OBTAINING COOPERATION FROM THE SCHOOLS

One of the most important problems in this study was gaining the cooperation of an official at each college who would give permission for the study to proceed. Since the random sample was only of the Ethics Department (pseudonym for a social science department), the researcher first contacted the department chairman who either gave his approval, requested further information (letter from thesis advisor and copies of the questionnaire) or referred the decision to the dean of students.

Table 27 presents the outcome of contacts with the twenty-nine colleges that were ever part of the sampling frame.

Table 27 shows that twenty-one of twenty-five, or 84%, of the contacted colleges participated in the study. Permission was usually obtained easily and quickly from the department chairman (fifteen of seventeen, or 88%) with the deans of students somewhat less willing to cooperate (five of seven, or 71%).

Chairmen who refused did so for the following reasons: One Ethics department chairman at a college in New York City refused because "there is too much use of students as guinea pigs for Ph.D. students who need some data for their thesis." That statement came at the end of a very brief telephone conversation in which the chairman told the researcher to make contact with one member of the department who was doing research on drug use, but that a random sample of classes in the Ethics Department was out of the question. Perhaps contact with the instructor mentioned might have provided access to the department at a later time, but since a very similar college was part of the sample, it was not felt to be crucial that this college be sampled. Another department chairman at a college outside New York City, after considering the questionnaire with department members and the dean of students, denied cooperation, because the college was planning its own briefer study of drug use on the campus and felt that there would be too much overlap.

The two deans who refused to permit the cooperation of their colleges did so because it was university policy not to allow surveys to take up class time, and this rule applied to their own faculty members as well. The researcher could have distributed questionnaires in classes and had the students return the answer sheets by mail, but this was not done.

At one college permission to conduct the study had been obtained and three classes had completed the questionnaire when a student strike was called to protest the extension of the Vietnam war into Cambodia and the killing of students at Kent State University. This prevented the inclusion of additional classes at the college. There were five other colleges included in the sample, but lack of time to make contact and administer the questionnaire prevented the researcher from ever contacting these schools. Thus, obtaining cooperation from most of the colleges in the sampling frame was relatively easy, and the few refusals did not greatly harm the study.

OBTAINING COOPERATION FROM CLASSROOM INSTRUCTORS

Once the department chairman and/or dean had given his permission to contact instructors, the problem of obtaining access to the college classes themselves arose. Each instructor was contacted by a letter that indicated which class was included in the sample, a suggested date, and a postcard for the teacher to indicate his cooperation or refusal. (The letter and postcard are reproduced on pp. 248 and 249.) Table 28 indicates the rate of refusals and cooperation of all classes that were ever part of the random sample (not including non-Ethics Department classes that were not part of the random sample, where the refusal rate was much higher).

Table 28 shows that cooperation was the most common response. Over 70% of all instructors contacted in the Ethics Department were willing or actually cooperated in the study. Most teachers who refused to cooperate did so on the grounds that they could not spare a whole class hour in the class chosen. Three of the professors who refused suggested one of their other classes as a substitute. The most interesting fact about teacher refusals is that 75% of them occurred after the first six weeks of class. Only 3% of the teachers who were contacted before that time refused. The vast majority of refusals came either during the midterm-exam time or afterwards, when teachers realized how far behind they were in their classes.

Many of those instructors who refused to set aside a whole hour would have allowed the researcher to distribute the questionnaires to be completed by the students at home. However, at the beginning of the semester the researcher did not permit students to take the questionnaires home because it would have meant lost questionnaires (which were valuable at the time) and a departure from the original research design. Toward the end of the semester, the only way any cooperation from some instructors could be obtained was by distributing the questionnaire in class. Students completed it at home, and returned the answer sheets to the instructor, who sent them to the researcher.

There was a greater tendency for instructors at upper-status (or quality) colleges to refuse either to set aside an hour of class time or to request that students be allowed to take the questionnaires home to complete. But cooperation was still the most frequent response even at these schools.

OBTAINING COOPERATION FROM STUDENTS IN THE CLASSES

The cooperation of students, however, was the most important factor in the success of the study. Most of the classes had been informed ahead of the actual administration of the questionnaire, and many instructors had let the students vote on taking the questionnaire. In some classes an undercurrent of hostility was evident when the class had not been informed beforehand. It would appear that the best way to obtain student consent is to have the instructor inform the students ahead of time and perhaps let the class vote on whether or not to take the questionnaire; in only three classes did the students vote not to take the questionnaire, while probably 30-50 classes voted to take it. However, it is not a good idea for the instructor to announce the exact date for the administration of the questionnaire. It appeared to the instructors that more of the conservative students and also radical students tended to be absent when they knew the questionnaire would be given on a particular day. But in terms of actual response rates and the rate of drug use in various classes, it did not make much difference whether the instructor announced the questionnaire ahead of time or on the day it was administered. The problem of students who stay away from class only to avoid the questionnaire appears to be a relatively minor problem.

