Content uploaded by D Joye Swan
Author content
All content in this area was uploaded by D Joye Swan on Jul 31, 2018
Content may be subject to copyright.
1
CHAPTER 2: Models and Measures of Sexual Orientation by D. Joye Swan, PhD
2
Chapter Abstract: This chapter provides an in-depth examination of the varied models and
measures most commonly used to determine sexual orientation, in general, or bisexuality, in
particular. The chapter identifies the components of and the strengths and weaknesses of each.
The chapter presents each chronologically starting with the Kinsey Heterosexuality-
Homosexuality Rating Scale, followed by the Klein Sexual Orientation Grid, Storms Erotic
Response and Orientation Scale, the Sexual Identity Model, the Multidimensional Scale of
Sexuality, Savin-Williams Assessment of Sexual Orientation, and, finally, Sexual Configurations
Theory. Following the individual assessment of each model, the chapter concludes with general
critiques of all existing models and offers some recommendations for the establishment of future
models and measures with the goal of better capturing the complexity of bisexual orientation and
the goal of creating greater synergy in research assessment and more accuracy in estimating the
number of bisexual individuals.
KEYWORDS: Bisexuality, sexual orientation, Kinsey, Klein, measurement, assessment,
assessment
3
Introduction
The history of bisexuality would be incomplete without an in-depth presentation and
discussion tracing the models and measures that have attempted to measure sexual orientation
and bisexuality. At latest count there are currently over 200 measures that exist to define sexual
orientation (Fisher, Davis, Yarber, & Davis, 2013). While a review and discussion of all of them
are beyond the scope of this chapter, we outline below several of the ones which attempt to
subsume sexual orientation, including bisexuality, into overarching identity measures. We begin
with the Kinsey Homosexual-Heterosexual Rating Scale of the 1940s and conclude with the
most recent model, Sexual Configurations Theory (van Anders, 2015). Within the discussion of
each, we highlight its strengths and weaknesses and conclude the chapter with some general
critiques of the existing models presented and offer some suggestions on future directions for
measuring bisexuality.
The Kinsey Homosexual-Heterosexual Rating Scale
The Kinsey Homosexual-Heterosexual Rating Scale, most commonly referred to as the
Kinsey Scale, is the most widely recognized measure of sexual orientation. It was the product of
Zoologist, Alfred Kinsey’s, observations in the animal kingdom that nothing in nature is
black/white, either/or, but that nature operates on a continuum. Just as there are gradations of
intelligence, speed, coloring, etc., so too, he argued, we should find sexual orientation to be on a
continuum rather than a strict dichotomy between heterosexual and gay or lesbian. Kinsey and
his colleagues developed his scale for his landmark qualitative studies in the 40’s and 50’s on sex
and the human male and human female (Kinsey, Pomeroy, & Martin, 1948; Kinsey, Pomeroy,
Martin, & Gebhard, 1953). The scale is a seven point measure (0-6) that places an individual on
4
a point between Exclusively Heterosexual and Exclusively Homosexual (see Figure 1). Each
point on the continuum corresponds to a person’s determined (as opposed to self-identified)
sexual orientation (see Figure 2).
INSERT FIGURE 2.1 HERE
INSERT FIGURE 2.2 HERE
In Kinsey’s view, most people were bisexual to some degree, and he actually felt that
bisexuality was the most natural state of orientation (Kinsey et al., 1948). Therefore, he can be
credited with being the first to attempt to bring bisexuality to the public conscious. However, as
it relates to defining someone as bisexual, the Kinsey Scale is quite problematic. First, rather
than being a self-identity measure, it involved participants filling out questionnaires and then
having a researcher conduct interviews where after, the researcher would decide where an
individual was to be placed on the continuum. This method allows for great subjectivity as to
number and extent of behaviors that would cause different individuals to be placed on the same
point on the scale (Whalen, Geary, & Johnson, 1990). For example, where would someone be
placed who had kissed many people of the same-sex yet not engaged in any other physical acts
with them versus where one would place someone who had engaged in intercourse one time with
one person of the same-sex? Although the first behavior may be ‘incidental’ in intimacy, it
occurred not infrequently. Alternately, intercourse is an intimate behavior that is much more than
incidental, yet, in our example, it only happened one time. Even Masters and Johnson (1979)
expressed frustrations over how difficult it was to objectively assign individuals to anything
other than categories 0 and 6. If someone is a 2 or a 3, what does this mean exactly? Are all 2’s
alike? One of the issues from a research design perspective is that what should be parallel
corresponding numbers on the continuum are not operationalized the same. For example, (1)
5
Predominantly heterosexual, incidentally homosexual and (5) Predominantly homosexual,
incidentally heterosexual are not operationalized the same (See the definitions of category
descriptions 1 and 5 in Figure 2). Without a standardization of these categories, there is no
quantifiable way to classify respondents, and, therefore, no way to use the measure except to
make broad conclusions.
