
   
2. A survey of American businessmen showed that they would support higher taxes if they could be assured that the quality of education would actually improve.
   
   
3. The principals who responded to the survey indicated strong opposition to corporal punishment.
   
   
4. The students in the experimental school that used a whole-language approach showed a strong tendency to want to share their opinions with their peers.

 

Each of the italicized control variables shows the limitations imposed by the context of the study that will restrict the degree to which the results of each of these descriptive studies can be generalized. If you noticed that many of these control variables were too vague to be of much use, you're right. The usefulness of these control variables would improve with operational definitions (the topic of chapter 4) or qualitative description (the topic of chapter 9).

INTERVENING VARIABLES

Intervening variables are hypothetical concepts assumed to be created by the treatment and to have an impact on the observable outcome. Treatments produce intervening variables (which are invisible, perhaps inside the brain of the learner), and intervening variables then produce observable outcomes. This variable intervenes in the sense that the treatment does not produce the observable outcome directly but rather through the mediation (intervention) of this invisible, conceptual, hypothetical, internalized process. The intervening variable is the basic subject matter of theoretical research (discussed in chapter 17).

To take a brief example, a researcher might find that computer simulations cause students to do well on science tests. In fact, the computer simulations themselves probably do not cause the improved performance on the tests. The simulations probably induce some internal change within the students (such as improved motivation, greater attention to the task, perceived relevance, or something else), and this internal change is what eventually leads to the difference in performance. Determining and examining intervening variables helps us understand educational principles better. In addition, such variables can be useful for practical reasons: we may be able to find less expensive ways to bring about these intervening variables than, for example, relatively expensive computer simulations.

In the following examples the hypothesis is stated (without the moderator variables or control variables), and then a possible intervening variable is stated in parentheses:

1. Studying Shakespeare leads to a greater appreciation of Western culture. (perception of important similarities within Elizabethan and contemporary Western culture.
   
   
2. Behavior modification leads to noncreative behavior among elementary school children. (an urge to conform in order to be reinforced as efficiently as possible)
   
   Now you try it. Identify possible intervening variables for the following hypotheses:
   
   
3. Because Logo offers opportunities for improved planning ability arising out of the involvement in microworlds, programming in Logo will cause improvements in higher-order thinking.
   
   
4. Delaying reading instruction until the sixth grade will have no adverse impact on reading ability by the time the child reaches adolescence.
   
   
5. Language-rich environments will make children more eager to share their opinions with their peers.

 

Here are possible intervening variables for the preceding hypotheses:

1. Background information that makes it easier to understand aspects of Western culture
   
2. A tendency to refrain from doing things that do not gain extrinsic reinforcement
   
3. Improved planning ability arising out of the involvement in microworlds offered by the programming language
   
4. Self-motivation, which replaces external requirements
   
5. In this case, none is stated. A reader or researcher could construct one through such processes as reasoning about the evidence arising from the study and reflecting on theoretical information from the research literature. Through such reasoning, stimulation of language and social processes might emerge as the intervening variable.

In these examples, the intervening variable has been appended to the barest possible statement of the hypothesis. This has been done for the sake of simplicity. In actual practice, the moderator variable should take into account all the variables in the research hypothesis.

Unlike the other variables, the intervening variable is usually not stated as part of the hypothesis. Rather, it is stated at the culmination of the review of the literature (prior to- the hypothesis) as the specific rationale behind why the hypothesis is going to be stated in the form it will take.

EXTRANEOUS VARIABLES

An extraneous variable is a factor that produces an uncontrolled, unpredictable impact upon the dependent variable. Extraneous variables weaken research studies because they introduce ambiguity into the research process. To the extent that they are uncontrolled, they render uncertain the conclusions we can draw from a study. The threats to internal and external validity, discussed in chapters 10 and 1S, are extraneous variables, and the whole purpose of the strategies discussed in those chapters is to minimize the impact of such extraneous variables. To the extent that extraneous variables can be brought under control, the results of a study are strengthened. One good reason for replicating research (discussed later in this chapter) is to reduce the possibility that conclusions will be weakened by extraneous factors.

THE RESEARCH VARIABLES COMBINED

Table 2.5 summarizes the major characteristics of each of the research variables. Our discussion of research variables has focused largely on experimental research. This is because only in experimental studies are we likely to find all the research variables in a single hypothesis. We may have given the false impression that most educational research is experimental. Actually, only about 10% of published educational research is experimental; the rest is either descriptive, qualitative, or correlational. Table 2.1 briefly described these various types of research and indicated where each is discussed in this book. Also, the categories in that table are not mutually exclusive; for example, a quantitative or experimental study may also employ some qualitative methods, and a meta-analysis often combines the results of several experimental studies.