A much more serious problem (far more serious than originally anticipated) is the low rate of class attendance at almost all colleges. It appears that a significant number of students either do not attend class at all or do so very irregularly. It is probable that many heavy drug users are among this group of nonattenders, although there are no good statistics to prove or disprove this hypothesis. At the beginning of the study it was anticipated that the rate of nonattendance would be from 15% to 30% on any given day. Experience tends to indicate, however, that the rate of nonattenance in college classes runs from about 20% to 50% with the average class being in the 24%-45% nonattendance range. Further, attendance slips toward the end of the semester. This high rate of nonattendance is the major unknown bias in the study. Almost all classes and colleges in this sample have similar problems of nonattendance but upper-quality colleges appear to have better attendance than medium- and low-quality colleges. Unfortunately, accurate statistics on the number of students registered in each class were not obtained, although it was available from the instructor, so that no accurate estimate of nonparticipation in this study can be made. Nonattendance was probably the most serious defect in the use of college classes as the basic sampling unit.

The main advantage of classroom administration lies in the low rate of direct refusals by students in the class. Of the students attending class on the day that the questionnaire was administered, only 3% either walked out after hearing about the questionnaire, failed to complete the questions on drug use, or handed in a blank answer sheet. Thus 97% of the students present actually completed the questionnaire, or enough of it for their responses to be included in our analysis.

On the whole the present study was reasonably successful in obtaining comparable data for a representative sample of students in the Ethics Department at each of the colleges included in the sample design. The best evidence that the sample was accurate is the small range of differences between colleges on selected variables such as sex, religiosity, sexual activities, drug use, etc. Colleges that are similar in quality or type of control have similar drug-use rates, etc. (Table 2).

NONCOOPERATION OF CLASSES OUTSIDE THE ETHICS DEPARTMENT

Although most of the researcher's time was spent obtaining cooperation from the randomly selected classes in the Ethics Department, he also tried to give the questionnaire in other departments but had little success. The amount of effort required to make and sustain contacts with department chairmen was consider. able. In one low-quality college where the dean's permission had been obtained, a serious effort was made to obtain cooperation of nonethics classes. The researcher made contact with department chairmen in mathematics, arts, English, physical education, and business, by telephone or in person. The chairman of the Math Department called and talked to the dean about the study but finally refused to cooperate, because he felt that instructors could not justify setting aside one hour of class for the questionnaire. The Art Department chairman refused to cooperate. The English, physical education, and accounting chairmen took the study under consideration, but none replied positively or negatively to repeated phone calls. The only cooperation obtained was in the Data Processing Department. At two other schools, the Physical Education Department was formally approached. The chairman at one school refused almost immediately. The chairman at the other college felt that approval from the dean was necessary. Such permission was eventually obtained and a couple of physical education classes were sampled. In short, the lack of cooperation in nonethics classes was great and the little cooperation that did occur was the result of considerable work; it is doubtful that it was worth the effort. Perhaps cooperation outside of the Ethics Department might have been higher at the beginning of the semester and particularly if the questionnaire had been shorter (five or ten minutes).

OBTAINING A SUBSAMPLE OF HEAVY-DRUG-USING FRIENDS

The original sample design called for a greater proportion of heavy drug users than would be obtained in the Ethics Department sample. The major technique for obtaining such heavy drug users was to ask students in all classes sup/eyed to take questionnaires to their drug-using friends. While the students were completing the questionnaire, the researcher would hold up a brown manila envelope, containing a copy of the questionnaire, three answer sheets, a pencil, and a stamped self-addressed envelope to be returned to the researcher, and say to the class, "If you have a friend who has used more than six times either methedrine, cocaine, LSD, other hallucinogens, or heroin, and you feel quite certain that he would complete this questionnaire, would you please pick up one of these brown envelopes which contain all the materials he needs to complete the questionnaire and return the answer sheet to me." This proved to be both a beneficial strategy for increasing the number of heavy drug users and a waste of resources. About 5%-15% of each class picked up the brown envelopes, which resulted in a total distribution of 400-500 questionnaires. Since there were three answer sheets in each packet, up to 1500 persons could have responded to the questionnaire. However, only 165 answer sheets from less than 100 questionnaire packets were received, and so, putting multiple answer sheets in the packets proved to be a good idea. However, the response rate (25% or less) from the 400-500 packets handed out was very low. It is not clear why the response rate was so low, but it is probable that many persons from the classes failed to get the questionnaire into the hands of drug-using friends. Although the response rate from all the questionnaires handed out was very unsatisfactory, the persons who did return the questionnaires were very heavy drug users when compared to the random sample. Thus, this strategy significantly increased the size of the heavy-drug-using sample.

