A GUIDE TO UNDERSTANDING CRIMINOLOGICAL THEORY

    This page provides practice at three important skills: (1) translation of abstract, general statements into measurable, plain English (operationalization); (2) identification of the Independent Variable (cause), Dependent Variable (effect), and Other Factor(s) (specification); and (3) figuring out other possible relationships between variables in a theory (elaboration). Integration is a fourth skill we could have added, but for now, let's concentrate on the three basic ones in italics. The SAMPLE we will work with is the following statement:

"Criminality is a probabilistic event determined by the frequency and quality of interaction with persons holding definitions favorable or unfavorable to violation of the law."

In unlocking the meaning of the above statement, we will use three (3) ordinary, everyday TOOLS at our disposal: (1) Name Recognition; (2) Sentence Analysis; and (3) Visualization. The insights these tools might yield about operationalization, specification, and elaboration may not come in any systematic order. That is, we may arrive at an understanding of elaboration first, specification second, and operationalization third; or any other order. We are concerned with understanding theory, not following any established procedure.

NAME RECOGNITION involves trying to connect the name of a theorist with the ideas of the theory; that is, who the inventor is, who made the statement, who might be associated with this kind of thought, etc. In our Sample Complex Theory, reproduced once again,

"Criminality is a probabilistic event determined by the frequency and quality of interaction with persons holding definitions favorable or unfavorable to violation of the law."

the phrases "frequency and quality of interaction" and "definitions favorable or unfavorable" should look familiar to any student of sociology or criminology as representative of the Interactionist approach. In fact, the word "interaction" is even in the statement. Further reading of the whole statement reveals words like "persons" which tell us it is probably a SOCIAL theory, having to do with the kinds of things, like "definitions", which people presumably pick up from other people in some way. These kinds of theories about things picked up from other people are called LEARNING theories. We therefore have a SOCIAL LEARNING theory which probably also adheres to an INTERACTIONIST or INTERPERSONAL approach. A look at our table of perspectives, approaches, and theories reveals that either Sutherland, Glaser, or Akers could have made this statement. Now it becomes a matter of narrowing down the field of names. You need to have some rough idea of how these three theorists (Sutherland, Glaser, and Akers) tend to write (one reason why professors sometimes make you read the original works), what they look like (one reason why textbooks often put their pictures in books -- who knows, maybe something in the way one looks provides a memory recall aid), and who the most important theorist is in the field or to your professor (information gained from lectures) in order to make a safe guess. If you guessed Sutherland (the "father" of Social Learning Theory), you made the right choice. Our sample is taken word-for-word from Sutherland's Differential Association Theory. With practice, and some knowledge of the oral tradition passed down about theorists, you can get good at this.

Sutherland, unlike the other two theorists, never made very clear, say-what-you-mean statements. As a rule of thumb, the more abstract and obtuse the statement, the more likely it was made by the founder or "father" of a particular approach. As an example, just look at the way Sutherland's statement is worded:

"Criminality is a probabilistic event determined by the frequency and quality of interaction with persons holding definitions favorable or unfavorable to violation of the law."

Some of the more abstract elements include phrases like "Criminality is a probabilistic event" and "definitions favorable or unfavorable". Nobody talks like this in real life. Why didn't he just say: "Crime is caused by interacting with others who believe in breaking the law"? This would be a nice, plain English translation. But it's too close to the kind of simplistic statements going around during Sutherland's time. He wanted to avoid simple "Crime is caused by..." statements, and also avoid a twist on that old adage "Birds of a feather flock together" (which in this case would be: "Those that flock together tend to become like birds of a feather"). Remember that theorists usually try to make their theories measurable, and Sutherland tried to do this with the "probabilistic event" language, which lends itself to mathematical testing. Every theorist has some quirk to help you get an idea of where they're coming from, but they also usually prepare their statements very, very carefully. Every phrase they put in their theoretical statements has some meaning.