Most studies do not contain every type of variable. Whatever variables emerge from the analysis of a problem setting should be clearly identified and appropriately stated in a succinct research question or hypothesis. Subsequent chapters will describe how to operationally define these variables and incorporate them into effective research studies.

REVIEW QUIZ 2.2

1. Examine the following research hypothesis and identify each of the research variables requested below:

Formal operational high school biology students will use problem-solving heuristics to solve genetics problems, but concrete operational high school biology students will use a rote-memory strategy to solve the same kind of problems.

Identify the following variables:

 

Dependent

Control

 

2. Examine the following research question and identify each of the research variables requested below:

How do K-3 students who use the whole-language method to study reading differ in their comprehension strategies from those who use a phonetic approach?

Identify the following variables:

 

Independent

Dependent

Moderator

Control

 

3. Examine the following research hypothesis and identify each of the research variables requested below:

High school students who study Latin for two years will develop better English vocabulary skills than those who do not study Latin. This difference will occur both among those who are in advanced-placement English classes and among those who are in regular English classes. This difference will occur because of an increased ability to break a word down into its component parts.

Identify the following variables:

 

Independent

Dependent

Moderator

Control

Intervening

 Click here for answers.

 

SELECTING RESEARCH PROBLEMS FOR YOUR OWN STUDIES

Many readers of this book will develop research proposals and conduct research projects of their own. An important consideration for these readers is the selection and exploration of their research problem. The following guidelines will be helpful:

It is not necessary to invent problems in education. Education is overwhelmed with problems. Pick one that is within your interests, abilities, and resources. To select a research problem, focus proper attention on questions and problems that already exist and that can feasibly be solved. Follow the guidelines in this chapter to state and clarify a problem that is meaningful, interesting, feasible, and important to you.

1. State the research question or hypothesis early, but be ready to modify, adapt, or expand it as your reading of the literature and early observations give you more information pertinent to the topic. This early focus will enable you to direct your attention to the problems of the research more effectively.
   
   
2. Use the library and reference sources discussed in chapter 3 to find out what has already been done with regard to your problem. Using reference materials will enable you to discover other variables. as well as how others have operationally defined and measured their variables, how they have carried out their research studies, and how they fit their ideas and methods into a theoretical framework.
   
   
3. It's often worthwhile to replicate previous research. As later chapters will show, replications not only check the accuracy of their original counterparts but also lead to refinement and expansion of generalizations. A replication can be almost a direct duplication, or you can make modifications, as the next guideline suggests.
   
   
4. Consider modifying previous research by using new control variables, moderator variables, or operational definitions of any of the variables.
   
   
5. If possible, conduct a pilot study before undertaking the full research project. Pilot studies often enable researchers to refine, clarify, and try out their variables, measurement processes, and other research strategies.

    

Strategies for getting research ideas from existing studies are summarized in Table 2.6. By following these guidelines and employing the strategies discussed throughout this book, you can plan and carry out an effective research project. The preceding list offers guidelines to stimulate your thinking, not steps to be followed in a linear sequence. The basic strategy is to formulate an effective plan, to be willing to modify and adapt it, and to carry it out successfully.

PUTTING IT ALL TOGETHER

As we saw in the first chapter, Mr. Anderson, our humane educator, was interested in improving attitudes toward animal life. He did many forms o f research, including descriptive, qualitative, and correlational studies, which will be described in subsequent chapters. The present chapter will focus on an experimental study he eventually conducted, because that research includes examples of a wide range of variables and is perhaps simpler to interpret.

The attitudes of children toward animal life were the outcome variable with which Mr. Anderson was concerned, and they became the dependent (or criterion) variable in his experimental research study. At the beginning of his work, he had no clear idea of how he could accomplish his goal. Nevertheless, he stated a rough hypothesis:

A program that I'll design will lead to improved attitudes toward animal life.

In this rough statement, Mr. Anderson's independent (or treatment) variable was a program that he would develop. He spent time searching library references, talking to colleagues, and using his imagination to find a program that he could present to children to help them improve their attitudes toward animal life. At the same time, he tried to find effective ways to measure attitudes toward animal life. After doing some preliminary descriptive and qualitative research, he eventually developed an Animal Life Program (ALP), which was based on theoretical principles of induced dissonance to produce attitude change. Once he had decided to test the ALP, it became his independent variable.