CODING AND CHECKING OF RESPONDENT'S ANSWER SHEETS

The respondents marked their answers on a specially printed answer sheet that could be read by an Optical Scanning Machine 100 (at the Teacher's College Computer Center, Columbia University). This machine is very accurate in its sensing of marks; it picks up the slightest mark in the sensing area and converts it into a punch. It does the mark-to-punch conversion with more than 99.5% accuracy. However, it is so sensitive that it will convert well-erased marks into punches; it is almost impossible to erase a mark well enough so that it will not be converted into a punch by the Optical Scanner. Thus, it was necessary for the researcher to eliminate such erasures with Snopake (a white liquid used by typists). At the same time, the researcher examined each answer sheet for obvious lying, extremely inconsistent answers, and for inaccurate answers to single item questions.

From the more than 3500 answer sheets obtained, only 9 had to be eliminated from the active sample. Of these answer sheets 5 were refusals of persons who answered very few questions, or failed to answer the questions on drug use, or several other important questions such as SES, age, sex, etc. Four answer sheets had to be eliminated because of extremely inconsistent answers or what appeared to be random marking of answers. For example, one respondent claimed qo be black, male, a conservative jew, sixteen years of age, but had started smoking pot at age eighteen. Another respondent marked that he had used almost every drug with considerable regularity, but at other points in the questionnaire where check questions were asked, he claimed never to have used any drugs. One respondent in an all-male class claimed to be female, fifteen years of age, but had started to use drugs at ages eighteen and nineteen. When a questionnaire was eliminated from the sample because of lying, therefore, the contradictions were obvious ones.

Actually, though, the rate of lying as measured by multiple checks on drug questions appears to have been very low. Contradictions were found in less than 2% of the questionnaires; this could have easily occurred if a few students mistakenly marked the wrong question. The answer sheets discarded from the sample made up less than 0.3% of the number of questionnaires returned, a very low rate of nonusable questionnaires in a self-administered interview.

A more difficult problem was that students did not always mark the questions correctly, and some of the marks they made had to be eliminated in order for the question to be answered properly. For example, the students were asked at what age they had first been involved in certain activities (see questionnaire, questions 42-49). Some students marked several ages for each activity. The researcher eliminated all answers but the youngest age at which a respondent said he had been involved in the activity. Questions on the recency and frequency of drug use (questions 81-87) were supposed to have one answer, but many respondents marked two answers (one for each time period, before and after July 1, 1969), so the researcher kept the most recent use of the drug and eliminated the answer for the period before July 1, 1969.

Thus, the sample actually obtained suffered from several problems that had to be dealt with as efficiently as the limited time and resources available to the researcher permitted. With more resources it might have been possible to adhere more closely to the original sample design. The sample appears to be adequate, however, to measure the intercorrelation of variables. In sum, the sample actually obtained was essentially what was sought.

Columbia University in the City of New York I New York. N.Y. 10025
BUREAU OF APPLIED SOCIAL RESEARCH    5O5 West 115th Street

Dear Professor John Doe,

I am presently involved in research which deals with the use of legal and illegal drugs among students attending several colleges in the New York City area. I would like to obtain your cooperation and the cooperation of your students in aiding me in this important research.

In February or March 1970, I will be administering precoded questionnaires in several classes in your college. The classes have been selected by using a table of random numbers. One of these randomly selected classes is your class of Soc. 50.1B meeting TThF 10 in 207 A.

Since your class was chosen as part of a scientific random sample, it is hoped that you will be able to cooperate. Most students have found participation to be an interesting experience and have willingly cooperated. The questionnaire takes about 40 - 50 minutes to complete.

Since drug use is a controversial issue, I realize that you are concerned about anonymity and confidentiality. I have carefully considered this problem and have devised means of assuring the complete anonymity of all respondents. The accompanying letter will be attached to the questionnaire to assure students of their anonymity.

Your cooperation and that of you± students is requested. Would you please indicate on the enclosed card whether you would be able to cooperate in this research and the time which you would prefer. I have indicated a date which would be most convenient for me; if you are unable to cooperate at that date, please indicate a time which would be most convenient for you.

If you have any questions or wish to obtain more information, please make a note on the enclosed card. I will be most happy to supply such information. I have discussed this study with your Department Chairman and have obtained his cooperation. His secretary has copies of the questionnaire if you wish to see it.

Sincerely yours,
Bruce Johnson
Lecturer
Department of Sociology
Barnard College
Phone
Days    - 280 - 2159.

This card was sent to each professor suggesting a date for sampling his class. He was to indicate his willingness (or not) to cooperate, sign, and return the card to the researcher.