SENTENCE ANALYSIS
Take the beginning of Sutherland's theory: "Criminality is a probabilistic event determined by..." Now, criminality is an old word leftover from the days of biological determinists who were trying not to sound so absolutely certain. They realized that heredity could only account for about 50% of something at best, so they started coming up with phrases like "propensity for", "predisposed to", "inclination towards", and "criminality". So when Sutherland uses this word, you get the idea that he is talking about being inclined toward crime, not actually committing crime. This is called SPECIFYING the Dependent Variable, or effect. For people new to the science of criminology, it is astonishing how many theories are not about crime at all, but the "readiness" to commit crime, "potential" crime, or "tipping points" in the community's tolerance. Anyway, the point is that there is more than one Dependent Variable in criminology. Few theories provide explanations for actual crime or specific criminal events.

From Sutherland's point of view, what is a "probabilistic event"? Well, probability is simply the idea that something can happen greater than chance; that is, greater than 50% of the time. If we were to test Sutherland's theory, all that would be needed is to take 100 people, expose them to what he says are the causal factors, and see if at least 51% of them are more inclined to commit crime than before. Case closed, theory proven. Theorists always hedge their bets like this in some way. No theorist claims to explain anything close to 100% of something. Many criminological theorists are perfectly happy with small to moderate effects in the 20-30% range, although predictions in the 50-70% range (or higher) are expected for policy relevance. You should get into the habit of estimating the percentage of explanation by what the theorist says. There are important research method reasons for this; a stronger theory requires a stronger research design.

Another part of SENTENCE ANALYSIS is to find what is called the "relational" word. A relational is simply the verb or verbs contained in the statement. Everyone uses verbs in sentences, but you will probably never see the verb "to cause" used by a criminologist. "Determines" is the closed thing to "cause", but it is also rarely used. I wish I could provide you with a standard list of the hundreds of verbs and the rules for interpreting them, but there are no such hard-and-fast rules. Below, I am providing you with some general interpretations, but do not assume any automatic translations:

• "varies with" -- this means things fluctuate together; as one thing goes up, the other thing goes down; usually used to describe a possible inverse relationship.
• "where..." -- while not technically a verb, this word usually indicates a feedback relationship, where things go up or down in response to one another. Often, but not always, the case involves an important Z factor which moderates, distorts, or confounds the relationship. Relationals like "varies", "fluctuates", "predominates", "associated with", and "overrepresented by" are usually found when the theorist is dealing with sociodemographic variables, like age or race.
• "seems to be" -- this wishy-washy language usually means that the theorist suspects a weak relationship, probably way less than 50%.
• "tends" -- this might mean, but not always, that there are important Z factors which are antecedent, intervening, or contingent; that is, that come before, in the middle, or after an X and Y relationship. Or, it may be a cojoint relation.
• "is conducive to" -- this usually means that the cause is a mysterious, unknown construct; typically found in highly abstract theories involving words like anomie, relative deprivation, norms, or controls. In some cases, it refers to a confounding or contextual relationship.

Don't let the words I just used, such as cojoint and confounding, intimidate you. They are related to our third, everyday tool, VISUALIZATION. Here, you want to look at the nouns in the theoretical statement, and try to figure out which ones were intended to be the X's (causes), the Y's (effects), and the Z's (other factors). At a minimum, all theories require at least two variables, an X and a Y, but in real life, most theories are complex and have at least one Z factor. In the interest of what is called parsimony, theorists try to keep their models down to 3-5 variables. In visualizing, also called MODELING, you can determine the relative importance of these Z factors and their impact on the X-Y relationship by looking at how the variables can be rearranged. Sometimes, the theorist has done this for you by "controlling", "partialling", or "factoring out" certain variables, but this is only a statistical solution. To understand all the possible ELABORATIONS of a model, you need to think through all the possibilities, especially those that the theorist forgot. It would be helpful to look at the page of modeling relationships between variables before reading further.

There are two kinds of MODELS: intrinsic and extrinsic. Intrinsic models usually refer to visualizations using the variables or implied variables provided by the theory. For example, you might be reading a book where the main theory is contained in chapter 2, but chapter 6 contains a discussion of the locus of control variable. Even though the author never incorporated locus of control into his or her theory, it was nonetheless implied as an important Z factor. An extrinsic model is one in which you, the reader, have added a possible Z factor. For example, you might be interested in whether the relationship predicted by the theory holds true for women or people of color. In this case, we would refer to sex or race as an extraneous variable to the model because you added it to the author's original work. There are also other words, such as exogenous, which apply to cases where you are modeling the impact of a criminal justice system factor, such as a new law or change in political environment. For the sake of simplicity, we will concern ourselves with visualizing intrinsic models only. Let's look again at Sutherland's theory:

"Criminality is a probabilistic event determined by the frequency and quality of interaction with persons holding definitions favorable or unfavorable to violation of the law."

