MEASURING HOW PEOPLE CONTROL THE AMOUNTS OF SUBSTANCES THEY USE

Robert Apsler

Robert Apsler, PhD is Assistant Professor in the Department of Psychiatry at the Harvard Medical School. He is involved in the study of psycho-social aspects of drug use and in the development of new techniques for evaluating drug treatment. The research was supported by a grant from the National Institute on Drug Abuse, #51181DA01519.

A new measure-control style-is presented as an alternative approach to measuring drug use in an effort to avoid the many limitations of existing methods. Control style and a traditional number of occasions of use measure are compared in their relationship to two dependent variables-drug dependency and drug problems. Results of a large, random household survey investigating individuals' use of many substances show that 1) responses on the control style measure are consistent and related to other measures in a meaningful way, 2) control style is more strongly associated with drug problems than is number of occasions, but the reverse holds for drug dependency, and the pattern varies with the particular substance; and 3) there is little overlap between the two measure in accounting for variation in drug problems and dependency. Thus, control style appears to be an important new tool for measuring drug use.

0ne of the most severe handicaps confronted by psycho-social investigators within the alcohol/drug field is the absence of adequate measures of drug use. The necessity for useful measures should be obvious. Without adequate measures of drug use, 1) existing theories cannot be tested effectively, and 2) data collected expressly for the purpose of generating new theories tends to be so inconsistent, misleading, or irrelevant that little progress can be made.

The purpose of this article, therefore, is first to describe and explore the reasons for the lack of adequate measures. Next, the role that the concept of control has played in the alcohol and drug literature is reviewed, and control style is introduced as a potentially important addition to the standard measures of drug use. Finally, results from a community survey which employed both the proposed measure of control style and a standard measure of drug use are presented in order to document empirically the utility of the control measure.

Three of the most serious difficulties with existing measures of drug use are: 1) lack of standardization, 2) absence of clear definitions of basic concepts, and 3) presence of serious practical obstacles to accurate measurement. First, reviewers of operational definitions in sociobehavioral drug research have reported on the diversity of operational measures of drug use utilized by survey researchers (Ellinson & Nurco, 1975; see also Bentler and Eichberg, 1976; Eichberg & Bentler, 1976). Since it appears that nearly every new drug survey employs a unique set of operational definitions, results from most surveys are not comparable with results from other surveys. Drug use questions differ mainly in 1) time period covered, 2) frequency categories presented to respondents, and 3) focus of items (number of occasions the drug was used versus the number of units of the drug that have been used) (Richards & Cisin 1975; Kandel, 1975). The problems of non-comparable data are compounded when investigators apply labels such as "experimental use" and "heavy use." These labels give the misleading impression that comparison between surveys is possible when, in fact, a close look at the operational definitions often reveals that one investigator's "experimental use" is another's "moderate use."

Second, the most fundamental concepts in the drug/alcohol fields, such as drug use, drug dependence, drug abuse, loss of control, alcoholism, etc., are extremely vague (Josephson, 1974; Smart, 1974; Rohan, 1975; Sobell & Sobell, 1975; Apsler, 1978; Zinberg, Harding, & Apsler, 1978). Conceptual definitions are obviously important for many reasons. The reason which pertains here is the necessity for specifying the population of operational definitions which can be used to measure a particular concept. Precise conceptual definitions are essential in order to delineate a set of rules which investigators can use to determine whether a particular operational definition does or does not constitute an appropriate measure of a particular concept. Since the conceptual definitions in the drug/alcohol fields are extremely vague, it should not be surprising that the outcome is a confusing array of measures, such as those for drug use. In the absence of clear definitions of drug use, but with the presence of strong pressures to obtain data on drug use, disagreement among operational definitions of drug use is almost unavoidable.

Third, a host of practical problems confront investigators who want to ascertain the frequency and quantity of drug use. To begin with, drugs purchased on the street are notoriously impure, often making it impossible for the user to know the actual contents without obtaining a chemical analysis. Even in those cases where the amount of drug contained in each unit is known, the drug's effects may cloud users' memory of the precise quantity ingested. Even without drug induced distortions, imperfect memories are another factor. Few people keep records of their drug use, thereby forcing investigators to rely solely on respondents' memories, sometimes of very distant events. It is not uncommon for investigators to ask drug users for a detailed accounting of their drug use during their first year of use, a time which might be many years in the past.

Another set of practical problems confronting researchers is that so many variables influence individuals' experiences with drugs. It is well known, for example, that an individual's physical characteristics such as body weight and rate of metabolism affect experiences with many drugs. Beliefs (especially expectations) and personality characteristics are other factors. The setting in which drugs are used (number and nature of others present, other activities going on, etc.) is another potent factor. Thus, even if accurate measures of frequency and quantity of drug use were available, these measures would not necessarily be adequate indices of users' reactions: to drugs.

At present, the many theoretical and practical problems which complicate measurement of drug use are largely unsolved. The result has been a proliferation of different approaches for assessing the quantity and frequency of drug use. Given the limitations of these measures, it is necessary to develop alternatives for measuring drug use. Keep in mind that a major objective in measuring drug use is to provide a means for relating drug use to other variables-most notably drug problems, drug dependency, etc. Thus, it is the discussion within the drug/alcohol field of this relationship between drug use and other variables that suggests an alternative measure of drug use. As will be seen below, the ways in which people control their use of drugs/alcohol is seen as an increasingly important factor in understanding the development of drug/alcohol problems. Conversely, it is becoming less common for the frequency and quantity of drug/alcohol use to be viewed as the only characteristic, or even as the most important characteristic, related to drug problems. Let us turn, therefore, to the literature on control of drug use.

The notion of "loss of control" is central to the concept of alcoholism. Keller (1972) states, for example, "He (the alcoholic) has become disabled from choosing invariably whether he will drink. That is the essential loss of control over drinking ... (T)here is no room for an alcoholism without loss of control. Without loss of control there is only a prealcoholism phase . . . " Since loss of control plays such a central role in the conceptualization of alcoholism, it is not surprising that loss of control is also the cornerstone of some treatment approaches to alcoholism, such as the Alcoholics Anonymous approach.

The data available on loss of control of alcohol use are contradictory, but they are contradictory in a consistent way. On the one hand, many experimental studies have failed to find evidence for craving or loss of control while investigating alcohol use (Cutter, Schwaab, and Nathan, 1970; Engle and Williams, 1972; Gottheil, Crawford, and Cornelison, 1973; Mello and Mendelson, 1971; Marlatt, Demming, and Reid, 1973; Ludwig, Wikler, and Stark, 1974; and MacDonough, 1976). Yet on the other hand, clinical reports continue to support the existence of the experiences of loss of control or of craving (Sobell, Sobell, and Christelman, 1972; Hore, 1974; Litman, 1974; and Ludwig, Wikler, and Stark, 1974; Madsen, 1974).

Further, though less direct, evidence for the importance of the concept of control in the use of alcohol comes from the research on personality variables. A large number of studies have repeatedly shown that alcoholics tend to be more field dependent than non-alcoholics; that is, they rely more on external cues in the environment than on internal cues in making judgments. (For extensive reviews of the research in this area, see Sugarman & Schneider, 1976; and Goldstein, 1976). Similarly, work with Rotter's locus of control test has demonstrated that alcoholics tend to score high on external locus of control, feeling that they are controlled by outside events (Distefano, Pryer, & Garrison, 1972; Lazarus, 1976; Oziel & Obitz, 1975; and Oziel, Obitz, & Keyson, 1972). All of this work on both field dependency and locus of control suggests that alcoholics see the world and react to the world differently from non alcoholics vis a vis control over their lives. Consequently, one would expect the manner in which alcoholics control or fail to control their drinking to differ from non-alcoholics.

Finally, there have been different approaches taken by survey researchers in their efforts to measure loss of control. On the one hand, Keller (1962) has argued that loss of control over alcohol use can be determined simply by measuring the problems that result from drinking. He assumes that only someone who has lost control will drink to such an extent that the drinking produces serious, continuing problems. On the other hand, investigators such as Cahalan & Room (1974) and Clark (1976) have attempted to measure directly respondents' experiences of losing control over drinking. However, there are several critical limitations to these important attempts. The items making up the loss of control scale, for example, may not be unidimensional. In fact, given the emphasis in the alcohol literature placed on the internal - external dimension of control, it is unfortunate that some items in the loss of control scale involve violation of an external standard (sometimes getting drunk at inappropriate times) while others deal with violation of internal standards (drinking "after I promised myself not to"). Furthermore, the loss of control scale raises an interesting paradox. A high score on the scale indicates that respondents drank even though they did not want to. The paradox is, how can individuals drink while not wanting to drink? The answer is probably that the respondents wanted to drink at the time that they were drinking even though they may have expressed the desire sometime before beginning or sometime afterwards of not wanting to drink. Is this loss of control? Perhaps it is merely evidence of inconsistent desires. In sum, it is not yet clear what the loss of control scales are actually measuring.

