<< Statistical Analyses >>

In our first volume, nearly all of the statistical calculations were based upon the male’s experience in sexual activity which had led to orgasm. Although the male is frequently aroused without completing his response, he rarely engages in such activities as masturbation or coitus without proceeding to the point of orgasm. On the other hand, a considerable portion of the female’s sexual activity does not result in orgasm. In consequence, statistical calculations throughout the second volume have shown, wherever the data were available, the incidences and frequencies both of the female’s sexual experience and of her experience in orgasm.

It is usually possible to secure data on the incidences of sexual experience that did not lead to orgasm, but it is often impossible to secure frequency data on such experience, because of the difficulty of distinguishing between non-erotic social activities—a simple kiss, for instance—and similar activities which do bring erotic arousal. On the other hand, orgasm is a distinct and specific phenomenon which is usually as identifiable in the female as in the male. It has served, therefore, as a concrete unit for determining both incidences and frequencies. The use of such a unit is justified by the fact that all orgasms, whether derived from masturbation, petting, marital coitus, or any other source, may provide a physiologic release from sexual arousal; but since there may be differences in the social significance of orgasms obtained from one or another type of activity, there has been some objection to the use of that phenomenon as a unit of measurement. There seems, however, no better unit for measuring the incidences and frequencies of sexual activity.

The following definitions are designed to give the non-statistical reader an acquaintance with the meanings of some standard statistical terms. For the technically trained reader, these definitions will show the way in which the terms have been applied in the male and female volumes.

Accumulative Incidence
An accumulative incidence curve shows, in terms of percentages, the number of persons who have ever engaged in the given type of activity by a given age. The calculations are, of course, based on the experience which the reporting subjects had had before contributing histories to the present study; but by securing information on the age at which each subject had first had experience, it is possible to determine the percentage of the total sample who had had experience at each age up to the time of interview. But an accumulative incidence curve may also be useful in indicating what percentages of any group might be expected to have experience if they were to live into the older age groups.

Most of the incidence data presented in research studies record the number of persons who are having experience in a given period of time, e.g., the current incidences of venereal disease, or the current incidences of marriage in the U. S. population. The accumulative incidences, on the other hand, show how many persons have ever had experience by a given age. This answers a type of question that is very frequently asked. One may want to know how many persons ever masturbate, how many persons ever have coitus before marriage, how many persons ever have homosexual experience.

A subject’s ability to report whether he has ever engaged in a given type of activity is not as liable to errors of memory and of judgment as his ability to recall frequencies of activity and to estimate the average frequencies of his activity. A subject’s willingness to admit experience, however, may be affected by society’s attitudes toward particular types of sexual activity—toward pre-marital coitus, extra-marital coitus, mouth-genital contacts, and homosexual contacts, for instance,—and in regard to such items, subjects may occasionally deny their experience. On the other hand, it is more difficult for a subject to exaggerate because of the difficulty in answering subsequent questions concerning the details of the professed experience. The accumulative incidence data are, therefore, probably a minimum record rather than an exaggeration of the experience actually had by the subjects in the study.

Active Incidence
The active incidences represent, in terms of percentages, the number of persons who have engaged in each type of sexual activity in a particular period of their lives. In our volumes, these have been standardly expressed as five-year periods, covering in most instances the years between adolescence and fifteen, sixteen and twenty, twenty-one and twenty-five, etc.

The active incidences for the various age groups provide highly significant data in any sex study, because the number of persons involved in a given type of activity depends upon age more than upon most other factors. However, it should be recognized that any individual who has had a single experience within the five-year period raises the percentage shown in an active incidence. In most instances the active incidences would have been lower if they had been calculated for one-year instead of five-year periods. A better comprehension of the extent of any type of sexual activity may be had if one considers the active incidences in conjunction with the average frequencies among those who are having any experience (the active median frequencies).

Frequency of Activity
The frequency of each subject’s activity is recorded as an average frequency for each of the five-year periods specified above. Throughout our volumes, frequencies have been expressed as average frequencies per week in each of those periods. The weeks or years in any five-year period which were without sexual activity have been averaged with the weeks or years in which there was activity, and in that way periods of inactivity have lowered the average rates in such a five-year period. Unless the data are specifically designated as applying to pre-adolescence, all of the frequency calculations have been based upon activities which occurred after the onset of adolescence. In consequence, the first age period extends from adolescence to fifteen, and in that case the average frequencies are based on the number of adolescent years and are not reduced by being averaged with the preadolescent years. If the last age period—the one in which the subject contributes his or her history, or changes his marital status—is less than a full five-year period, it has been treated in the same fashion as the first adolescent period. No frequency calculations have been made for persons who belonged to a given age, adolescent, or marriage period for less than six months.

