Defining Variables and Formulating Hypotheses
The purpose of a research study is to discover unknown qualities of persons or things. To measure these qualities we define variables. In a study there are several classes of variables.
1. Independent (or experimental) variable: There are two types of independent variables: Active and attribute. If the independent variable is an active variable then we manipulate the values of the variable to study its affect on another variable. In the above example, we alter anxiety level to see if responsiveness to pain reduction medication is enhanced. Anxiety level is the active independent variable. An attribute variable is a variable where we do not alter the variable during the study. For example, we might want to study the effect of age on weight. We cannot change a person's age, but we can study people of different ages and weights.
2. Dependent variable (or Criterion measure): This is the variable that is affected by the independent variable. Responsiveness to pain reduction medication is the dependent variable in the above example. The dependent variable is dependent on the independent variable. Another example: If I praise you, you will probably feel good, but if I am critical of you, you will probably feel angry. My response to you is the independent variable, and your response to me is the dependent variable, because what I say influences how you respond.
3. Control variable: A control variable is a variable that effects the dependent variable. When we "control a variable" we wish to balance its effect across subjects and groups so that we can ignore it, and just study the relationship between the independent and the dependent variables. You control for a variable by holding it constant, e.g., keep humidity the same, and vary temperature, to study comfort levels.
4. Extraneous variable: This is a variable that probably does influence the relationship between the independent and dependent variables, but it is one that we do not control or manipulate. For example, barometric pressure may effect pain thresholds in some clients but we do not know how this operates or how to control for it. Thus, we note that this variable might effect our results, and then ignore it. Often research studies do not find evidence to support the hypotheses because of unnoticed extraneous variables that influenced the results. Extraneous variables which influence the study in a negative manner are often called confounding variables.
Variables must be defined in terms of measurable behaviors. The operational definition of a variable describes the variable. There are two ways by which we can operationally define a variable; by how it is measured and by how it is used to classify subjects. Later we will use specialized terms for how variables are defined (continuous or categorical) and the nature of the data obtained (nominal, ordinal, or interval). These terms are discussed in Chapters 6 and 7.
The first way of defining a variable is to describe how we measure it. We cannot just say we will "reduce anxiety." We must define how anxiety will be measured and just what is a reduction in anxiety. In our example we define anxiety as a change in galvanic skin response generated by the discussion of potentially emotional content (i.e., one's pending death). Another example would be range of motion. In general range of motion deals with the amount of mobility one has of their limbs. Actually, we define ROM as the movement of a specific limb through so many degrees as measured by a goniometer with the limb held in a specific way and moved in a prescribed arc.
The second way of defining a variable is to describe how you have classified subjects (people) into groups or categories. This is important since two researchers could be studying the same variable but if they each classify their subjects differently they may get different results. For example, suppose we wanted to study the income levels of single adults. If one researcher classified his single adult subjects into these three categories (17 through 22, 23 through 27, and 28 through 33), he would get different results than this second researcher who used three different categories (20 through 40, 41 through 60, and 61 and over). The first researcher is interested in young adults and the second in all ages. Thus, without operational definitions we could think that they both were studying the same variable.
When we use behavioral (operational) definitions for variables, we define exactly what we are studying and enable others to understand our work. This is called operationalism.
Once the research question has been stated, the next step is to define testable hypotheses. Usually a research question is a broad statement, that is not directly measurable by a research study. The research question needs to be broken down into smaller units, called hypotheses, that can be studied. A hypothesis is a statement that expresses the probable relationship between variables.
There are two types of hypotheses: descriptive and directional. Descriptive hypotheses ask a specific question regarding some phenomenon. For example, we might want to study this research question: what are the social and economic characteristics of patients who have high blood pressure? A descriptive hypotheses that would test a part of the above research question is: what is the distribution of hypertensive patients by income level? Descriptive hypotheses are always phrased in the form of a question regarding some aspect of the research question. Usually a descriptive hypothesis does not include an active independent variable. When we use an independent variable, a directional hypothesis is usually needed.
The second type of hypothesis is a directional hypothesis. Directional hypotheses are never phrased as a question, but always as a statement. Directional hypotheses always express the effect of an independent on a dependent variable. For example, hypotheses drawn from the anxiety and dying research question above would be:
1. A client who is at the "acceptance" stage will exhibit less anxiety, as measured by GSR recordings, when discussing their pending death than clients in the other stages.
IV - stage of client (attribute)
DV - anxiety level
2. Client anxiety levels, as measured by GSR recordings, will be lower at the end of any state (denial, anger, bargaining, depression, acceptance) than at the beginning.
IV - client at beginning, middle, or end of stage (attribute)
DV - anxiety level
3. Some clients will reach their lowest anxiety levels, as measured by GSR recordings, after progressing through only one stage. Others will require two or more stages before achieving their lowest anxiety levels.
IV - number of stages client has gone through since learning of their condition (attribute)
DV - anxiety level
4. A planned program of counseling interventions will enable clients to achieve low anxiety levels more rapidly than clients who receive normal nursing and medical care.
IV - presence or absence of counseling (active)
DV - time taken to reach a low anxiety level
5. Clients who achieve and maintain low levels of anxiety when discussing their pending death will be more responsive to pain medication.
IV - anxiety level (attribute)
DV - effectiveness of pain medication
Hypotheses also are as specific as possible and deal in behaviors rather than attitudes or general feelings. Any time a non-behavioral term, such as anxiety or happiness, is used, the researcher needs to define that term as behaviors that can be measured or observed. Also, the hypothesis needs to include the conditions under which the behaviors will be seen. In the above hypotheses we planned to study anxiety levels, as measured by GSR recordings under the conditions of the client discussing his pending death. We do not hypothesize if the client's anxiety will be lower at other times.
The above hypotheses express the relationships between several variables. The independent variables are: 1) stage client is in, 2) stages client has passed through, 3) counseling vs. normal nursing care, and 4) anxiety level. Dependent variables are: 1) anxiety levels, 2) time, and 3) responsiveness to pain medication.
Suppose we were interested in the influence of environmental variables on a person's typing accuracy. The research question is: What is the effect of environmental variables on typing accuracy? We could then operationally define typing accuracy as the number of errors made while typing for two minutes at a rate of at least 40 w.p.m. For environmental variables let us use room temperature and humidity. These can be operationally defined as follows:
Temperature - Amount of warmth in the air. Measured with a mercury thermometer using the Fahrenheit scale.
Humidity - Amount of water vapor in the air. Measured with a wet bulb thermometer and stated as the percent of saturation.
In this study we will vary the temperature and humidity and see the effect those changes have on typing accuracy. Thus, temperature and humidity are the independent variables and typing accuracy is the dependent variable. For several of the hypotheses below we will NOT vary humidity, but will hold it constant. In those cases when humidity does not vary, it is a control variable instead of an independent variable.
Hypothesis 1: Given that the humidity of the room remains constant at 40% of saturation, as room temperature is increased from 65 degrees to 95 degrees, in increments of 5 degrees, the number of typing errors will increase.
Hypothesis 2: Subjects will make more typing errors when the room is 90 degrees than when the room is 70 degrees, holding humidity constant at 40%.
Hypothesis 3: Subjects will make more typing errors when the temperature is 90 degrees and the humidity is 80%, than when the temperature is 90 degrees and the humidity is 40%.
We have made humidity the independent variable, and by holding temperature constant it is a control variable. Now another example.
Hypothesis 4: As temperature and humidity increase from 65 degrees to 90 degrees, and 40% to 90% respectively, the number of typing errors will increase.
Temperature and humidity are now both used as independent variables. In all four hypotheses the independent variables are active since the researcher manipulates them.
Research Question and Hypotheses
Often readers wonder if there is a difference between the Research Question (RQ) and a Hypothesis. Usually the RQ is a more general statement than a hypothesis. A RQ can usually be divided, through careful definition of variables, into several hypotheses. Let's look at an example.
In Chapter 2, we wrote this Topic Sentence (a one sentence statement of the RQ): "As a person gets older the amount of their body that is muscle mass decreases."
Now, let's define each of the variables:
Now, what are the effects we might see? Will the decrease in muscle mass be linear, e.g., will it decrease in gradual, equal steps from age 20 through 80? Will the muscle mass loss accelerate as the person passes 60? Might other variables affect muscle mass? Do we need to add exercise level and diet as controlling variables?
Finally, here are several hypotheses that came to mind after thinking about the above elements.