We already know that Sutherland and the Social Learning Perspective draw heavily upon the Interactionist or Interpersonal approach in Sociology. After all, Sociology is what puts the S in Social. Therefore, we would not expect a sociologically oriented criminologist to use a psychological variable as their X. Theorists usually devote their first, or antecedent, variable to the discipline in which they were trained. So we should scan his theoretical statement to see what other sociological words are used; to see if nouns like "norms", "values", or "socialization" are used. A quick look reveals that nothing of the sort appears outside of "interaction" and "definitions". But it is not interaction itself that Sutherland is pointing to, but its "frequency" and "quality". Likewise with definitions (a strange word in itself), he points to those that are "favorable" and "unfavorable". Which is more important? Which comes first? Which is more measurable? We need to keep asking these questions because that is why theories (as puzzles) are there; for each successive generation to find out if anything new has changed the way the theory can be elaborated. There are some rules of thumb about estimating the relative importance of variables with intrinsic models, but again, these are just interpretive guidelines, not hard-and-fast rules:

• X is usually a variable from the discipline a theorist was trained in.
• X is usually more important than Y, especially if the theorist seems more interested in collecting causes (i.e., constructs an index).
• Y is more important than X if the theorist seems more interested in predicting symptoms, syndromes, or effects (i.e., constructs a scale).
• X is always more important than Z (a third factor), always.
• X is usually less measurable than Y or Z, because it's an unknown construct.
• X and Z are usually hard to tell apart, and may seem like the same thing.
• When more than one Z is in the statement, the first one mentioned is the Z in estimating a parsimonious model.

With regard to measurability, "frequency" is obviously easier to measure than "quality", and I wouldn't even bother with trying to figure out whether "unfavorable" definitions are easier than "favorable" ones since they are about equally difficult. But "definitions" (that strange word) appear to be the least measurable thing in the whole sentence. Definitions might be the main construct (the idea in the theorist's mind) which is our X (causal factor).

Since we already know the Y (effect) factor: probability of criminal inclinations, from an earlier discussion, we need to estimate the Z (other) factor. If we followed our last rule of thumb (rule #7), we would choose frequency over quality, but the same reasoning we used in determining our construct applies (as well as rule #5), and we should choose quality over frequency. Why? Because the harder-to-measure rule is more important than the position-in-sentence rule. Other reasons: quality is a more sociological word than frequency; Sutherland seems to like more unmeasurable concepts; and Sutherland seems more interested in collecting causes of which quality might be a part instead of frequency which runs the risk of being associated with symptoms.

There are, of course, other ways to arrive at these same conclusions. We can draw inferences out of the Approach and Perspective the theory is located under. Since approaches contain assumptions about human nature, models of society, and so forth, we can use this valuable information to rule out other possibilities. We know that learning theories, in general, hold to a blank slate position on human nature. We know that they hold to a consensus model of society, a process (motion picture) orientation toward human action , and a micro-level explanation of social forces. Any one of these provide clues to unlock the theory. The consensus assumption, for example, tells us that "definitions" might be important building blocks in social order, without which, society might collapse. This reassures us that "definitions" are the X in Sutherland's theory. In fact, definitions are beginning to sound an awful lot like norms, but they're not. They might play some part in an interactionist account of social order, but this would be a micro-macro link issue which is beyond our concern. Adding norms would create an extrinsic model; adding a social order variable like degree of solidarity would be adding an exogenous factor.

Now, we need to visualize the model. We have some idea of the important variables, but what is their logical order? It helps to ask some causal sequence questions, some which-came-first, some chicken-or-egg questions. Can definitions exist without people? Probably not. It takes people to come up with definitions in the first place. This is exactly what we would expect from a social theory. OK, so you've got to have some high-quality interaction going first. This means that although definitions are the primary causal factor, quality of interaction must be an antecedent variable. In other words, Z must precede X before Y in our model. Therefore, we have:

Z - X - Y
Z leads to X which leads to Y
An Antecedent Variable or Conditional Relationship