Our control group was obtained primarily through a sampling technique described in our previous volumes. We selected a well-defined social or professional group and attempted to interview every member. Realizing the danger of selectivity in sampling on a voluntary basis, we worked with each chosen group, returning again and again, if necessary, until we had interviewed at least half (and usually nearer three quarters) of its members. We do not stop when we have interviewed those who came to us freely or with little prompting; we continue with what at times must be annoying persistence, obtaining the case histories of persons who come to us as a result of the pressure exerted by other group members, their conscience, and our polite harassment. Once we have the opportunity to talk with these reluctant and hence especially valuable individuals, they almost always become quite cooperative and ascribe their previous foot-dragging to their ignorance of our methods and attitudes. Approximately one third of the control-group individuals derive from groups wherein we interviewed 90 to 100 per cent of the constituent members.
Dr. Alfred C. Kinsey, who began and directed this research until his death in 1956, was never impressed by the desirability of keeping a record of refusal rates—the proportion of those who were asked for an interview but who refused. He felt that an individual should not be considered a refusal until he or she had been the recipient of Dr. Kinsey’s persuasiveness, which was extraordinary. However, throughout most of the history of the research, and particularly in the earlier years, the interviewers were frequently in a position which made the recording of refusals almost meaningless; when beginning with a new group of subjects it is imperative to interview as large a number as possible in as short a time as possible so that there will be a sizable proportion of the group that can report favorably on the experience and influence others to contribute their histories. To use a military analogy, once a breakthrough has been achieved there must be no delay in pressing onward. This means that one must take first those who are most willing to be interviewed. Afterward one can afford to return to those who originally were reluctant and spend a considerable amount of time on a second or even third solicitation.
Let us say, for example, that we have been working with a large group, one too large to interview in its entirety. We have interviewed 50 per cent of the members; we feel that we have sufficiently diluted the volunteer bias found among those who were anxious to contribute their histories; we are not particularly interested in having more from the group, hence we stop. Now the question arises, inasmuch as the whole group was solicited (even though indirectly), is the refusal rate
50 per cent? Or are only those who were individually and briefly solicited to be considered refusals? Or should we count only those who rejected the second and serious solicitation? The problem is in some instances hopelessly confused by the fact that we often do not record the name of the person solicited and cannot determine whether he or she subsequently gave a history to another interviewer.
Nevertheless, it is indefensible thus to excuse ourselves from considering the problem of refusals as a statistical factor. Consequently, in obtaining the “hospital sample,” the only major interviewing endeavor in recent years, we arranged our sampling procedure so that an accurate account of refusals could be made. In this sample we first ascertained the males who were eligible for inclusion in the control group, and we then solicited each person in specific terms so that a clear-cut acceptance or refusal resulted. Out of 164 requests we suffered six refusals; also two men began interviews but refused to finish.
While we have dealt thus far with the question of sampling within the groups chosen for this purpose, we have not touched on the equally important question of which groups were chosen and why. Since at this point our concern is with the 477 males constituting the control group, this is no place in which to discuss in detail the sampling procedure of the entire research, and a brief description must suffice. The sampling procedure was to ascertain the weaknesses of our existing sample and with these in mind to select, from whatever groups seemed available to us, those that would best remedy the weaknesses. For example, at one time we realized that we had very few older, never-married women in our total sample; consequently, we arranged matters so that we were invited to address an educational conference whose membership included a large number of never-married female teachers, many of whom were of the age we desired.
While we are satisfied with our sampling within each chosen group, we are not satisfied with our sample of groups. Various imbalances in terms of age, education, religion, and other factors exist. In selecting our control group—which we wished to make as similar as possible to the prison group and sex offenders with regard to socioeconomic status —these defects hampered us. However, in view of the magnitude of our task and the limited means at our disposal we feel no apologies are called for.
With two exceptions, the term “sampling procedure” can scarcely be applied to our prison group. In the early stages of the research, when much interviewing was being done at Indiana correctional institutions, Dr. Kinsey did not view the inmates as a discrete group that should be differentiated from people outside; instead, he looked upon the institutions as reservoirs of potential interviewees, literally captive subjects.
This viewpoint resulted in there being no differentiation in our 1948 volume between persons with and without prison experience. By 1953 when we published our second volume we had found that the sexual behavior outside prison displayed by persons previously or subsequently imprisoned differed from the sexual behavior of persons never imprisoned; consequently, persons with prison experience have since been treated separately. At any rate, the great majority of the prison group was collected omnivorously without any sampling plan—we simply interviewed all who volunteered and when this supply of subjects was exhausted we solicited other inmates essentially at random.
One instance of planned sampling in the prison group occurred at San Quentin when Dr. Gebhard and Dr. Pomeroy asked the prison administration to select an inmate group it felt would best represent the total inmate population. The administrators chose those who were working at the quarry; here the only known bias was that the men could not be serious escape risks and had to be in at least average physical condition. There were some 50 in all, and we made an attempt to interview each one. Of these men, three refused and one, through some oversight, was not solicited; counting him, this gives a refusal rate of 8 per cent. This quarry sample we compared to the total prison group on a small number of behavioral items and the differences appear inconsequential.
The second instance of planned sampling took place at Soledad where all males who appeared obviously homosexual or who proved to be problems because of their homosexuality were segregated in one wing of a building—Z wing. We decided to sample Z wing as thoroughly as possible, and succeeded in interviewing 111 males—everyone in Z wing—plus two who had recently been returned to the general inmate population.
The procedures used in accumulating our sample of sex offenders varied according to circumstances, and each institution must therefore be dealt with separately. In each we found more sex offenders than were officially listed; men are listed in prisons on the basis of their current offense. Prior offenses, unless notable, are often forgotten and some minor ones may not appear on official records.
At the Indiana State Farm we had no plan of sampling—we simply sought out sex offenders and, after a time, avoided the more common types of offense (e.g., statutory rape) and directed our efforts toward the rarer types.
At San Quentin, where we interviewed intermittently from 1949 to 1954, we originally followed a rather simple pragmatic plan: to obtain a numerically adequate sample of every type of sex offender. This led, naturally, to an eventual emphasis on the less common types. Toward the end of the San Quentin interviewing, in November 1953, we inaugurated a more rigorous sampling plan. We first obtained a census of the inmates currently serving sentences for sex offenses and then decided to attempt to interview all the men convicted of certain unusual offenses and to make a random sample of men convicted of other commoner offenses.
*15\161\2*
You must be logged in to post a comment.