© 2004, 2007 Mark L. Mitchell & Janina M. Jolley. All rights reserved.
First, theories tend to be more internally consistent than common sense. That is, a theory usually doesn’t contradict itself. Common sense, on the other hand, often contradicts itself (“absence makes the heart grow fonder,” but “out of sight, out of mind”). Researchers find it easier to make clear, consistent predictions from a consistent theory rather than from inconsistent common sense.
Second, theories tend to be more consistent with existing facts than common sense. Often, theories are constructed by systematically collecting data and carefully analyzing the data for patterns. But even when facts do not play a dominant role in giving birth to a theory, facts will usually shape the theory’s development. Generally, if deductions from a theory are incorrect, the theory will be changed or abandoned. Thus, unlike common sense, theories do not ignore facts. Consequently, a hypothesis based on an established theory is a more educated guess and should have a greater chance of being correct than one based on common sense.
Third, theories are not restricted to making commonsense or intuitively obvious predictions. Theories can make predictions that are counter-intuitive. For example, social learning theory predicts that rewarding a child for a behavior could make the child like doing the behavior less (because the child may decide that he or she does the behavior because of the reward, rather than because the child likes it). Because theories are not limited to making predictions that are consistent with common sense, a theory may suggest controversial, new ways of viewing the world. For instance, Darwin’s theory on evolution had us look at apes as relatives, Einstein’s theory of relativity had us look at matter and energy as being the same thing, Freud’s theory had us look at ourselves as being motivated by forces of which we weren’t aware, and Watson’s theory had us look at ourselves as a set of reflexes.
Fourth, theories summarize and organize a great deal of information. Just as the plot of a movie may connect thousands of otherwise unrelated images, theories connect individual facts and give them meaning. That is, theories try to explain facts. The ability of theories to connect facts means that theory-based research will not produce isolated bits of trivia. Instead, the findings will fit into a framework that connects many other studies. In other words, the facts revealed by theory-based research are not merely of interest for their own sake, but also for how they relate to the theory’s explanation of how the world works. For example, consider the following fact: around age 7, children stop believing in Santa Claus. In its own right, this is a relatively trivial fact. However, when put in the context of Piaget’s theory, which states that around age 7, children are able to think logically about concrete events (and thus realize that Santa Claus can’t be everywhere at once and can’t carry that many toys), the finding has deeper significance.
Fifth, in addition to giving individual facts a meaningful context, theories focus research. Because many researchers try to test theories, findings from theory-based research are not only relevant to the theory’s explanation of events, but also to the findings of other researchers. Because progress in science comes from researchers building on each other’s work, the importance of a theory’s ability to coordinate individual scientists’ efforts should not be underestimated.
Sixth, theories are often broad in scope. Because theories can be applied to a wide range of situations, researchers can generate a wide variety of studies from a single theory. For example, social learning theory can be applied to prisons, businesses, advertising, politics, schizophrenics, smokers, librarians, mad dogs, and Englishmen. Similarly, Freud’s theory of the unconscious can be applied to virtually any situation.
Seventh, theories try to explain the facts with only a few core ideas. That is, they tend to be parsimonious: explaining a broad range of phenomena with a few principles. The value of parsimony is evident when you consider that a major function of science is to simplify our world. The parsimonious theory provides a few simple rules that summarize hundreds of observations. These general rules making existing knowledge easier to understand, remember, and use. Therefore, scientists prefer theories with a few far-reaching principles to theories that require a different principle to explain each new phenomenon. Thus, it should be no surprise that two theories that have enjoyed great popularity--evolutionary theory and social learning theory--possess only a few, broad-ranging principles.
Finally, theories are often more testable than common sense. That is, by talking about variables that can be objectively measured and by making specific predictions, a good theory is easy to test.
Despite their similarities, all theories are not equally good. Some are more parsimonious than others, some are broader than others, some are more logically consistent than others, some make more interesting predictions than others, and some are more consistent with the facts than others. However, if you are trying to develop a research hypothesis, the most important difference between theories is that some theories are more testable--and thus more useful--than others. Therefore, when choosing a theory, make sure that it is testable.
