General Designs for Research
Having selected the way we want to measure the topic of interest, researchers must embed this measure in a research design that yields useful, relevant results. Gerontologists rely on primary designs in planning their work: experimental studies, correlational studies, and case studies. The specific design chosen for research depends in large part on the questions the researchers are trying to address.
Experimental Design. To find out whether Leah’s or Sarah’s approach to remembering works better, we could gather groups of older adults and try the following. We could randomly assign the participants into three groups: those who are taught to use imagery, those who are taught to use lists, and those who are not taught to use anything. After giving all the groups time to learn the new technique (where appropriate), we could test each group on a new grocery list to see who does better.
What we have done is an example of an experiment, which involves manipulating a key factor that the researcher believes is responsible for a particular behavior and randomly assigning participants to the experimental and control groups. In our case, the key variable being manipulated (termed the independent variable) is the instructions for how to study. In a study of memory, a typical behavior that is observed (termed the dependent variable) is the amount of information actually remembered.
More generally, in an experiment the researcher is most interested in identifying differences between groups of people. One group, the experimental group, receives the manipulation; another group, the control group, does not. This sets up a situation in which the level of the key variable of interest differs across groups. In addition, the investigator exerts precise control over all important aspects of the study, including the variable of interest, the setting, and the participants. Because the key variable is systematically manipulated in an experiment, researchers can infer cause-and-effect relations about that variable. In our example, we can conclude that type of instruction (how people study) causes better or worse performance on a memory test. Discovering such cause-and-effect relations is important if we are to understand the underlying processes of adult development and aging.
Finally, we must note that age cannot be an independent variable, because we cannot manipulate it. Consequently, we cannot conduct true experiments to examine the effects of age on a particular person’s behavior. At best, we can find age-related effects of an independent variable on dependent variables.
Correlational Design. In a correlational study, investigators examine relations between variables as they exist naturally in the world. In the simplest correlational study, a researcher measures two variables, and then sees how they are related. Suppose we wanted to know whether the amount of time spent studying a grocery list such as one that Sarah might create was related to how many items people remember at the store. To find out, the researcher would measure two things for each person in the study: the length of study time and the number of items purchased correctly.
The results of a correlational study usually are measured by computing a correlation coefficient, abbreviated r. Correlations can range from -1.0 to 1.0, reflecting three different types of relations between study time and number of groceries remembered.
• When r = 0, the two variables are unrelated: Study time has no relation to remembering groceries.
When r > 0, the variables are positively related: As study time increases (or decreases), the number
of grocery items remembered also increases (or decreases).
• When r < 0, the variables are inversely related:
When study time increases (or decreases), the number of groceries remembered decreases (or increases).
Correlational studies do not give definitive information about cause-and-effect relations; for example, the correlation between study time and the number of groceries remembered does not mean that one variable caused the other, regardless of how large the relation was. However, correlational studies do provide important information about the strength of the relation between variables, which is reflected in the absolute value of the correlation coefficient. Moreover, because developmental researchers are interested in how variables are related to factors that are very difficult, if not impossible, to manipulate, correlational techniques are used a great deal. In fact, most developmental research is correlational at some level because age cannot be manipulated within an individual. This means we can describe a great many developmental phenomena, but we cannot explain very many of them.
Case Studies. Sometimes researchers cannot obtain measures directly from people and are able only to watch them carefully. In certain situations, researchers may be able to study a single individual in great detail in a case study. This technique is especially useful when researchers want to investigate very rare phenomena, such as uncommon diseases or people with extremely high ability. Identifying new diseases, for example, begins with a case study of one individual who has a pattern of symptoms that is different from any known syndrome. Case studies are also very valuable for opening new areas of study, which can be followed by larger studies using other methods (e. g., experiments). However, their primary limitation is figuring out whether the information gleaned from one individual holds for others as well.