Although Kinsey asserted that sexuality was not categorical, he felt there was still value
in classifying people according to their behavior. Therefore, despite Kinsey’s assertions about
nature being on a continuum, the Kinsey Scale is, in fact, not a true continuum in that people can
only be orientated on one of seven, finite, possible points. However, given the difficulty with
placing people on the scale, it is perhaps a blessing; infinite points would make classifying
people and the utility of the scale infinitely more difficult (Sell, 1997).
Further, of particular concern in our attempt to define bisexuality, is that the Kinsey Scale
posits sexual orientation as a bipolar, unidimensional battle of behavioral frequency between
heterosexuality and homosexuality. Accordingly, it is assumed that more cross-sex behavior will
correspond to less same-sex behavior and vice versa. But is bisexuality about trading one sexual
attraction for another; would it be possible for bisexual people to be high on both same- and
cross- sexual behavior, or any combination thereof acknowledging that behaviors may change
over time as a product of opportunity and relationship status?
Epstein and Robertson (2014) recently proposed a way to remedy the limitation of using
a unidimensional model. They suggest measuring same-sex and cross-sex attractions separately
(using a 0-13 score) and then calculating the mean of the two resulting scores, after reverse
coding one of them. This gives a mean sexual orientation score (MSO) that will still have the
problem mentioned above so Epstein and Robertson then create a Sexual Orientation Range
6
(SOR), or confidence interval if you will, bracketing the MSO which identifies a range of scores
on the unidimensional continuum that could identify the individuals ‘true’ sexual orientation.
This partially remedies the problem of both asexual and bisexual individuals being located in the
middle. While the initial MSO will have them both as 6.5, an asexual person’s SOR would be 0
while a bisexual person’s range could be as large as 6.5- so the asexual doesn’t get any closer to
either the heterosexual or homosexual poles and a bisexual person’s score would branch out
toward both poles. However, this conception, while interesting, still defines bisexuality as “the
thing” in between heterosexuality and homosexuality (Weinrich, 2014).
Finally, although Kinsey says that during the interviews he took affective and emotional
attraction into account when he classified people, the Kinsey Scale only vaguely speaks of
“psychic response” which is taken to mean desires and fantasies rather than to imply feelings of
love and emotional attraction (Klein, 1993). For all intents and purposes, the Kinsey Scale is a
behavior-based measure of sexual orientation that, in its original format, ignored how an
individual self-identified and his or her affective motivations. However, affect is considered of
paramount importance to individuals’ self-identified definitions of their sexual orientation
(Baldwin, et al., 2016). In the end, Hanson and Evans (1985) saw the behavior focused Kinsey
Scale as appropriate if one is measuring the 0’s and 6’s of the world but as a barrier to
enlightening us on bisexuality.
Klein Sexual Orientation Grid (KSOG)
The second most widely known measure of sexual orientation is the Klein Sexual
Orientation Grid (KSOG, Klein, 1993). Although it provided several improvements over The
Kinsey Scale, it has less commonly been used in actual research (Sell, 1997). Fritz Klein was a
psychiatrist and founder of the American Institute of Bisexuality. In his seminal book, The
7
Bisexual Option (Klein, 1993), Klein critiques the Kinsey Scale as being too obtuse to capture
the complex nature of human sexuality. Instead of Kinsey’s focus on behavior, Klein identified
seven variables that he believed were vital to a measure of sexual orientation (see Figure 3).
