An operational definition is a detailed specification of how one would go about measuring a given variable. Operational definitions can range from very simple and straightforward to quite complex, depending on the nature of the variable and the needs of the researcher. Operational definitions should be tied to the theoretical constructs under study. The theory behind the research often clarifies the nature of the variables involved and, therefore, would guide the development of operational definitions that would tap the critical variables.
There is an old saying that you can never be too rich. When it comes to operational definitions, you can never be too detailed. The more clearly you specify the procedures, the more likely that the procedures will be carried out precisely and the more likely that researchers who attempt to replicate your work will use the same procedures.
Even for simple things, it it best to specify procedures in detail. For example, if you want to weigh people in a study, you could just say "weigh them." That procedure seems obvious on the surface, but how do you guarantee that the scale is working properly (standardizing the measure) and what should you have your participants wear when they are being weighed. The weight of street clothes can vary significantly depending on how cold it is or what you might be carrying in your pockets. Those variation add error variance.
If you are having your participants take a psychological measure, you want to specify the conditions under which the measure should be given. Some measures, for example, will be answered differently depending on whether they are filled out privately or in a group. Some measures are affected by distractions in the environment. That is one reason why admissions exams like the SAT are often given under very strict conditions that are spelled out in detail for the examiners.
If you are unsure of whether certain variations in procedures will affect the scores, err on the side of caution and try to hold the procedures constant by specifying precise instructions for measurement. You can often learn what factors are known to affect a measure by doing a thorough library search of how that measure, or similar measures, have been used in the past.
The best way to illustrate the process of developing operational definitions for variables is to identify several theoretical constructs and develop multiple operational definitions of each. This not only illustrates how the process is done, but also shows that most constructs can be measured in more than one way. The research literature also shows that it is common for different operational definitions to tap different aspects of a construct and thus react differently to experimental manipulations.
Anxiety is a concept that most of us are all too familiar with. It is an unpleasant feeling that occurs in certain situations. It can disrupt our functioning if it is excessive, but it also motivates behavior.
So how do you measure anxiety? How do you operationally define anxiety? Now this is a problem that has challenged researchers for years, and many fine operational definitions of anxiety are already available for our use. For the sake of this exercise, however, we will assume that we have to develop our own measure without benefit of much of this existing research.
Since this is a concept that we have first-hand knowledge of, we might start the process of operationally defining anxiety by asking ourselves what it is like. What do we feel? How do we react? How do others react? What features in other people would suggest to us that they are anxious? These are all excellent ways to start this process. It is especially useful to focus on factors that indicate anxiety in others, because those are likely to be more objective and observable factors and would provide higher reliability.
When we think of anxiety, we think first about the "feeling" of being anxious. We know what it is like and we can easily tell when we are experiencing it. It is less clear whether others would be able to tell that we are anxious just by looking at us. If fact, our own experience suggests that we may be effectively hiding our anxiety, because some people have told us they were impressed with how calm we were at a time when we felt anything but calm. Furthermore, others have told us they were very anxious in a situation in which we had observed them and they did not look anxious to us. Nevertheless, the feeling of anxiety is distinctive, even if it is not always public, so it provides one way of measuring anxiety.
Since feelings are internal events, apparently without consistent external features, we will have to rely on self-reports to find out whether a person is feeling anxious. We could simply ask people to rate their level of anxiety on a 100-point scale, a technique that is commonly used. These are often referred to as SUDS (subjective units of distress) ratings. We could instead ask people a number of questions about their feelings, questions that tap elements of anxious feelings. These might include things like "I am worried about what might happen." or "I can feel my heart pound." The number of such items endorsed by the person would likely indicate the level of anxiety. With mild anxiety, a few might be endorsed, but as the anxiety became more intense, more and more of the items would be endorsed, because more of the anxiety symptoms would be intense enough that the person noticed them.
We just mentioned something that probably resonated with many of you. When you are anxious, your heart feels as if it is pounding, and when you are very anxious, you almost always experience this sensation. This is a real effect. Anxiety is not just a feeling; it is also a physiological response. When we are anxious, our heart beats faster and stronger, our muscles tense and we shake, our palms sweat and sometimes even our face sweats, our voice may crack or our face flush. Sometimes these effects are visible to others; often they are not unless the anxiety is very strong.
We all have witnessed someone giving a talk in class who was visibly shaking, whose voice was cracking, and whose face lit up the entire room with a red glow. We can use these responses to provide another set of ways of operationally defining anxiety. We can measure the physiological changes in people as an indication of their anxiety. If their heart rate increases, we would take that as a sign of anxiety. If their palm sweats, that is another sign of anxiety. Without going into the complexities of how one measures each of these things, we will just say that it is relatively easy to do so, and that these measures have often been used to index the anxiety level of participants in studies. With modern telemetry, it is even possible to monitor many of these physiological responses while the person is carrying out everyday activities in his or her natural environment.
Most of the people who were obviously nervous about giving a talk in school somehow got through the talks, but a few quit in the middle, sometimes even leaving the room. This is yet another indicator of anxiety--in this case, the behavior of fleeing the situation. We do not see it often in classroom situations, but people who are anxious of snakes will often run away or at least step back from the object of their fear. Furthermore, we often see avoidance of situations that produce anxiety. Someone who has been very anxious giving talks in public may chose to only take classes that do not require a presentation. He or she may even chose jobs later that are unlikely to require a presentation, even though it may mean making considerably less or having a less prestigious job. So behavior, both escape and avoidance, is yet another indicator of anxiety.
We have outlined three separate strategies for operationally defining anxiety. They include (1) asking people how anxious they are feeling, (2) measuring their physiological response, and (3) observing their behavior, especially their escape and avoidance behavior. The natural question for most students is which of these is the BEST measure of anxiety. In essence, which of the measures captures true anxiety most precisely.
The answer to this question for anxiety is often frustrating to students, but reflects the complex reality of human emotions. The answer is "It depends." Most students seem to prefer the physiological measures, because they seem more "basic." Certainly, the physiological measures have the advantage that we cannot deliberately lie about them. If we are anxious and we don't want people to know that we are anxious, we can always lie about how we feel, provided our anxiety is not so obvious that everyone can see signs of it. We can also stay in situations in spite of intense anxiety to avoid losing face or to do something that we feel is critical. Many nervous parents have spoken up at PTO meetings, because they thought it was important to the well being of their children.
But physiological measures also have their problems. The heart rate will indeed go up when we are anxious, but it also goes up for lots of other reasons as well. Walk up a flight of stairs and your heart rate will have increased several beats a minute to meet the aerobic demand. Your palms will sweat from nervousness, but they also sweat, along with the rest of your body, when you are hot. The same is true of face flushing. Your muscles will tighten when nervous, but they also tighten when you are expecting to act or are engaged in physical action. So none of our measures of anxiety is ideal.
If none of our measures of anxiety is ideal, which one should we use. The best answer is "as many as we can." The truth is that each of these measures capture a different aspect of the construct of anxiety, and therefore they do not always agree with one another. For example, people can avoid a situation without showing visible signs of anxiety, but the avoidance is a strong indicator of their feeling about the situation. Even though there may be little physiological arousal and they may claim to not be anxious, their avoidance is telling another story. The validity of that other story can often be confirmed if the person is required to face what they have been avoiding.
Looking at it from another perspective, we often see people with considerable anxiety, as measured by their physiological responses, performing all of the things required of them. Golfers might calmly sink a 10-foot putt to win a tournament, even though their heart might be racing and their palms are dripping wet. So are they anxious or not? Scientifically, the fact that these various measures of anxiety do not always agree has led to a much more thorough understanding of anxiety. We now know that it is not a single construct, but rather represents a complex collection of responses, and that the pattern that we will see will depend on the situation that the person is in. We would never have been able to recognize that if we had not operationally defined anxiety in several different ways and used all of those various definitions in our research studies.
Civic responsibility seems like a clear construct. People who are civic-minded are likely to do what is expected of them by society. But what is it that is expected of a responsible citizen? Do responsible citizens vote regularly? Do they agree to serve on juries? Do they donate time to the Scouts or Little League baseball? Do they drive within the speed limit? Do they work to help solve world hunger? Do they pay all the taxes that they owe? Must they do all of these things in order to be a responsible citizen, or would a certain subset of these activities be sufficient? Should some of these activities be considered mandatory of responsible citizens, like voting for example? Would the late Harry Chapin, a well known song writer and performer, be considered a responsible citizen? He gave as many as a hundred benefit concerts a year to combat world hunger, but by most accounts drove like a maniac, collecting frequent speeding tickets. His reckless driving eventually took his life in a fiery crash on the Long Island Expressway.
This rather clear construct suddenly gets fuzzy when you start wondering about how to measure it? What behaviors should be included? What should be excluded? Do some behaviors bias you against certain people. For example, would doctors who try to avoid jury duty, because of the demands of caring for their patients, be responsible or irresponsible for their decision? Would people who avoid activities, because they are uncomfortable around large groups of people, appear to be irresponsible because they do not engage in important civic activities? The best solution is to provide a standard set of behaviors that will be taken into account. As we will see, these behaviors either can be taken from the person's everyday life or can be determined by specified laboratory procedures.
Natural Environment Measures. Many psychological measures are based on looking at samples of relevant behavior in natural environments. Sometimes these measure rely on the self-report of behavior. Other times the behavior can be measured through standardized observations of activities. For example, we could construct a measure of civic responsibility with 10 items that represent things that one believes that a responsible citizen would do. We could ask people to rate on a scale how frequently they do each of those behaviors. The behaviors may include things like voting, learning about the qualifications of candidates running for office, staying informed about civic matters, supporting the activities of those who are working for the community, and so on.
The more clearly we specify the items, the more likely each participant taking the measure will interpret it the same. For example, an item like "I vote in most elections" is more ambiguous than "I have voted in at least 4 of the last 5 elections." Items such as "I support the efforts of community leaders" is so vague and open to interpretation that you would have no idea what endorsing that item would say about the individual. If measures of this construct already existed and the reliability and validity data for those measures were adequate, you would definitely want to go with them. If not, you would have to develop your own measure, and then it would be your responsibility to gather reliability and validity data as part of that process.
You might be asking yourself what would stop people from lying about their activities? The answer is "not much," and of course some people would lie, or at least would try to put their activities in the best possible light. There are ways to deal with this problem. One way is to have some items that measure this tendency to place oneself in an overly favorable light. For example, including items such as "I thoroughly investigate the qualifications of all candidates before voting" would likely pick up the extent of such self-promotion. It is unlikely that anyone THOROUGHLY investigates the qualifications of ALL candidates before voting, so those who claim to do that are probably claiming other things that are less than accurate.
Some questionnaires are much more effective in giving an accurate indication of behavior if they are completed anonymously. Structuring the research study so that such anonymous responses are possible may dramatically improve the effectiveness of some measures. Clearly in this situation, the operational definition includes not only the items of the psychological measure, but also the conditions under which the measure should be administered.
Another way to guard against participants presenting themselves more favorably than is warranted is to base your measure on publicly recorded data. For example, voting records and jury participation records can often be accessed. We won't be able to know how people voted, but we will know if they voted. Some public service activities, such a being on advisory boards, are a matter of public record. Detailing a list of such public service activities as your operational definition of civic responsibility, and then doing the leg work to track down the requisite information, can provide a measure that is not contaminated by a tendency to inflate one's actual civic-minded activity level. Of course, a disadvantage of this approach is that many civic-minded activities are not public and therefore could not be included in this measure.
Laboratory Measures. If one thinks of civic responsibility as a trait, one would expect that it would occur in many different settings. Therefore, one does not have to sample every possible setting to get an idea of how civic-minded a person is. This opens the possibility that a laboratory analogue could prove to be a very satisfactory measure of this construct.
Laboratory analogues are laboratory tasks that represent behavior that is believed to be similar conceptually to the natural behavior in the community. The task need not take place in the laboratory, but it will be under the control of the researcher. For example, the researcher could call participants in the study to see if they are willing to do something that one would expect a civic-minded individual to do. It might be something like agreeing to help on a project or lending their name to a worthy cause. Of course, people may be unwilling to help on a particular project, whereas they are willing to help on other projects, so such a standardized measure will not be a perfect indicator of civic-mindedness. Nevertheless, it is a reasonable operational definition of this construct, and it has the advantage of being under the control of the researcher.
One must accept that there is no perfect operational definition for a given construct. Each operational definition will have advantages and disadvantages. A self-report measure is often quick and easy, but it is subject to presentational biases by the participants who take it. Actual counts of behavior are less affected by presentational biases, but they are much more time consuming and often miss critical behavior that is private. Physiological measures can tap some constructs, but physiological changes occur for many different reasons; therefore, it is hard to know if observed physiological changes are an indication of the construct of interested. Laboratory analogues have the advantage of experimental control, but there is always the question of how closely they relate to real world behavior.
Because no single operational definition is likely to provide the perfect measure of the construct of interested, it is wise to consider using more than one operational definition in a given research study. If you randomly select a dozen research studies from the best journals, you may be surprised to see how often this approach is used. Multiple operational definitions help us to zero in on the constructs that we are studying, and they often give us insights into the complexity of those constructs. We already discussed how anxiety researchers now recognize that the feelings, behavior, and physiology of anxiety are not just alternate ways of tapping anxiety, but represent distinctly different aspects of anxiety. By recognizing this basic fact, we can begin to identify how these various aspects of anxiety fit together. This is science at its best--a concerted effort at zeroing in on the workings of nature.