Defining, Measurement, and Scaling
In this unit we will discuss the ways in which a researcher observes information about human and animal behavior. We will also describe some of the experimental procedures used to ensure that observations are made in an accurate and orderly manner. This discussion is not meant to be comprehensive, but it will provide an understanding of the types of questions, procedures, and rules that concern researchers.
Try to answer the following questions as you read this module. What general rules do scientists follow when conducting scientific inquiry?
Rules represent scientific disciplines required for credibility in research
Researchers have established over the years a set of practical rules and guidelines, generally unwritten, which they follow in conducting research projects. These rules are dynamic, in that they have undergone many changes and additions through time, and there are undoubtedly more changes to come. Such changes have included the use of control groups, the idea of operationalism, random samples, and so on. Such rules must be followed if data purporting to be scientific fact is to be accepted as such. When one ignores these rules, research results have no credibility among scientists.
Science is an endeavor somewhat like the game of chess; one must follow the rules in order to be admitted as a player. For example, a knight moves three squares, two in a horizontal or vertical row, and one perpendicular to the row. One who ignores this convention will probably find it difficult, if not impossible, to be accepted as a chess opponent by any but the totally unwary. In the same way, credibility in the scientific establishment cannot be achieved without a willingness and ability to follow scientific disciplines in one's contributions to scientific knowledge.
The first major problem in building a body of knowledge concerning the behavior of organisms is in deciding how to communicate theories and evidence to other scientists. One approach to communication has been labeled operationalism. There are other approaches, such as logical positivism, but our discussion of operationalism will include all of the major factors to be considered in deciding how to accumulate, record, and present scientific information.
PROBLEMS OF DEFINITION
Most terms in psychology can be interpreted in many different ways
Do terms such as anxiety, intelligence, or acceptable behavior mean the same thing to everyone? Surely not, yet these are the types of terms with which psychologists must work and communicate. Anxiety might be recognized as overt behavior, such as sweaty palms, rapid speech, knocking knees, or even rapid respiration rate. One might also recognize anxiety in less explicit behaviors, such as a lack of poise in new situations. How can psychologists speak meaningfully about anxiety when they do not agree on an acceptable definition of the term? It can be done, but in order to do so anxiety must be defined in terms of public events that all can observe and agree on. Therefore, one will find some of the following used as definitions of high anxiety in journals of experimental psychology:
A scientist may not wish to accept any of these as adequate definitions of anxiety, even though they meet the requirements of being objective and open to public inspection. As long as a scientist reports as fact only what he and others can observe, the evidence (and the scientist) are credible, even if inconclusive. An important rule of scientific inquiry is to always distinguish between fact and interpretation, so that each can be evaluated independently.
Operationalism can best be described by example. Suppose a researcher says that he used ten motivated rats as subjects in an experiment. Understanding the experimental results requires an understanding of the definition of "motivated rat." But suppose the researcher said, "I used ten rats that had been deprived of food, but had water freely available, for twenty-three hours prior to the experimental session." The researcher in this case stated the procedures, or operations, used to produce ten rats motivated to secure food. His operational definition of motivation is: 23 hours of food deprivation.
Operational definitions emphasize actions rather than conditions
Notice that in this case a procedure has been indicated whereby another psychologist can produce a similar state of hunger in other experimental subjects. Operationalism requires that an event be replicable, that is, capable of being repeated. A replicable event does not require that two people form the same subjective judgment. If learning is found to be a function of the number of M&M's given upon making a correct response in one laboratory, yet this same finding never occurs in any other laboratory, then we probably would agree that the relationship between learning and M&M's has not been established as a scientific fact. The finding was probably due to chance.
Stating definitions in terms of operations avoids communication problems. Misunderstanding is often created as a result of using terms that have no empirical definition. Everyone knows what "success" means, yet look at the various images of successful people that different people carry in their imaginations. Success to a doctor's son may not be the same as success to an itinerant farmer's son. A few years ago, cartoonist Charles Schultz published a book entitled / Happiness Is a Warm Puppy. This definition comes very close to being operational. That is, rather than defining ``happiness" as an inner state of the child, Schultz defines it in terms of conditions which are identifiable and replicable. Operationally, we might define happiness for a child (a /a Schultz) as "an opportunity to play with a puppy for 10 minutes."