Because of the difficulties which most persons have in estimating average frequencies for experience which may have been sporadic or irregular in its occurrence, frequency data are subject to much greater error than incidence data. This explains why there have been few attempts in previous studies to determine the frequencies of sexual behavior. But even though considerable allowance must be made for errors in the frequency data, they show what appear to be significant correlations with age, the decade of birth, the religious associations, and still other social factors in the backgrounds of the females in the sample.

Frequency Classes
In all calculations in the first and second volumes, individuals have been grouped in frequency classes which have been named, throughout both our volumes, for their upper limits. Since it is current practice in most statistical publication to name frequency classes for their lower limits, attention should be drawn to our different practice. The ranges and the mean values of the frequency classes as we have defined them in our preceding and second volumes, are as follows:

CLASS	RANGE	MEAN VALUE
0	0	0
0.09	0.0-0.09	0.05
0.5	0.1 -0.5	0.3
1.0	0.6-1.0	0.8
1.5	1.1-1.5	1.3
2.0	1.6-2.0	1.8
etc.
10.0	9.6-10.0	9.8
11.0	10.1-11.0	10.5
12.0	11.1-12.0	11.5
etc.
28.0	27.1-28.0	27.5
29.0+	28.1 and higher	28.5

Since most persons report frequencies in terms of whole integers, and since we have used the upper limits of each class to designate the frequency classes into which such reported data are placed, our calculations of both median and mean frequencies have, in actuality, been more conservative than they would have been if we had used the lower limits to designate each class. An individual who reported an average frequency of a given type of sexual activity at 2.0 per week would go, in our calculations, into a class which had a mean value of 1.8 per week. If the lower limits had been used to designate each class, that same individual would have gone into a group which had a mean frequency value of 2.2 per week. However, the differences in the averages obtained by these two methods of calculation are slight and usually immaterial in terms of the quantities being measured.

Median Frequency
When the individuals in any group are arranged in order according to the frequencies of their sexual experience, the individual who stands midway in the group, the median individual, may be located by the formula: Md = (n+1)/2

Throughout such statistical formulae, the symbol N or n stands for the number of individuals in the group. While the median is an average which is not often calculated by people in their everyday affairs, it is a useful statistic because it is unaffected by the frequencies of activity of the extreme individuals in any sample. The mean, which is the sort of average that most persons ordinarily calculate, is affected by extreme individuals.

Active Median Frequency
In any group and especially in any group of females, there may be some individuals who are not having any sexual activity of the sort with which the calculation is concerned. That portion of the total sample which is actually having experience or reaching orgasm has been identified in the present study as the active sample. The median individual of this active sample has a frequency of experience or orgasm which we have identified throughout this volume as the active median frequency.

Total Median Frequency
The entire sample in any group, including both those individuals who are not having experience of the sort under consideration and those who are involved in that particular type of sexual activity, constitutes the total sample as we have used the term in the present study. The median individual in such a total sample has a frequency of activity which we have identified as the total median frequency.

Where less than half of the individuals in a sample are involved in a given type of activity—where, for instance, less than half of the individuals in a given group are petting to the point of orgasm—the median individual is, of course, not having any activity at all. In consequence, the median frequency for such a group is zero. Since the active incidences of many types of sexual activity among females are frequently less than 50 per cent, it has not been possible to calculate total median frequencies on more than a few of the types of sexual activity discussed in the second volume. They have been calculated chiefly for marital coitus and for the total sexual outlet, because more than 50 per cent of the females are actively involved in those activities in most of the age groups.

Mean Frequency
A mean frequency may be determined by totaling the measurements (in the present study, the total number of experiences or total number of orgasms) in each group, divided by the number of individuals in the group. The process is summarized in the formula: M=(Efv)/n

The mean is the sort of average which is most commonly employed by most persons in their everyday affairs. If one wants to find the average price which has been paid for a number of articles, this is done by totaling the individual prices and dividing by the number of objects bought. Such an average is the mean of the various prices which were paid. Conversely, the total amount of money spent may be calculated by multiplying the mean price by the number of objects which were bought. In the same fashion the total number of orgasms experienced in any group may be determined by multiplying the mean frequencies of orgasm by the total number of persons in each group. The mean, therefore, serves a function which is not served by the median.

On the other hand, means often give a distorted picture because their values may be considerably raised by a few high-rating individuals in the group. Since there is usually a tremendous range of variation in the frequencies of sexual activity or of orgasm in any group of females, the mean frequencies are uniformly higher than the median frequencies of sexual activity. Since the range of individual variation is more extreme among females than it is among males, mean frequencies calculated for females are even less adequate as measures of sexual activity than they are for males. Consequently mean frequencies have been used in the second volume only when we wished to calculate the total number of experiences, or the total number of orgasms occurring in a whole group.