To be testable, a theory must:
1. make predictions rather than rely entirely on after-the-fact explanations;
2. predict one outcome rather than several contradictory outcomes;
3. make a specific prediction, rather than an extremely vague one; and
4. make a prediction that can be verified through objective observation.
To be testable, a theory must tell you about events that have not yet been observed. Unfortunately, not all theories make such predictions. Instead, some, such as McDougall’s (1908) instinct theory, only explain what happened after the fact. For example, after a woman picked apples from her orchard, McDougall might say, “she picked apples because the instinct to pick apples from an orchard was activated.” However,
McDougall’s theory could not make before-the-fact predictions because his theory didn’t tell us when to expect instincts to be aroused or how to tell whether someone would inherit a high level of an instinct.
To be testable, a theory must be capable of making one and only one prediction about what would happen in a certain situation. To illustrate the problem of making more than one prediction, consider Freudian theory. According to Freudian theory, receiving a severe beating from one’s father could result in any of the following outcomes:
1. no apparent effect (we try not to think about it: repression or
2. deep anger and resentment at people similar to our father
3. great love for our father (reaction formation); or
4. hate for ourselves (internalization).
Given all these predictions, it is hard to imagine an outcome that would not agree with one of them. Freudian theory would be more testable if it made one prediction.
Almost as useless as making many predictions about what would happen in a certain situation is making one extremely vague prediction. Some theories purport to make predictions about the future, but these predictions are so vague that they are untestable. An extremely vague prediction may remind us of the fortune cookies that read, “You will make a decision soon.”
Precision is the reason we often like to see quantitative statements in theories. For instance, the statement, “People taking drug A will remember twice as much as those not taking A,” is more precise than the statement, “People taking drug A will remember more than those not taking A.”
Even if a theory makes specific, unambiguous predictions about the future, these predictions must involve publicly observable events--if the theory is to be testable. That is, for the relevant variables, we must be able to provide operational definitions: publicly observable sets of procedures (operations) to manipulate or measure variables.
To illustrate the importance of operational definitions, consider the statement: “When you die, you will go to heaven.” Although this is a prediction about future events, it cannot be scientifically tested because we cannot find any publicly observable, physical evidence that would help us determine whether a person has gone to heaven. Since religion makes such metaphysical (beyond the physical world) statements, science and religion usually do not mix. Analogously, a few scientists have argued that science and psychoanalysis do not mix because we cannot observe the unconscious. Such is the fate of theories whose variables cannot be operationalized.
We should caution, however, that not all variables in a theory must be directly observable. Many theories discuss hypothetical constructs: entities that we cannot, with our present technology, observe directly. Gravity, electrons, love, learning, and memory are all hypothetical constructs because they are invisible. Although hypothetical constructs can’t be seen, we may be able to infer their presence from their traces or impact. With enough indirect, physical evidence, scientists can make a very convincing case for the existence of an invisible entity (a hypothetical construct). Thus, although no one has ever seen a quark, physicists have demonstrated that quarks exist.
In psychology, the challenge has not been to see inside the atom, but to see inside the head. Like quarks, mental states cannot be directly observed. For example, we cannot directly observe learning. However, we can see its effect on performance. That is, we can operationally define learning as an increase in performance. Thus, if we see someone improve their performance after practicing a task, we would conclude that learning has occurred. Similarly, we can provide operational definitions for such intangible hypothetical constructs as hunger, thirst, mood, love, etc.
You now know how to judge whether a theory can help you generate research ideas. But where do you find theories?
To find a useful theory, start by reading textbook summaries of theories. Reading a textbook summary should at least acquaint you with some of the theory’s propositions (to go beyond your text's brief explanation of dissonance theory, you can go to the article that launched dissonance theory ,or, to get an online summary of the theory's relevance to modern life, try this link or this one) . Although these summaries will allow you to select a theory, do not rely exclusively on textbook summaries--such summaries may oversimplify the theory. Thus, the researcher who relies exclusively on textbook summaries may be accused of ignoring key propositions of the theory or of using a straw theory: an exaggerated, oversimplified caricature of the theory. Therefore, in addition to reading textbook summaries, you should also see how other researchers have summarized the theory. To find these summaries, consult journal articles that describe studies based on the theory (e.g., “Elation and Depression: A test of opponent process theory”). Usually, the beginnings of these articles include a brief description of the theory that the study tests.
Once you have selected a theory, read the original statement of the theory (the citation will be in the texts or articles that you read). Then, to keep up-to-date about changes in the theory, use Psychological Abstracts to find books and review articles devoted to the theory (see Appendix B).
Once you understand the theory, your task is to apply your powers of deduction. You have these powers or you wouldn’t have passed high school geometry and you wouldn’t be able to write an essay. In fact, much of your everyday thinking involves deductive logic. For example, you may say, “The important thing about a college education is to learn how to think. This assignment doesn’t help me learn how to think. Therefore, this assignment is not important to my college education.” If your premises were sound, your statement would be an example of sound, deductive logic.
In deducing hypotheses from theory, you will use the same deductive logic illustrated above. That is, you will apply a general rule to a specific instance. The only difference is that the general rule comes from a theory instead of from the top of your head. To reassure yourself that you can apply deductive reasoning to propositions that were made up by someone else, try this deductive reasoning test:
1. All people treated like b turn out c.
2. Person a is being treated like b.
3. Person a will turn out ___________.
1. All behavior can be changed by controlling its consequences.
2. Al’s behavior is bad.
3. Al’s bad behavior can be changed by ___________.
As this “test” illustrates, if you know the premises and set them up correctly, deductive logic can be as simple as 1-2-3. Thus, because you know common sense’s premises, you probably had no problem deducing research ideas from common sense. Consequently, once you know what a theory’s premises are, your problem will not be how to think deductively, but what to think about. In the next few pages, we will give you some strategies that will help you focus your deductive reasoning.
Contrary to common stereotypes about theories, theories can be applied to practical situations. As Kurt Lewin said, “there is nothing so practical as a good theory.” For example, social learning theory has been used to cure shyness, promote energy conservation, address speech problems, reduce violence, and improve studying behavior.
To take a closer look at how theory can help you attack a practical problem, consider cognitive dissonance theory (Festinger, 1957). According to cognitive dissonance theory, if a person holds two thoughts that he considers inconsistent, he will experience dissonance (see Table T-2). Since dissonance is unpleasant, the person will try to reduce it, much as the person would try to reduce hunger, thirst, or anxiety (Aronson, 1990).
Let’s see how dissonance theory was used to get people to conserve energy. After being told they would get their names in the paper if they conserved energy, people cut back on their energy use. Then, dissonance was induced by telling them that their names would not be printed. This created dissonance between two inconsistent ideas: (1) I do things for a reason, and (2) I went without air conditioning for no reason. Participants resolved the dissonance by cutting energy use even more! That is, they decided: (1) I do things for a reason, and (2) I went without air conditioning because I believe in energy conservation (Pallak, Cook, & Sullivan, 1980). Similarly, Stone, Aronson, Crain, Winslow, and Fried (1994) applied dissonance theory to getting people to engage in safe sex. Specifically, they created dissonance by (1) having participants publicly advocate the importance of safe sex and then (2) reminding each participant about times when that participant had failed to use condoms. Stone and his colleagues found that participants reduced this feeling of dissonance by buying condoms.
Many researchers take advantage of the fact that a major purpose of theories is to explain what happens in the world. For example, researchers wanted to understand why fraternities engage in hazing (Aronson & Mills, 1959; Gerard & Mathewson, 1966). They wondered if cognitive dissonance theory could explain hazing. Consequently, they tried to induce dissonance in some participants by having them suffer electrical shocks as a requirement for being accepted into a “boring” group, whereas other participants were able to join the group without an “initiation.” The researchers found that the participants who received shocks resolved the dissonance caused by the opposing thoughts, “I am a logical person,” and, “I went through unpleasantness to join a boring group,” by deciding that the boring group was a pretty interesting group after all.
Theories are general rules that, ideally, hold most of the time under specific conditions. Therefore, ask yourself, “under what situations or conditions, does the theory not apply?” That is, has the theory neglected to specify important moderator variables: variables that can intensify, weaken, or reverse the relationship between two other variables?
Because researchers asked this question about cognitive dissonance theory, we now know that people do not change their attitude every time they behave in a way that goes against their attitudes. Instead, the presence of certain moderator variables will determine whether performing a counter-attitudinal behavior will change participants’ attitudes (Aronson, 1989; Brehm & Cohen, 1962; Festinger & Carlsmith, 1959). Specifically, if participants are going to change their attitudes after doing a counter-attitudinal behavior, the following conditions must be met:
1. participants must believe that they engaged in the behavior of their own free will (perceived freedom);
2. they do not receive a large reward for doing the behavior (insufficient justification); and
3. they view the attitude as important to their self-concept (self-relevance).
For example, if a smoker is forced at gunpoint to say smoking is bad, or given $10,000 for saying smoking is bad, or does not view smoking as important to his self-concept, the smoker will not change his or her attitude about smoking. That is, perceived freedom, insufficient justification, and self-relevance are all variables that moderate the relationship between doing a counter-attitudinal behavior and changing one’s attitude. Can you think of other moderator variables that should be included in dissonance theory? To answer this question, think about factors, situations, or circumstances that might prevent people from trying to eliminate inconsistencies between their attitudes and actions.
When looking for moderating variables, ask yourself whether the theory might be too parsimonious. For example, operant conditioning theory’s rule that a behavior reinforced under a partial reinforcement schedule is more resistant to extinction is too simple. A behavior reinforced under a partial reinforcement schedule is more resistant to extinction only when the person believes the reward is controlled by external forces, such as chance, fate, or the experimenter’s whim. Partially reinforced behaviors are not more resistant to extinction when the participant believes that getting reinforcement depends on skill (Rotter, 1990). In other words, the relationship between reinforcement schedules and extinction is moderated by the variable of perceived control.
Another way to generate research ideas from a theory is to design a study that tests the accuracy of the theory’s core assumptions. Often, attacking the heart of the theory involves examining the physiological or cognitive events that are--according to the theory--the underlying causes (mediators) of a phenomenon. For instance, cognitive dissonance theory assumes that when people have two beliefs that they see as contradictory, they experience an unpleasant, anxiety-provoking state called dissonance. To reduce dissonance, people will reconcile the inconsistency. In other words, dissonance theorists assume that dissonance mediates attitude change.
To test this assumption, you might try to induce and maintain dissonance in participants and determine whether they do find dissonance an unpleasant, anxiety-provoking state (Elliot & Devine, 1994). If participants felt decreased arousal, it would seem that you had disproved a core assumption of the theory. That is, you would have cast doubt on the belief that the state of dissonance is a mediating variable--physiological process or mental state that is the mechanism for how an event has its effect--for attitude change.
In addition to trying to measure an alleged mediating variable, you may try to manipulate it. For example, suppose that a certain manipulation tends to cause attitude change, presumably because it creates a cognitive and physiological state of dissonance that people then reduce. If the psychological state of dissonance is really the mechanism by which attitude change occurs, interfering with that psychological state should reduce attitude change. Therefore, you might expose all your participants to the attitude change manipulation, but do something so that half of your participants would be less likely to experience the physiological arousal of the dissonance state. For example, you might reduce any feelings of dissonance-related arousal by giving one group
a tranquilizing drug. If dissonance really is the mediating variable for attitude change, your tranquilized participants should experience less dissonance-related arousal and thus less attitude change than your other participants.