An operational definition of a variable is usually one of many possible definitions in common usage
One of psychology's major difficulties is that specific variables often have names that are defined subjectively in common usage, e.g., reward, punishment, or frustration. Even if everyone agreed on general meanings, most variable names carry "surplus" meanings that are irrelevant to the experiment. We could rename the variables, but the results would become inaccessible to all but an inner circle of researchers who knew the jargon, which is enough of a problem already. Operational definitions provide meanings that are limited to the scope of the research being described. For example, we might induce frustration by blocking an organism for 15 minutes from a desired goal. Such definitions aid in communicating our ideas to other scientists. They may make results less interesting to nonscientists, but they also make it possible for another psychologist to repeat a study, with large or small variations, to see if its results are stable.
MEASUREMENT AND SCALING
Having determined an operational definition of variables, the next objective is to measure them accurately. In day-to-day conversation, most things are measured in terms of labels, using what we call a nominal scale. The operational definition of anxiety as a "verbal-report" meets the requirements of a nominal scale; subjects attach either an "anxious" or a "non-anxious" label to themselves. All subjects belong to one or the other group. In the same way, the identity of one's sex is a nominal variable (unless we are concerned with how much of it one has). When the sex variable is measured on a nominal scale, subjects are divided into two categories or groups according to gender. The numbers used to identify football players also form a nominal scale. There is no order, or sequence, or quality, implied by such a designation. Player number 60 is not twice as good or twice as large as player number 30. Nominal scales provide very little information; they do little more than make it possible to identify and count the number of objects, persons, or events that bear a label.
An ordinal scale permits a variable to be identified in some sort of sequence, or ranking. Sometimes teachers rank the performances of their students by assigning a rank of "1 " to the best student, "2" to the second- best student, etc. Notice that with an ordinal scale it becomes possible to make descriptive statements about the subjects within a given category. For instance, dividing a class into two categories, boys and girls, makes it possible to rank the boys in terms of their classroom performance and to do the same for the girls. An ordinal scale may also establish a hierarchy, or "pecking order" in the case of chickens. The limitation of this scale is that, although it indicates who is best in a class, say, it does not describe how much better the best is than the second or third best. The second best may be light years better than the third best, but just slightly poorer than the very best. An ordinal scale gives only comparative information; no assumption can be made about how much lower one score is than another. An illustration is provided by the annual winners of each of the major football conferences in the United States. Each conference winner has a rank of "1 " within its own conference; it is the best within its conference. But there is still no suitable way to compare the winners of the conferences. Therefore, many people suggest playoff games among the conference winners in order to determine the national champion.
An interval scale is a scale on which the distances between numbers are equal. The Fahrenheit temperature scale is an interval scale. On an examination with a maximum of 100 points, the distance between scores of 90 and 95 equals the distance between scores of 45 and 50, which would not be the case if these numbers were class rankings. We can order and count scores on interval scales as we do for ordinal scales. In fact, each scale has the properties of the scales that occur before it; a set of nominal, ordinal, interval and ratio scales (discussed below) forms an ordinal scale, in that each successive scale incorporates more of the formal properties of mathematics than the preceding scales. No ratios are implied by an interval scale because no true zero point exists. IQ scales are often considered interval scales because they do not have a true zero point. The complete lack of intelligence has not been defined.
A ratio scale is basically an interval scale with a true zero point. With this scale it becomes possible to make statements about ratios. We can say that a father, six feet tall, is twice the size of his son, who is three feet tall. With IO scores, however, we could not meaningfully say that a score of 150 is twice as great as a score of 75, since no true zero point has been established for intelligence. Other examples of ratio scales are the absolute temperature scale, time, weight, and most other physical scales. Psychologists may use time to measure a rat's speed in running an alley. Time is usually used as a ratio scale; that is, a rat who ran an alley in 20 seconds is twice as fast as one who ran it in 40 seconds. However, if the running speed is taken as a measure of some covert psychological variable, such as motivation, then it is no longer a ratio measurement. A 20-second rat is not necessarily twice as motivated as a 40- second rat.
Nominal and ordinal scales are the most freguently used scales in psychological research
Many psychological variables are measured by nominal or ordinal scales, and a few are measured by interval scales. Some behavior can be measured by ratio scales, but, as we saw in the previous example, these contribute little to descriptions of underlying psychological dimensions. Such measurements give true ratio scales only when used with quantifiable behaviors. When we carry out psychological research, we should always be aware of the characteristics of the scale we use, since the type of scale employed influences the statistical treatment of the data.
Now take Progress Check 1.
ANSWER KEY PAGE 71
6 OR MORE CORRECT PAGE 39
FEWER THAN 6 CORRECT PAGE 35
Reverence may be defined in several ways. If we define it in terms
of the number of times a person goes to church, which of these
people would be considered reverent by our definition?
a. "I don't believe in God, but I go to church every Sun- day."
b. "I never get to church as I'm crippled, but I pray at home daily."
c. "I believe in God, but I never go to church."
Obedience may be objectively defined as compliance with the com-
mands of an authority. According to this definition, which of the
following would be an example of obedience?
a. A man saying, "I always support the police" as he runs a stop sign.
b. A district attorney that prosecutes cases and accepts bribes on the side.
c. A fellow that hates the police, but always follows their directions
|Group||Order Implied||Intervals Equal||True Zero Point|
Referring to the text if necessary, put an X in each column if the characteristic applies to the scale on the left. For example, ratio scales have a true zero point, so we mark an X in that box. _____________________________________________6
Indicate the type of scale used in the next seven examples.
An instructor assigned "1 " to the highest score on a test, "2" to the next highest, and so on. ____________________________________ 3
Having an IQ of 200 does not mean that a person is twice as intelli- gent as one whose IQ is 100. The distance between 100 and 101, however, is the same as between 140 and 141. _____________________________________________ 1
3 Ordinal 4 c
SCALE CHARACTERISTIC Group Order Implied Intervals Equal True Zero Point Nominal X Ordinal X X Interval X X X Ratio X X X X
Each car in an auto race has a different number assigned to it arbitrarily. __________________________________ 4
Temperature is measured using a centigrade thermometer. __________________________________3
The tallest person in a room is assigned " 1," the next tallest "2," and so on. __________________________________5
A psychologist measures mass using a scale calibrated in grams _______________________________1
An experimenter analyzes the results of a test designed to discover attitudes toward war. He then assigns "1 " to subjects in favor of war "2" to those opposed, and "3" to those who do not care. __________________________________ 2
1 Ratio 2 Nominal 3 Interval 4 Nominal 5 Ordinal.
NOW TAKE PROGRESS CHECK 2
PROGRESS CHECK 2
1. Match the following variables with the appropriate scales.
d)______ IQ scores
e) ______Pages in a book
f)_________ Number of responses made per unit of time
2. Which of these may be an operational definition of motivation?
a. The spring to action
b. Being deprived for 23 hours
c. Having the "urge" to accomplish something
d. A personality trait
3. With nominal scales we may ___________________________ responses, events, or persons within a given category.
d. determine interval equality or inequality of
4. The blue, red, and white ribbons presented to the winning entrants in a contest represent a(n):
a. nominal scale.
b. ordinal scale.
c. interval scale.
d. ratio scale.
5. Repeatability, or replicability, should be possible for most scientific data.
6. Psychological variables often do not show a perfect relationship to the physical, particularly ratio, scale that is measuring it.
7. Each psychologist establishes his own rules and procedures for determining scientific truths.
8. Scientific data are generally subjective.
7 OR MORE CORRECT PAGE 39
FEWER THAN 7 CORRECT INSTRUCTOR CONFERENCE
ANSWER KEY Page 71
Unit 13 Table of Contents