The concept of control has also become important in the study of narcotics use. Several investigators have discovered that, contrary to almost universally held beliefs, there exist "occasional" (controlled) users of heroin (Lindesmith, 1957; Scher, 1961, 1966; Chein et a, 1964; Alksne et a], 1967; Gay & Senay, 1973; Minkowski et al, 1972, Newmeyer. 1974; Powell, 1973; Levengood et al, 1973; Graeven & Jones, 1976; Robins, 1976; Robins et al, 1975; Zinberg, 1974; Jacobson & Zinberg, 1975; Zinberg, Jacobson, & Harding, 1975; Zinberg, 1975; Zinberg & Jacobson, 1976; Zinberg, Harding, & Winkeller, 1977). Following the initial recognition that controlled use of heroin is possible, Zinberg and his colleagues (see references above) are engaging in a series of studies aimed at understanding the nature of control, the various ways in which heroin use is controlled, and the many drug, situational, and personality factors that govern control.

Thus, the concept of control continues to play a central role in the study and treatment of alcoholism. At the same time, a new area of investigation, the study of controlled heroin use, has been attracting increasing attention. Nevertheless, the few efforts to date at devising measures of control (or more accurately, loss of control) have not proved very productive. Measures of how individuals control their use of substances, however, are becoming increasingly important. Further conceptual advances require additional data on control, and it is for this reason that a measure of control style was devised and tested in the present study.

METHOD

Measures

Consider two very different ways of controlling drug use. First, take the individuals who employ an external standard, such as tailoring the frequency and quantity of their drug use to the amount of drug use by others around them. In other words, if others are having a couple of joints of marijuana, an individual will have a couple of joints also, while if others have just one drink, then an individual will, too, have a single drink. Contrast these individuals with others who use an internal guide such as feeling high, mellow, etc., or those who do not stop using a substance until they feel guilty about using too much. The two types of individuals posed above may well differ in the amount of drug used, but their different styles of controlling use are probably related directly to whether they experience drug problems, even after the quantity and frequency of drugs used is taken into account. There is certainly ample anecdotal evidence that one way of avoiding substance -related problems is to regulate substance use with a relatively inflexible external standard. Conversely, much common wisdom holds that highly subjective internal standards become highly variable and are most likely to change in the direction of greater substance use. The classic example, of course, is tolerance, where ever larger quantities of a substance are necessary to produce the desired effect.

As a result of this line of reasoning, an exploratory measure was developed with the aim of obtaining an index of rules or styles that individuals use in judging how much of a substance to use. A list of 5 different rules was arrived at with the intention of representing the most common control styles. The following item was read to respondents for each of several substances that they used:

"We would like to know the way you decide on how much (insert name of substance) you use. Look over the choices on this card and tell me which number best describes how much (insert name of substance) you use on a typical occasion.

1. Roughly whatever amount others with me are taking.

2. The amount that I always take; like 1 cocktail, 2 cups of coffee, handful of nuts, 1/2 candybar.

3. 1 take until I get a strong feeling that I should stop; like feeling uncomfortably full or high, feeling guilty, feeling worried, and so on.

4. The amount that is supposed to do what I want, that is, the amount that instructions, doctors, friends, or others say will relieve pain, help me sleep, reduce hanger, and so forth.

5. 1 take until I feel the way I want such as good, full, relaxed, or until I have the experiences I want such as high, creative thoughts, and so forth."

The control measure was also compared with several other measures. First, a standard measure of drug use was included in the questionnaire. Respondents were asked:

On how many different occasions in the past 30 days, that is since (DATE) did you (insert name of substance)? An occasion is a period of use. For example, during lunch you might begin with a candy bar and finish with a pastry. That would still be just one occasion. But if you had a danish roll for breakfast & some cookies for a coffee break, that would count as 2 separate occasions. Respondents were then asked about drug problems in general and about dependency in particular. The questions read to respondents for each of the substances used were:

"Think about any problems or dissatisfactions that you are having-they could involve troublesome thoughts or feelings, difficulties with your health work, friendships, family, standard of living, and so on. Has your use of (SUBSTANCE) been a cause of any problems or dissatisfactions?"

"Would you say you feel completely dependent on (SUBSTANCE), very dependent, somewhat dependent, a little dependent, or not dependent at all?"

The Survey

The study consisted of a systematic random household survey of the Boston Standard Metropolitan Statistical Area in 1976. A total of 1,087 face-to-face interviews were conducted with adults (18 years or older) randomly selected from each household in the sample. The interview averaged about an hour in length, and questions covered use of a wide range of substances (common snacks, deserts, and drinks, tobacco, alcohol, non-prescription and prescription medications, and illicit drugs). To aid respondents with recognition of which drugs they had taken, they were shown charts containing color pictures of the more commonly prescribed drugs in each category of medications.