Rather than trying to torpedo a theory, some researchers think the best hypotheses are those in which two theories make opposite predictions. Ideally, these studies, called critical experiments, try to settle the question of which theory’s view of the world is more correct. One of the first critical experiments was simple but persuasive. Participants looked at two lights. Almost as soon as one went on, the other went off. According to structuralism, the person should see one light going on, another going off. However, according to Gestalt theory, participants should see the illusion of a single light moving back and forth. Gestalt theory was supported.
More recently, cognitive dissonance theorists have taken on psychoanalysis. Specifically, dissonance researchers tested the psychoanalytic position that if you express hostility towards a person, you’ll release pent up anger and consequently feel better about the person. Dissonance theory, on the other hand, predicts that if people are mad at someone and then hurt that person, then people will justify their aggression by denigrating that person. Consequently, after expressing their aggression toward a person, people will feel more hostility toward that person. Experiments support the dissonance prediction (Aronson, 1990).
If you can devise a situation where two theories make different predictions, you have probably designed a study your professor will want to hear about. However, even if you perform a critical experiment, do not expect the “loser” of your study to be replaced. The loser has only lost a battle, not a war. There is usually enough vagueness in any theory for its arch-supporters to
minimize the extent of the damage. That is, they may argue that their theory wouldn’t necessarily make the prediction that you claimed it would. In other words, they may say that you put words in their theory’s mouth. If they can’t claim that you put words in their theory’s mouth, they may concede that their theory applies to a more limited set of situations than they thought or they may modify the theory to account for the results (Greenwald, 1975). Because scientists usually respond to a damaging set of findings by modifying an established theory rather than “throwing the baby out with the bath water,” Darwin’s theory of evolution and Festinger’s theory of cognitive dissonance survive today, but not in their original form. That is, by adapting to new data, theories evolve.
As you have seen, theory is a very useful tool for developing research ideas and tying those ideas to existing knowledge. Without research based on theory, psychology would chaotically move in every direction with little purpose, like a chicken with its head cut off. Indeed, theory-based research is responsible for much of psychology’s progress since 1892, when psychology was described by William James (p. 468) as:
a string of raw facts; . . . but not a single law in the sense in which physics shows us laws, not a single proposition from which any consequence can causally be deduced . . . This is no science. . . .
Yet not everyone believes that theory-based research is always best (Greenwald et al., 1986; Kuhn, 1970; Skinner, 1956). (See Table T-3 for pros and cons of theory-based research.) Thomas Kuhn (1970) argues that theories can serve as blinders, causing us to ignore problems that don’t fit nicely into existing theory. Skinner (1956) also argues that sticking to a theory’s narrow path may cause us to ignore interesting side streets. Specifically, Skinner’s advice to investigators was “when you find something of interest, study it.”
In addition to stopping us from seeking new facts, theories may also stop us from seeing old facts in new ways. Thus, we may fail to make the kind of discoveries Darwin, Freud, and Skinner made--the ones that result from seeing what everyone else has seen, but thinking what no one else has thought. As physicists learned when Newton’s theory was largely overturned by Einstein, looking at things exclusively through one theory’s perspective is especially dangerous when the theory has not been extensively tested. In other words, some experts (Greenwald et al., 1986; Kuhn, 1970; Skinner, 1968) would agree with Sherlock Holmes’ statement about the danger of premature theorizing: “One begins to twist facts to fit theories rather than theories to fit facts.”
Fortunately, as Aronson (1989) points out, science is like a big circus tent. Under the tent, there is room for research derived from theory, but there is also room in the tent for researchers who follow hunches.
Not only can hunch-based and theory-based research share the same circus tent, but they can sometimes share the same ring. For example, suppose that an intuitive hunch led you to predict that having pets would cause the elderly to be more mentally alert and healthy. You might then use theory to help you clearly articulate a logical rationale for your prediction--or even to help you refine your prediction. For example, according to learned helplessness theory, a lack of control over outcomes may cause depression. Therefore, having a pet, or even a plant, may give one more of a sense of control and thus make one less vulnerable to helplessness (Langer & Rodin, 1976).
Back to Research Design Explained Student Home Page