INSERT FIGURE 2.3 HERE
These seven variables and Klein’s definition of them are:
1. Sexual attraction: To whom one is sexually attracted.
2. Sexual behavior: Whom one actually has sex with.
3. Sexual fantasies: Whom you fantasize about when you masturbate, daydream, etc.
4. Emotional preference: Whom you love or to whom you are emotionally attracted.
5. Social Preference: The sex of the people you hang out with socially.
6. Heterosexual-homosexual Lifestyle: The sexual identity of the people with whom you
hang out.
7. Self-Identification: How you think of yourself.
In addition to these seven variables, which were measured on two separate 7-point
assessment (see Figure 3), Klein, who believed that sexual orientation could change over time,
calling it an “ongoing, dynamic process” (p. 19), included a temporal component to his measure
asking individuals to rate themselves on the seven items in their past (> 1 year ago) and present
(last 12 months) lives. He also included a column asking participants to rate their ideal
distribution on each variable allowing that there may be psychic and social barriers to ones
feelings, thoughts and behaviors. The KSOG was a self-administered measure allowing
individuals to rate themselves in each area. Klein (1978) used the following example as a
conclusion about the sexual orientation of an individual named Kevin;
Using the numbers of the Klein Sexual Orientation Grid, Kevin’s profile is,
therefore, 6, 4, 7, 1, 4, 2, and 5. This example highlights the complexity of the
concept of sexual orientation. Given this complexity, which one number would we
assign to Kevin according to the Kinsey scale? How much more difficult still, then,
8
to fit him into one of the three categories of heterosexual, bisexual, or homosexual.
(p. 18)
Instead, Klein (1993) argued that by using the KSOG an individual’s numbers on each
cell of the grid would give someone an immediate visual idea of that person’s orientation.
Compared to the Kinsey Scale, it allowed for a much broader range of identities and accounted
for more of the nuances of sexuality. However, this strength is also one of the barriers to its use.
How does one scientifically study sexual orientation with a measure that can literally result in
thousands of orientation combinations? Its complexity of assigning seven different orientation
numbers to each individual makes it impractical for most research. In the original scale, Klein
delineated no way to combine the variables or to assign a weight to them. For example, would
behavior be a stronger indicator of sexual orientation than social preference?
In an attempt to address this criticism, Klein’s colleagues recently published a study
which sought to devise a method to place individuals into discrete categories of sexual
orientation based on their KSOG scores (Weinrich, Klein, McCutchen, & Grant, 2014). Male and
female participants completed the 21 KSOG items and the researchers conducted a cluster
analysis on the scores for each person on each variable. From this they were able to identify four
sexual orientation categories for women and five for men. Both sexes had a heterosexual and a
gay/lesbian category. Women had two bisexual categories, Bi-lesbian and Bi-heterosexual,
whereas men had three, Bi-homosexual, Bisexual, and Bi-heterosexual. By using standard
deviation analysis, they identified that sexual attraction, sexual behavior, and self-identity were
consistently important whereas, socialize (what are the sexual orientations of the people with
whom you socialize?) consistently was not, and past emotional preference was mixed for males.
For bisexual women, the clusters showed that, while their scores fell between those of the
9
heterosexual and lesbian clusters on all 21 items, the standard deviations (how much the scores
varied within each cluster) were very large on all of them suggesting that, although they were
clustered together as either bi-lesbian or bi-heterosexual, the individuals within each category
varied greatly from one another. For bisexual men, their scores on sexual attraction and sexual
fantasies toward men closely resembled the scores for bi-gay and gay men, but their actual
sexual behavior with men more closely matched heterosexual and bi-heterosexual men.
Additionally, for men as a whole, emotional preference did not really predict any cluster. It
might be worth noting, in regards to the binegative belief that bisexual men are “simply a gay
man in a different closet” (Swan & Habibi, in press, p. 7), that this study found bisexual men to
be statistically distinct from gay men when asked to rate each variable in relation to their ideal
self.
In sum, what the research showed was that the KSOG, like the Kinsey Scale, did a very
good job of categorizing heterosexual, gay, and lesbian individuals, while the results for bisexual
identifying individuals were less clear. While the cluster analysis identified two and three
bisexual groups for women and men, respectively, the variability was such as to say that those
within these clusters differed greatly from each other. In the end, Klein et al. (2014) concluded
that “bisexuality itself is a continuum” (p. 371). This conclusion seems to be becoming clearer
the more we research bisexualtiy and may well be the central message to those trying to define it.
Therefore, as Klein himself alludes to, the measure might be better used on an individual
basis for self-examination of one’s place on the various components of sexual orientation to
elicit, what he calls, the “a-ha reaction” (Klein, 1993, p. 20) to seeing how one’s sexual
orientation falls within the individual components of the KSOG. However, the KSOG is still
useful to research in that it recognizes behavior, affect, and self-identity as key components of
10
sexual orientation. Rather than using the entire grid, research could benefit from studying these
three components and distilling the weights of each in the organization of sexual orientation
measures. Further, the KSOG is a multi-dimensional and multi-temporal measure which better
accounts for the complexities of measuring sexual orientation. Further, in his use of a temporal
measure he was acknowledging what some have called situational bisexuality where someone
may behave bisexually for a given period of time (e.g., during college, prison inmates) or only in
specific situations (e.g. female swingers). However, rather than stemming, as Klein argued from
people’s changing sexual orientation, temporal changes might reflect changes in opportunity,
relationship status, or social and religious impediments to having one’s sexual behavior or
identity match one’s ideals. In other words, someone may behave and self-identify as
heterosexual despite having attraction, affect, and/or fantasies toward same-sex individuals due
to stigma-avoidance, social or religious judgments or internalized binegativity. If any or all of
these factors change, the individual’s behavior might change in a way that was more in line with
the other components of their sexual orientation. But, I would argue the individual’s sexual
orientation didn’t change, rather, what changed was their ability or willingness to express that
orientation.
Klein (1993) himself noted several other limitations of his model including- not
addressing the impact of age, the lack of weighting of love versus friendship or of lust versus
infatuation, and the quantifying of behavior (e.g. does frequency refer to number of sexual
behaviors themselves or number of partners with whom someone is sexually behaving?).
Although Klein made later attempts to demonstrate that the KSOG was reliable (Klein, Sepekoff,
& Wolf, 1985), his manuscripts simply stated that he had found the variables of the KSOG to be
“generally reliable” without any statistical data given to corroborate this assertion.
11
In addition, to Klein’s noted limitations, there are several other limitations of the model.
The most important of which is that, despite his own self-identity as bisexual, Klein failed to
include bisexuality or bisexual partners as part of the KSOG. In essence, it, like the Kinsey
Scale, sees bisexuality as a trade-off between the two monosexual orientations. Specifically, in
looking at the assessment for items F and G the seven points show a trade-off between
heterosexual and gay/lesbian. For example, item F asks the sexual identities of the people with
whom you socialize. Notice that one is able to hang out with heterosexual or gay/lesbian
identified individuals but not bisexual ones. Further, in item G, measuring one’s own self-
identity, you can only choose between degrees of identifying as heterosexual, gay, or lesbian.
Even, the mid-point, rather than being able to identify as bisexual, instead imposes a monosexual
label on the identity (hetero/gay-lesbian equally). It seems that as we unpack what bisexuality
means to people, it is not simply a byproduct of various amounts of cross-sex and same-sex
attractions but, rather, an independent orientation that overlaps in some ways with, but is distinct
from, monosexual orientations. Finally, subsequent researchers have found that study
participants have noted that some of the KSOG variables are confusing (Galupo, Mitchell, &
Grynkiewicz, 2014). In sum, while the KSOG improved upon the Kinsey Scale in important
ways, especially in adding affect and self-identity to its measure, and has great utility for an
individual’s self-examination of their sexual orientation, it is still not immune from the invalidity
of using monosexual anchors as the confines to defining bisexuality. Further, its construct
validity is just now being tested and early findings show that each of its variables may not have
the same degree of importance in determining sexual orientation. Finally, as Klein and his
colleagues found (2014), bisexuality itself, appears to be a continuum. If we average all bisexual
people into a single category, the different degrees and shadings of bisexuality are lost. (Note:
12
Weinrich [2014] has proposed that one could expand the KSOG to include parsing out the
bisexual types [hetero-bi, bi-bi, and homo-bi] Klein had elucidated. However, this would result
in empty categories on the grid and, ultimately, this was a musing on the part of the author and
has not been put to use.)
