Shape in the Kinsey interviews was recorded in a 3x3 category system consisting of 3 variations of shape when viewed from the side (curved up, straight, curved down) and 3 when viewed from above from the subject's point of view (curved left, straight, curved right).  The 9 resulting categories were retained for the present research except that the labels for the side-view categories, curved up and curved down, were changed in this report to u-curve and n-curve respectively.  This change was made because we felt the words "up" and "down" were likely to elicit in the reader associations related to the angle rather than to the curve of the erection.   Shape in the new data of the present research was determined from drawings the respondent made on the questionnaire, from an answer to a multiple choice question, and from photographic evidence.  Values analyzed on shape (a categorical variable) were the category of agreement or best evidence.

Angle in the Kinsey interviews referred to the angle of the erection from the body when the man was standing.  The self-report of the subject was recorded in 30º increments, with one major exception.  The horizontal position was reserved as a separate category and covered only 10º.  The two lowest 30º segments were combined in the Kinsey Institute data summary, presumably because the research team found so few erections reported in this range.  Figure 1 (imagine a man standing in profile with the front of his body against the flat side of the figure) illustrates the resulting Kinsey angle categories.  In the present research we did not constrain the reporting of angles to predetermined categories but allowed each subject to draw the angle of his erection on a diagram of a clock face which contained the hour numbers from 12 to 6 o'clock.  Since clock hours are 30º apart, this made for relatively easy conversion to the Kinsey categories.

Figure 1. Degree Intervals
Used in the Kinsey Research
 

To be included in the photo sample, the self-reported (questionnaire) angle of erection and the verified (photo) angle of erection had to agree within 30º.  Of 99 subjects who volunteered to be photographed, 18 did not get fully erect and thus did not meet this questionnaire-photo angle agreement criterion.  This left an active sample of 81.  The 18 whose questionnaire and photo data did not agree, were about 9 years older (mean age = 49.4 years) than those with agreeing data.  They reported an average erection length (5.9") that was close to and slightly under the mean for the total sample.  However, being older, lower than average angles might have been expected—but the opposite was reported on the questionnaires.  In fact, the questionnaire angles of these eliminated subjects were reported to be 22.4º higher, on average, than the documented angles of the 81 subjects (mean age = 40.4 years) who met the criterion of agreement and 13.8º higher than the average for the Kinsey census sample (mean age = 39.8 years).  The combination of being older, reporting higher than average angles, and yet not obtaining a full erection for documentation, suggests some degree of internal conflict that was not compatible with the present research.  It was probably inevitable and appropriate that these candidates were screened out of the active subject pool.

Length was measured to the nearest ¼ inch at home by the Kinsey subjects and the results mailed to the researcher.  Measurements were on file for only 3,081 of the 9,865 subjects reviewed, posing the unanswerable question of whether those who measured and those who did not were systematically different on this variable.  Kinsey subjects also estimated their erection length during the interview, but this estimate was not used in the present research.  For length data, respondents in the photo sample entered a dimension in inches or centimeters on the questionnaire, using any fractions they wished and indicating whether this was an estimate or a careful measurement.  Centimeter values were converted later to inches.  For the photo subjects, the values used for analysis of angle and length consisted of the mean of the questionnaire and the photo values—except in the 26 instances when the photo value was higher (in degrees) or longer (in inches), and then the photo value was taken alone.  Mean values were used because we reasoned that self report would tend to over estimate and photography had the potential to under estimate lengths and angles.

A substudy on length was conducted in which the researcher made a direct measurement using a centimeter tape on 12 of the subjects in the photo sample.  In 4 cases the typical study measure (questionnaire + photo averaged) was longer, by an average of 0.2 in.  In 8 cases the centimeter tape measurement was longer, by an average of 0.1 in.  As a validity check on the present study's measurement methodology, this small exercise suggests that the questionnaire/photo method gives values that are within ±0.15 in. of a more traditional direct measurement procedure.  Pomeroy has been quoted as saying that were a researcher to be present to measure a man's erection, the man would probably lose at least part of it because of the researcher's presence.⁸  This may have been a factor affecting the present data collection.  However, in our study, the 18% of subjects not meeting the questionnaire/photo agreement criterion for inclusion, did not appear to lose part of their erection, but rather did not fully acquire it in the first place.  Each subject in this study determined when he had an optimal erection and if  he were ready for photography or measurement—which required only a matter of seconds.  This worked for the majority (81%), for whom any reduction in length or angle from nerves or from waiting was probably no more than the expected error of measurement.