RESULTS

Results from the survey appear to successfully represent the adult population in the Boston Standard Metropolitan Statistical Area. Selected demographic variables, for example, are comparable to the 1970 census data, taking into account expected changes in the population over the past seven years. The response rate was 75%.

Since only one adult was selected from each household, regardless of the size of the household, a test was run comparing results based on the data weighted by the number of adults in the household with unweighted data. The differences were negligible. Hence, all analyses in this report are based on unweighted data.

Judgment of the utility of the new measure of control over amount of drug used depends on two types of issues. First, the patterns of responses on the measure must be meaningful and must be related to other measures in a sensible way. Second, the control measure must produce useful information beyond that provided by the traditional measure of occasions of use. Introduction of a new measure is unwarranted if it merely duplicates an existing measure.

The Control Style Measure

Responses to the control style measure do indeed form meaningful patterns. To begin with, the measure discriminates between the different substances asked about in the interview. As expected, respondents control their use of different types of substances in very different ways. Table 1 shows the percent of current (within the past 8 months) users of each of 20 substances who employ each of the five control styles. The modal response for all prescription and non-prescription drugs is "The amount that is supposed to do what I want, that is, the amount that instructions, doctors, friends, or others say will relieve pain, help me sleep, reduce hunger and so forth." By way of contrast, only 2% of marijuana users select this response. Instead, the modal response for marijuana users is "I take until I feel the way I want such as good, full, relaxed, or until I have the experiences I want such as high, creative thoughts, and so forth." Conversely, the modal response for the more common, legal recreational substances such as coffee, tobacco, and alcohol, is "The amount that I always take: like 1 cocktail, 2 cups of coffee, handful of nuts, 1/2 candybar." These responses are consistent and meet the expectations about differences between substances. Most people take the amount of medicine that their physician prescribes; most individuals use marijuana in order to get high, and many people regularly have their one cocktail every evening, a soft drink with lunch, and/or their 3 cups of coffee every day.

Responses to the control style measure are also related to the drug problem measures in a consistent, meaningful way. The pattern is most readily seen by first collapsing the five styles of control into two categories. The collapsing is warranted on both conceptual and empirical grounds. Three of the control styles involve reliance on an external standard for making the decision of how much of a substance to use. The three are a reliance on 1) the amount others are using, 2) a regular amount that is commonly used, and 3) the amount specified by someone else, such as a physician. Conversely, the two remaining control styles involve reliance on an internal standard: using until 1) 1 feel the way I want, or 2) 1 get a strong feeling that I should stop. Thus the internal controls require that the individual test some internal index that changes as a substance is used.

With few expectations, individuals who employ internal control styles score more negatively on the dependent measures, regardless of substance, than those relying on external control styles. In other words, respondents who rely on internal controls also tend to feel that they are more dependent on substances and that the substances are more likely to cause them problems in their lives, in comparison with respondents who rely on external controls. (More details of this relationship are spelled out below in the discussion of the multiple regression analyses.)

These results are certainly consistent with the point of view taken by Zinberg (for example Zinberg et al, 1977) who argues that the presence of social regulatory mechanisms results in the controlled use of substances, but when they are absent, compulsive use tends to occur. Prominent examples of these social regulatory mechanisms are external standards such as those which most individuals use to control their drinking-"Never drink alone; never drink in the morning; always have something to eat when drinking," and so on.

Finally, the control style measure may also serve in the study of drug use as a personality characteristic. Twenty-four percent of the sample employ a single style of control on all of the common recreational substances they use. These substances are sweets, ice cream, salty snacks, soft drinks, coffee-tea-cocoa, tobacco, and alcohol. The remainder of the sample uses different styles of control with different substances. This finding--that fully a quarter of the sample uses the same control style across several substances-suggests that efforts be made to integrate the present approach with personality research on locus of control and dependency.

In sum, the measure of control style meets the first criterion-the pattern of results Is consistent and meaningful. The next question, then, is whether the measure of control style adds to the traditional measure of number of occasions of use.

Table 1
Percent of Respondents Employing Each Control Style

Alternative control styles
1. Roughly whatever amount others with me are taking.
2. The amount that I always take; like 1 cocktail, 2 cups of coffee, handful of nuts, half a candy bar.
3. I take until I get a strong feeling that I should stop; like feeling uncomfortably full or high, feeling uilty, feeling worried, and so on.
4. The amount that is supposed to do what I want, that is the amount that instructions, doctors, friends or others say will relieve pain, help me sleep, reduce hunger and so forth.
5. I take until I feel the way I want such as good, full, relaxed, or until I have the experiences I want such as high, creative thoughts and so on.