Since he planned to use his program with fourth, fifth, and sixth graders, the grade level of the children became Mr. Anderson's control variable. He also realized that children who owned pets may react differently to the ALP than other children, so he decided to examine this possibility during his research. Therefore, pet ownership became a moderator variable. His intervening variable was induced dissonance-the underlying principle that was supposed to make the ALP work. At this point Mr. Anderson's hypothesis could be stated like this:

Among fourth through sixth graders, the Animal Life Program (ALP) will lead to improved attitudes toward animal life. The program will lead to greater improvements among persons who do not own pets than among those who do own pets. (This impact will result from induced dissonance produced by the ALP, which will lead the children to change their attitudes.)

Subsequent chapters will describe how Mr. Anderson carried out his experimental study.

By stating his research variables and hypothesis in rough form at an early point in the project, Mr. Anderson could use them as a guide in his selection of measurement devices and in his choice of a research design. The earliest formulation would probably be changed and refined several times as he encountered new ideas and problems. The final product would be a guide for him in conducting the experiment, analyzing the results, and drawing conclusions.

SUMMARY

The first step in the educational research process involves identifying and clarifying the problem. We can begin to do this by specifying the variables involved in that problem and stating these variables in unambiguous research questions or hypotheses. Research questions ask about the nature of a variable or concept or about the relationship among two or more variables. Research hypotheses state the expected answers to these research questions. Research questions and hypotheses give focus, structure, and organization to the collection, analysis, and interpretation of the data collected by -.he person conducting the research and also help develop theory with regard to the variables in the study.

In many educational research studies, the researcher is interested in measuring or describing an outcome. Descriptive studies focus on the measurement, description, and interpretation of context and outcome variables. In experimental studies, the researcher is interested not only in the outcome (dependent) variable but also in the causal nature of relationships among variables. In experimental studies the outcomes are referred to as the dependent or criterion variables and the treatment as the independent variable.

Correlational studies also enable the researcher to examine relationships among variables. However, in this case the researcher merely describes the nature of the relationships, without making statements about whether one variable may have caused another.

Moderator variables are factors that influence the impact of the independent variable on the dependent variable. Control variables describe extraneous factors and provide a means to help reduce the impact of extraneous factors on a study. In all studies, control variables help determine the degree to which results can be generalized, and in many studies they enable the researcher or reader to understand better the nature of the relationship between the other variables in the study. Intervening variables are hypothetical concepts that mediate between the independent and the dependent variables.

By clearly understanding and identifying these variables, researchers can strengthen the quality of their work. In addition, a thorough understanding of these variables and their interactions will enable consumers of research to make more effective use of research that is performed by others and then communicated to us in some fashion.

What Comes Next

The next few chapters will focus on ways to discover, operationally define, and measure research variables. The book will then focus on how to design and carry out studies that collect data and test research hypotheses, how to analyze the results of these studies, and how to evaluate the quality of research results.

DOING YOUR OWN RESEARCH

In the first edition of this book, this chapter came near the end of the book. A major reason for moving it to the front of the book is to help readers begin to conduct a research study of their own. If you need to select a research topic and begin planning a research project, here are some strategies you can employ right now:

1. You can choose a problem that interests you, find a study that has already been done, and directly replicate that study. As later chapters will show, replication is an important part of the educational research process. Strategies for carrying out the replication will be discussed later in the book.
   
   
2. You can find a study that has already been done and replicate it using new operational definitions of at least some of the variables in the study. Strategies for operationally defining variables will be discussed in chapter 4.
   
   
3. You can find a study that has already been done and replicate it using additional moderator or control variables. As chapter 15 will show, replication with different moderator and control variables plays an important role in determining how far the results of educational research can be generalized.
   
   
4. You can find an existing study and develop a different way to test the same hypothesis.
   
   
5. You can choose a problem, analyze it as discussed in this chapter, and invent a completely new study of your own.
   
   
6. You can employ a combination of the preceding five strategies.

 

To actually carry out a study, you will need certain information that is presented later in this book. Right now, you can select a topic, determine what the variables are, and begin to think about what

problems you are likely to encounter in conducting your own study. Be aware, however, that as you read subsequent chapters in this book or begin to read the literature related to your topic, You may change your mind. That's OK. What is important is that you eventually develop a good plan and carry it out.

Chapter 17 integrates the methods and principles throughout this book; chapter 19 describes appropriate strategies for carrying out and reporting a formal research project. if you read this book in one course and later write a research paper in a subsequent Course, then you should use those chapters-rather than the preceding list-as your primary source of guidelines for carrying out that project.

