On the Science of Changing Sex


Posted in Confirming Two Type Taxonomy by Kay Brown on November 4, 2010

critical-thinkingIn one of the most misunderstood of Blanchard’s papers, he and his colleagues, Clemmensen and Steiner, explored the likelihood and areas in which MTF transsexuals might selectively ‘color’ their presentation of their sexual and gendered behavior history.  Many people on both sides of the debate have assumed that Blanchard was making a judgment upon “non-homosexual” transsexuals, saying that they were more prone to lying than “homosexual” transsexuals.  Actually, the data says just the opposite, that neither group is more naturally inclined to such distortion, generally.  However, the data does say that the more an individual is inclined to color their responses to questionnaires in such a way as to present as more socially desirable, the more likely that they will answer questions regarding sexual history and gendered behavior is such a way as to increase the likelihood of being accepted for SRS.  Is that so shocking?  That ‘trannies’ might ‘enhance’ their chances of getting past the gate keepers? (Gasp!)

But there is more detail to the study, that is important to note;  While both AGP and androphilic transsexuals were just as likely to color and shade their history, the AGP  transsexuals did so along all eight of the study’s scales, while HSTS did it only on one of the scales.

But, before we get to that, I think we need to explore how we know this, how Blanchard, et al. determined this.  They sought correlations between their sexological scales and the Crowne-Marlow Social Desirability Scale.

The Crowne-Marlow scale is a set of 32 statements that are answered true or false for one’s self.  Each statement is scored with either a zero or a plus one, depending if the answer indicates a tendency to color one’s socially desired behaviors.  Thus, the scale goes from zero to thirty-two (0-32).  The statements are very clever in that each statement, if answered in the non-socially desirable fashion, would still not be indicative of any pathology, and in fact might indicate self-honesty.  For example, one of the statements reads, “I have never intensely disliked anyone.”  If one answers “true” this is a socially desirable answer, most saintly indeed.  However, how many of us can honestly answer that there has never been someone, some time, that pissed us off so badly, that we still hold an intense and personal hatred for them?  (I can think of several such individuals instantly.)  The statements are also chosen to be “graded” from not likely to be that good, to likely to be that good, in that some statements might be honestly answered in the socially desirable manner by many, if not most people, for example, “I would never think of letting someone else be punished for my wrongdoings.”

Thus, the Crowne-Marlow scale has the unusual property that an honest saint may give the same high score as a dishonest sociopath.  So, a high score in no way indicates that one is a liar per se.  In fact, the scale is nearly useless as an individual test.  It is only in groups, large groups can we use the scale to look for meaningful inferences, in either the mean scores or in the correlations with other scales.

In his chapter comparing various scales of social desirability Paulhus noted that:

Crowne and Marlowe (1964) reported a mean of 15.5 (s.d. = 4.4) in a sample of 300
college students. In a more recent study of 100 students, Paulhus (1984) reported means of 13.3 (s.d. = 4.3) and 15.5 (s.d. = 4.6) in anonymous and public disclosure conditions, respectively. In a sample of 503 students, Tanaka-Matsumi and Kameoka (1986) reported means of l4.0 and 12.3 for normal and depressed respondents, respectively. In a sample of 650 Peace Corps volunteers (90% college graduates), Fisher (1967) found means of 16.1 (s.d. = 6.8) and 16.4 (s.d. = 6.5) for males and females, respectively.

Thus, we see that Peace Corps volunteers, probably the closest sample that we will ever find to saintly people, give scores in the range of 16.1-16.4.  But, Paulhus found that college students gave mean scores of 13.3 in an anonymous situation and 15.5 when they knew that someone they knew would be reading their answers.  So, folks tend to ‘color’ their answers when they feel that they might be judged in some manner by their answers?  Not much surprise there!

Compare these scores to the mean scores of the two types of transsexuals in Blanchard’s study of 17.68 for the “heterosexual” (non-homosexual) and 20.02 for the “homosexual” transsexuals.  Given that the scores for female Peace Corps volunteers was only 16.4, do we really believe that these transsexuals were more saintly?  Are we surprised that these transsexuals would be assuming that their answers to the other eight scales would be used to judge them, possibly used to deny them essential medical services, that they might wish to color their responses?  Also, please note, as did Blanchard, that if anything, the HSTS group was more likely to color their answers than the non-hsts group.