Active Mean Frequency
The mean frequencies for the females in an active sample—those females who were having any experience, or experience to the point of orgasm—have been designated as the active mean frequencies.

Total Mean Frequency
The mean frequencies of the females in any total sample, including those who were not having experience or orgasm as well as those who were having such experience or orgasm, represent the total mean frequencies, as they are designated in the first and second volumes.

Percentage of Total Outlet
In the second volume we have systematically calculated what proportion of the orgasms experienced by each group had been derived from each type of sexual activity. Thus, in a given age group of white females of a particular educational, re-ligious, or other background, we have calculated what proportion of the total number of orgasms occurring in the group had been derived from masturbation, from nocturnal dreams to orgasm, from pre-marital coitus, etc. Obviously, the sums of the percentages shown for the several types of sexual activity must constitute one hundred per cent of the total outlet (the total number of orgasms) of the group.

We have not calculated the percentage of the total outlet which was derived by each individual from each type of sexual activity, and then calculated averages based on those individual data, as we did at a few points in our volume on the male.

Age
For each type of activity, the ages at first experience and the ages during which there was subsequent experience have been standardly recorded and calculated for each year in each history. However, as a matter of economy, and as an aid to the comprehension of the total picture, in our volume on the female the accumulative incidence data have been published only for each fifth year, except when there were unusual or marked developments in the intervening years. In the latter event, we have published the record for those years.

Marital Status
In most calculations in this volume, sexual activities have been classified as occurring among single, married, or previously married females. Individuals were identified as single up to the time they were first married. They were identified as married if they were living with their spouses either in formally consummated legal marriages, or in common-law relationships which had lasted for at least a year. They were classified as previously married if they were no longer living with a spouse because they were widowed, divorced, or permanently separated. These definitions are more or less in accord with those used in the U. S. Census for 1950, except that common-law relationships have been more frequently accepted as marriages in our data, and we have considered any permanent separation of spouses the equivalent of a divorce.

Educational Level
On the basis of the educational levels which they had attained before completing their schooling, the females in the sample have been classified in four categories, as follows:
   0-8: those who had never gone beyond grade school
   9-12: those who had gone into high school, but never beyond
   13-16: those who had gone into college, but had not had more than four years of college
   17+: those who had gone beyond college into post-graduate or professional training

It was obviously impossible to determine the educational level that would ultimately be attained by subjects who were still in grade school or high school at the time they contributed their histories, and they were consequently unavailable for any calculation that involved an educational breakdown. Persons still in college were classified among those having 13 to 16 years of schooling, and this may have involved a small error because a portion of them would ultimately go on into graduate work.

Upon calculation, we find that educational backgrounds do not seem to have been correlated with the patterns of sexual behavior among females as they were among the males covered by our first volume. Consequently in the second volume we have published direct correlations with most of the other factors, such as decade of birth and religious background, without showing the preliminary classifications which we have made on the basis of the educational levels.

Occupational Class of Parental Home
Since the female’s social status after marriage depends on the occupational class of her husband as well as upon her own social background, it has not proved feasible to make correlations, as we did in the case of the male, with the female’s own occupational rating. However, in the second volume we have correlated the incidences and frequencies of her sexual activities with the occupational class of the parental home in which she was raised. If the parental home had changed its social status during the time that the female lived in it, she was given a rating in each of the occupational classes. The occupational classes have been defined as they were defined in the case of the male:
   (1)    Underworld: deriving a significant portion of the income from illicit activities
   (2)    Unskilled laborers: persons employed by the hour for labor which does not require special training
   (3)    Semi-skilled laborers: persons employed by the hour or on other temporary bases for tasks involving some minimum of training
   (4)    Skilled laborers: persons involved in manual activities which require training and experience
   (5)    Lower white collar groups: persons involved in small businesses, or in clerical or similar work which is not primarily manual and which depends upon some educational background
   (6)    Upper white collar groups: persons in more responsible, administrative white collar positions
   (7)    Professional groups: persons holding positions that depend upon professional training that is beyond the college level
   (8)    Persons holding important executive offices, or holding high social rank because of their financial status or hereditary family position
   (9)    Persons living primarily on income and occupying high social status because of their monied position and/or their family backgrounds

Decade of Birth
Correlations with the decade of birth have been based, in the second volume, primarily on the four following groups:
   Bom before 1900
   Born between 1900 and 1909
   Born between 1910 and 1919
   Born between 1920 and 1929

The decade of birth has proved to be one of the most significant social items correlating with the patterns of sexual behavior among American females. In many instances the youngest females in the sample, born since 1929, had not developed their patterns of sexual behavior far enough or been married long enough to warrant their inclusion in any of the comparisons of generations.