A few words of caution: Some studies may sound interesting, but they may involve the use of advanced methods which will not be covered in this book. Therefore, there is a possibility that you will find an interesting hypothesis and discover (weeks later) that you are not going to be able to test that hypothesis with strategies covered in this book. The best way to deal with such a problem is to confer with your instructor and modify your hypothesis as you move on to appropriate chapters in this book.

FOR FURTHER THOUGHT

1. Complete this sentence by answering the designated questions: "The independent variable... (does what?) (To what or whom?) (When?) (Where?) (How?) (Why?)"

2. Complete this sentence by answering the designated questions: "The dependent variable... (does what?) (To what or whom?) (When?) (Where?) (How?) (Why?)"

3. Complete this sentence by answering the designated questions: "The moderator variable... (does what?) (To what or whom?) (When?) (Where?) (How?) (Why?)"

4. Complete this sentence by answering the designated questions: "The control variable... (does what?) (To what or whom?) (When?) (Where?) (How?) (Why?)"

5. Complete this sentence by answering the designated questions: "The independent variable... (does what?) (To what or whom?) (When?) (Where?) (How?) (Why?)"

6. Complete this sentence by answering the designated questions: "The research hypothesis... (does what?) (To what or whom?) (When?) (Where?) (How?) (Why?)"

1. Complete this sentence by answering the designated questions: "The research prediction... (does what?) (To what or whom?) (When?) (Where?) (How?) (Why?)"

ANNOTATED BIBLIOGRAPHY

Martin, D. V. (1991). Doing psychology experiments (3rd ed.) Monterey, CA: Brooks/Cole. Chapter 4 provides good guidelines on how to select variables for formal research projects.

Smith, M. L., & Glass, G. V. (1987). Research and evaluation in education and the social sciences. Englewood Cliffs, NJ: Prentice-Hall. This up-to-date book gives good illustrations and different points of view regarding the research process.

Tuckman, B. W. (1988). Conducting educational research . (3rd ed.) New York: Harcourt Brace Jovanovich. Chapter 4 discusses the formulation of hypotheses, and chapter 5 focuses on identifying and labeling variables.

ANSWERS TO QUIZZES

Review Quiz 2.1

l. Dependent variable: appreciation of Western culture
Independent variable: studying Shakespeare

2. Dependent variable: noncreative behavior
Independent variable: behavior modification

3. Dependent variable: higher-order thinking skills
Independent variable: programming in Logo

4. Dependent variable: reading ability during adolescence
Independent variable: delaying reading instruction until the sixth grade

5. Dependent variable: eagerness to share opinions with peers
Independent variable: language-rich environment

Review Quiz 2.2

1.Dependent: strategies (heuristic versus rote)
Control: High school biology students

2. Independent: reading method
Dependent: comprehension strategies
Moderator: (none stated)
Control: K-3 students
Intervening: (none stated)

3. Independent: studying Latin (versus not studying it)
Dependent: vocabulary skills
Moderator: English class placement (advanced placement versus regular)
Control: high school students
Intervening: increased ability to break a word into its component parts

RESEARCH REPORT ANALYSIS

The sample research report in Appendix C deals with computer simulations that are used to stimulate scientific problem solving. It provides evidence to support the theory that guided use of computer simulations gives students opportunities to practice thinking skills and to receive reinforcement for successful use of these skills. This reinforced practice leads to improved problem-solving skills. Try answering the following questions:

1. What is the hypothesis of this study?

2. What is the independent variable?

3. What is the dependent variable?

4. What is the moderator variable?

5. What is the control variable?

6. What is the intervening variable?

ANSWERS:

1. Guided use of computer simulations will cause students to develop problem-solving skills to a greater degree than either unguided use of computer simulations or the use of noncomputerized approaches to attain the same objectives. (This hypothesis is briefly stated on page 458 and is clarified on page 459.)
   
   
2. The independent variable is the way students study the instructional material (guided use of simulations versus unguided use of simulations versus noncomputerized approaches to the same objectives). (This variable is described on page 460 of the report.)
   
   
3. The dependent variable is problem-solving ability. (This variable is described on page 462 of the report.)
   
   
4. The moderator variable is gender of students. (Th is variable is described on page 463 of the report.)
   
5. The control variable is the type of students in the study. The subject matter content could also be considered a control variable. (This variable is described on page 459 of the report.)
   
6. The intervening variable is reinforced practice for the effective use of thinking skills. (This variable is described on page 464 of the report.)