It is in the correlations with the individual scores on the Crowne-Marlow scale and the scores on the other eight scales used in the study that we learn something really interesting about each group (taken verbatim from the paper):

Correlations of Demographic Variables and Questionnaire Measures with
Social Desirability Scale ~
–                                                                                      All                Hetero        Homo
Variable                                                                  r           p            r       p           r      p
Age                                                                        -.04       NS     -.13     NS      .23  NS
Education                                                             .01       NS     -.04    NS      .18   NS
Item: Felt like a woman                                  .30     .001    .29   .011     .26  .034
Item: Rather live as female                           .27     .002    .34   .003   .01  NS
Feminine Gender Identity Scale                 .35     .001     .37  .001    .16   NS
Modified Androphilia Scale                          .28     .001     .25  .022    .02  NS
Modified Gynephilia Scale                           -.30    .001   -.38   .001    .18   NS
Cross-Gender Fetishism Scale                    -.35    .001   -.48   .001    .08   NS
Item: Aroused by cross-dressing              -.29    .001   -.34   .003    .02   NS
Item: Masturbated cross-dressed             -.27   .002   -.34   .003    .06   NS

~The abbreviations Hetero and Homo refer to heterosexual and homosexual subjects.
Columns headed r are correlation coefficients; columns headed p are their associated one-tailed probabilities. The abbreviation NS means that the associated correlation coefficient was not statistically significant at the 0.05 level. (The smaller the number, the more “statistically significant; that is to say, that it is more likely to be “real” and not just chance.)

Looking at the two groups and correlations, one notices that on all eight of the sexualogical scales, for the “heterosexual” group the correlations are all statistically significant.  Further, the single highest correlation was on the Cross-Gender Fetishism Scale (a measure of autogynephilia) at -0.48.  For those familiar with psychological research and statistics, this number screams!  (No correlation would be 0.00 and perfect, one to one, correlation would be 1.00; so this number is half way between.)  That is a very high correlation telling us that this group, as a group, would like to color this scale.  That is, that the more likely that an individual is to have a high score on the Crowne-Marlow scale, the more likely they will have a low score on this autogynephilia scale!  Ok, this can be interpreted that individuals who wish others to see them as having socially desirable traits are more likely to minimize or deny experiencing autogynephilia.

Similarly, scores for gynephila and androphilia are colored to minimize their attraction to women, while maximizing their attraction to men, and so on down the line, to seem more “classically” transsexual (more like HSTS) perhaps?

In contrast, for the “homosexual” transsexual group, there was only one scale that has a statistically significant correlation, “Felt like a woman”, and only just barely “significant”.  This was a scale from one to three that indicates under what state of dress that they felt like a woman, with three being dressed as either a man or a woman, to never, which excluded the subject as not being “transsexual”:

Item: Have you ever felt like a woman?
a. Only if you were wearing at least one piece of female underwear or clothing (1)
b. While wearing at least one piece of female underwear or clothing and only occasionally at other times as well (2)
c. At all times and for at least one year (3)
d. Never felt like a woman (exclude subject)

But… BUT… do the math… there were only fifty-one “homosexual” subjects (N=51) which gave a mean score of 2.96 on this scale.  That would come from two subjects giving a score of 2, while the rest, all forty-nine of the others, scored 3.  Also note that that one other correlation almost reached the threshold for statistical significance: age, at 0.23.  That is to say, that a weak correlation was found with older subjects being more likely to have a higher score on the Crowne-Marlow scale.  This suggests to me that age will have a weak correlation with higher scores on the “felt like a woman scale”… thus… we might guess that those two subjects that answered “2” instead of “3” were younger than the average of the “homosexual” group… perhaps they were more tentative in their answers?  Overall, this isn’t much of a strong signal.  In spite of the higher mean score on the Crowne-Marlow scale, there doesn’t seem to be much evidence for coloring their answers.  Perhaps that’s because they didn’t feel that they needed to?

Addendum 1/15/2018:

Study after study has shown that around 80% to 85% of “non-homosexual” transwomen readily acknowledge experiencing sexual arousal to cross-dressing, at least in adolescence.  This leaves 15% or so who say that they didn’t.  As we’ve seen, social desirability bias is strongly operating.  At least some percentage of the population is not being honest.  Interestingly, there is a new study exploring the subject of honesty and dishonesty.  It was discovered that there are three types of people that I shall paraphrase as “always honest”, “mostly honest”, and the “never honest”.

Guess what the percentages were?

“always honest” = 50%
“mostly honest” = 35%
“never honest” = 15%.

Wow, what an interesting coincidence.

Many transwomen who are critical of the two type taxonomy have specifically called out Blanchard and any who accept and advocate the taxonomy as being ugly transphobes for having called transwomen liars.  But are we to suppose that transwomen, as a population, are somehow more saintly and honest than the general population?  Phhfft !


Addendum 2/16/2019:

I found another paper dealing with dishonesty in research studies that showed that around 10% to 20% of people were willing to lie in a survey just to make a few bucks, even as little as $5.  Wow, that’s an interesting coincidence; around an average of 15% of people are willing to lie for so small an incentive.  Can we really believe that transwomen, as a population, are somehow more saintly and honest than the general population?



Ray Blanchard, Leonard H. Clemmensen, Betty W. Steiner, “Social Desirability Response Set and Systematic Distortion in the Self-Report of Adult Male Gender Patients

Douglas P. Crowne, David Marlowe, “A New Scale of Social Desirability Independent of Psychopathology”

Delroy L. Paulhus, Chapter 2: “Measurement and Control of Response Bias”
J. P. Robinson, P. R. Shaver & L. Wrightsman (Eds), Measures of personality and social psychological attitudes (pp. 17-59)., Academic Press, Inc.

Comments Off on Misunderstood

%d bloggers like this: