INTRODUCTION TO SOME BASIC ISSUES CONCERNING RESEARCH METHODS IN PSYCHOLOGY Variables and Constants Much of the scientific study of psychology involves the investigation of what are called variables. Unfortunately, variables are a bit difficult to define, without using the word variable. One way to approach this is to contrast a variable with its opposite, a constant. For our purposes, a constant is some characteristic that does not vary among the members of a group we wish to study. A variable is a characteristic that does vary among members of that group. That characteristic can be a behavior or some psychological characteristic such as intelligence, friendliness or preference for music. A situation may also be a variable because some people may be in a situation that others are not. I do not like dictionary definitions but if I were to attempt one for the word variable, it would be something like: a behavior, psychological characteristic, or situation that is not the same for all members of a group who are being studied. Among members of a PSY 101 class we might be interested in investigating Grade Point Average. GPA is a variable because it is a characteristic that varies among members of the class. A typical question a psychologist might ask is why some students have higher GPAs than others. To answer that question, the psychologist would probably measure GPA in some way, measure the levels of one or more other variables, and compare the outcomes. It should be easy for you to imagine some other variables that might have some relationship to GPA. Some examples are: time spent studying, time spent working at outside jobs, chronological age, class level, high school GPA. Any of these, and many more that might occur to you, could be part of the explanation for GPA differences within members of a class. All of these are variables and any of them could be investigated. When you get to class I will ask you for some more variables --however far-fetched--that might have a relationship to GPA. In contrast, no useful information would result from investigating the effect of a constant upon GPA. For example, all members of the class are breathing, so an investigation of the effect of breathing or not upon GPA would be useless. All members of the class are wearing clothes so there would be no way to use this group to investigate the effects of nakedness upon GPA. If this discussion of constants seems strange, you should remember that the primary reason for introducing the concept of the constant here is to help to define variable. Replication In scientific accounts of research, descriptions of variables must be sufficiently specific so that another researcher could re-do the study using the same variables. It is very basic to science that the events studied by one researcher could also be studied by another scientist in another time and place. This is called replication. Confirmation of findings through replication is an important way in which the power, or certainty, of scientific findings can be increased. Specificity of Variables If you look at the variables that might help to explain GPA listed above, you will notice that they are quite specific. Science can only deal with specific, observable events. Even though the examples given above are quite specific, they could be even more so. For example, take the variable “time spent working at outside jobs.” Try to imagine some ways in which that variable could be made so specific that two people in different colleges could measure it in PSY 101 classes and have confidence that the same thing is being measured. Does volunteer work count? This variable could be made more specific by saying “time spent working at outside jobs for money”. Alternatively, the variable could be restated “time spent working at outside jobs whether volunteering or being paid.” Either way, the variable is more likely to be measuring the same thing if it were to be used by different researchers. You should be able to find other ways to make this variable more specific and I will ask for some when you get to class. Measurement of Variables Another issue surrounding the definitions of variables is the question of measurement. Even if the variable itself is defined quite specifically, it is also important to be specific about the way in which it will be measured. For example, in the investigation of GPA and other variables, it is important to know how GPA will be measured. The students could be asked to self-report their GPA to two decimal places. Alternatively they could be asked to self-report it within ranges involving no decimals, i.e. 0 to 1, 1 to 2, 2 to 3 and 3 to 4. In some circumstances it might be possible to get lists of GPAs from the college registrar. These three tactics would almost certainly yield different results and so it is important to be specific about the measurement of the variable if we want to be sure that studies of GPA at two different colleges were measuring the same thing. Measurement of GPA is easy compared to the other variables, listed above, that might be studied for their effect upon GPA. Look at some of them and come to class prepared to suggest ways in which they could be measured. Operational definitions In summary, scientific research reports define variables specifically and in terms of measurement. These definitions are called operational definitions, because the variable is defined in terms of the operations needed to measure it. The primary goal of this course is to teach you to critically evaluate assertions about behavior and one of the most basic ways to be critical is to ask about the operational definitions of variables. Operational definitions are particularly important when discussing difficult-to-observe psychological characteristics such as being depressed, being smart or being sensitive. Someone might say, for example, that a new study had shown that first-born children are more likely to be withdrawn whereas second-born children are more likely to be outgoing. When you hear something like that you can begin to evaluate the assertion by asking for the operational definitions. When you come to class, I am going to ask you to make some suggestions about how these things might be operationally defined and I will also ask you to consider potential pitfalls in the operational definitions that might have been used in this study of birth order and personality. One reason why an understanding of operational definitions is such an important critical tool is that operational definitions can vary considerably in quality. If you were doing a study of depression and air temperature, a thermometer might provide a good operational definition of air temperature. Alternatively, you could open the window and stick your head out and make a guess based on this experience. During this course we will often evaluate the adequacy of operational definitions. A Short Guide to Some Common Types of Research Methods There are many different types of research methods, also called research designs, that are used by psychologists in trying to find things out about behavior. This is just a quick aid to the identification of research designs. In real life, some studies may combine the features of several research designs or may contain elements not included below. Experiment: Participants randomly assigned to different groups being studied. Groups are treated differently in one or a few very specific ways--the independent variable. Behavior resulting from this treatment difference is measured--the dependent variable. If one group gets a specific treatment and ones does not, usually the treated group is called the experimental group and other groups are called control groups. Conditions other than the independent variable are held as constant as possible for all groups. These constant conditions are called controls. If participants are their own control group, that is, they receive both research treatments, the design is called a within-subjects experiment. Conclusions can be taken to indicate a cause and effect relationship between the independent and dependent variables. Because of this, the experiment is in a class by itself and it is a very special type of research procedure. Quasi-experiment: Participants achieve membership in different groups as a result of characteristics other than random assignment, for example: gender, age, socioeconomic status, athletic ability, or ethnic identification. A link may be found between one or more of these characteristics and some outcome variables, but cause and effect relationships are not clearly identified. Without random assignment to groups, a researcher cannot clearly demonstrate cause. Correlational study: In the most general sense, a correlational study investigates the relationship between two variables. Usually the data are reported as correlation coefficients. Strength and direction (positive or negative) of relationships can be demonstrated by correlational studies but causal links remain an open question. Longitudinal study: A longitudinal study follows a group composed of the same people across a period of the life span. The behavior of these individuals is observed and/or measured at several intervals over time in an attempt to study the changes in their behavior. Longitudinal studies may cover a short time, such as a few weeks, or a long time, such as the entire life span. Longitudinal studies may additionally employ other methods, such as quasi-experimental or correlational approaches, but the defining characteristic is that the same people are studied repeatedly across time. Cross sectional study: A cross sectional study usually examines groups of different people who belong to different age groups as a means of studying behavior development across part or all of the life span. These studies can usually be done more easily and quickly than longitudinal studies but the resulting data may be of lower quality. More rarely, the term cross sectional may be used to describe studies which divide and examine segments of society based on variables other than age, such as income, educational level or family size. Survey: A survey is a structured list of questions presented to people. Surveys may be written or oral, face to face or over the phone. It is possible to cheaply survey large numbers of people, but the data quality may be lower than some other methods because people do not always answer questions accurately. Interview: An interview may be highly structured or it may involve less structured narrative. It may include survey methodology. It usually involves people responding orally to questions or talking about their thoughts on a topic. Case study: A case study involves extensive observations of a few individuals. Data collection may include watching behavior, interviews and record searching. Case studies may be retrospective and/or prospective. Usually case studies are employed where the behavior or situation is so rare that other methods, involving larger groups of participants, are not possible. Naturalistic observation: Naturalistic observations can range from unstructured observations of humans or other animals to situations involving hypothesis testing or some manipulations of a natural setting. If you wanted to know if males are likely to hold doors open for females, you could watch until you had seen a number of natural occurrences of this, or you could get a female helper to follow males into buildings and watch to see what happens. It can be difficult to precisely define the natural setting, particularly when the participants are humans. Placing an actual research procedure into this category or others can involve a judgment call which might be debatable. Demonstration: An unsystematically engineered observation of behavior, sometimes involving only one participant. The demonstration is remarkably common in the history of psychology, even though it provides only very weak evidence. It is not a recognized research method but it is a term which can be quite useful as a descriptor for studies that seem to employ no established method.