Age at Onset of Adolescence
Correlations have been made with the age at which the female showed the first adolescent developments. The classifications have been as follows:
   Before and at 11 years of age
   At 12 years of age
   At 13 years of age
   At 14 years of age
   At 15 years of age or later

Rural-Urban Background
The subjects in the present study have been classified as having rural backgrounds if they lived on an operating farm for an appreciable portion of the time between the ages of twelve and eighteen. This is the pre-adolescent and adolescent period which is of maximum importance in the shaping of sexual patterns. The more extensive classification of rural and urban backgrounds given in our volume on the male would have provided a more satisfactory basis for correlations, but unfortunately we do not yet have enough histories of rural females to allow us to make such an intensive study.

Religious Background
Subjects in the present study have been classified as Protestant, Catholic, or Jewish, or as belonging to some other group. In these other groups (1.3 per cent of the total sample) the number of histories is too small to allow analyses. In each of the three religious groups, the subject has been classified as devout, moderately religious, or religiously inactive, in accordance with the following definitions:
   (1)    DEVOUT: if the subject is regularly attending church, and/or actively participating in church activities. If Catholic, frequent attendance at confession is a criterion; if Jewish, frequent attendance in the synagogue, or the observation of a significant portion of the Orthodox custom.
   (2)    MODERATELY RELIGIOUS: if the subject attends church or engages in church activities with fair frequency, or attends confession in the Catholic church, or follows the Orthodox custom to some degree which, however, is less than that of strictly devout groups.
   (3)    RELIGIOUSLY INACTIVE: if the subject only infrequently, or rarely, or never attends church or engages in church activities, or infrequently or never goes to confession, or observes few if any of the Orthodox Jewish customs. Such persons may still, however, be related to one or another of the religious groups through their parents, through their own earlier training, or through their own current thinking. There have been exceedingly few persons in the study who are as completely irreligious or agnostic as they often insist.

It has been possible to correlate the sexual data on these histories with the current religious status of each subject, but it has not yet been possible to make correlations with their earlier religious connections. The earlier connections may have been the more significant in affecting the subsequent patterns of sexual behavior, but we will need a more extensive series than we yet have before we can undertake further analyses.

There are obviously still other factors which affect patterns of sexual behavior and which deserve investigation. For instance, the age at which the female marries may markedly affect the nature and the extent of her pre-marital experience, particularly in pre-marital petting and in pre-marital coitus. Consequently we have made correlations with the age of marriage at certain points in the second volume. On the other hand, some of the other correlations that may ultimately deserve investigation, such as the correlation with the geographic location and the type of community in which the subject was raised, cannot be made until we have secured additional data.

U. S. Estimates
In our volume on the male we attempted to estimate the incidences and frequencies of the various types of sexual activity in the total U. S. population, by weighting and combining the data for the age groups, the single, married, and previously married groups, the educational groups, and the rural and urban groups in the sample, in accordance with the incidences of those groups in the United States Census of 1940. Such a correction seemed desirable in order to prevent the general reader from making his own extensions of the data without any correction for census distributions.

The research scientist may choose between two possible procedures in presenting his data. He may present the material as a description of his findings on the particular individuals with which he worked, and make no suggestion concerning any possible extension of his generalizations to any larger group; or he may attempt, by some statistical or other technique, to discuss the applicability of his findings to some group larger than the one with which he actually worked. If he follows the first procedure, he imposes upon the reader the task of deciding how far to extend the specific findings, and the reader is usually more poorly equipped than the scientist to decide that. For instance, if we had not attempted a U. S. correction in the male volume, our data on the incidences and frequencies of masturbation, because of the larger representation of college males in the sample, would have given an exaggerated picture to any reader who took them to be typical of American males in general; and our data on the incidences and frequencies of pre-marital coitus would have represented a gross understatement for the total population because of the lesser representation of grade school and high school groups in the sample.

We have not, however, undertaken to do U. S. corrections in the second volume, because our sample of females is even more inadequate than our sample of males in representing lower educational levels, rural groups, and some of the other segments of the population. The generalizations made throughout the female volume have, therefore, been restricted to the particular samples that we have had available. Major changes might have been introduced into the generalizations if we had had a larger sample of females who had never gone beyond grade school, but we cannot suggest what those changes might have been. Meanwhile, the samples of high school, college, and post-graduate groups are of some size, and the generalizations drawn for the sample may not be too far from the actuality for those segments of the American population.

>>