Storms Erotic Response and Orientation Scale
A third model of sexual orientation was proposed by Michael Storms (1980).
During the late 1970s as researchers were questioning the use of one-dimensional measures of
masculinity and femininity, Storms, too, questioned the Kinsey Scale’s single dimension to
measure sexual orientation. As we’ve already noted, one problem identified with the Kinsey
Scale is that it placed heterosexuality and homosexuality at opposite poles whereby if one was
more ‘heterosexual’ it meant that one was less ‘homosexual.’ Therefore, while the Kinsey Scale
did a good job of identifying people at the two poles, its predictive validity was greatly reduced
for those in the middle. Further, strongly dissimilar people could end up in the middle points on
the scale. For example, a person with high sexual desire for both men and women would score in
the middle of the Kinsey scale, however, so too, would a person with low sexual desire for both
men and women. In answer to this, Storms (1980) developed a two-dimensional model of sexual
orientation called the Erotic Response and Orientation Scale (EROS, see Figure 4).
INSERT FIGURE 2.4 HERE
Storms (1980) based his model on the premise that one’s erotic fantasies were the sole
factor necessary to measure sexual orientation; a belief partially held by Kinsey. The EROS is
composed of a seven-point scale (0 = never – 7 = almost daily) and 16 questions, eight
measuring an individual’s male targeted erotic fantasies (androerotic) and eight measuring an
13
individual’s female targeted erotic fantasies (gynoerotic). The fantasies ranged from what he
called “low intensity” (e.g., finding a man/woman sexually attractive), to “moderate intensity”
(e.g., daydreaming about having sex with someone), to “high intensity” (e.g., masturbating while
fantasizing about having sex with someone; p. 786). In a test of his model, Storms (1980)
directly compared the efficacy of EROS to Kinsey’s model in predicting the magnitude of the
sexual fantasies of bisexual individuals in comparison to heterosexual, gay, and lesbian
individuals. He asked individuals (70 college students and 115 participants recruited from a
college’s ‘gay’ club and gay social networks) to self-identify their sexual orientation as
heterosexual, gay (for both gay men and lesbian women), or bisexual, then had them complete
the EROS. Analyses, which were based on creating dichotomous categories from his 1-7 scale (0
= never, 1 = all other answers) revealed that self-identified bisexual participants (n=22) reported
heteroerotic fantasies at the same level as heterosexual individuals and as many homoerotic
fantasies as gay participants.
This model allows for the recognition that “bisexuality is the combination of homoerotic
and heteroerotic attractions, not a compromise between the two” (Weinrich, et al., 2014, p. 350).
When heterosexuality and homosexuality are seen as independent constructs, hetereroticism and
homoeroticism are allowed to vary independently from one another (Storms, 1980). Therefore,
an individual can have high erotic desires for both same-sex and cross-sex individuals rather than
forcing a tradeoff between the two. The idea that bisexuality is not just a tradeoff between a
cross-sex and a same-sex orientation, is the greatest strength of the EROS (Udis-Kessler, 2013;
Weinrich, 2014) and aligns with our current conceptualizations of sexual orientation. Recently,
EROS has gained some validation showing that the model has empirical promise (Weinrich,
2014).
14
However, there are several limitations and weaknesses with the EROS that have, perhaps,
kept it from being more widely acknowledged and used by researchers. Methodological issues
with the EROS include that it was only tested on a small, highly skewed sample greatly limiting
its generalizability. Additionally, reducing a 7-point scale to a simple dichotomy for analyses,
removes all the nuances that might have been revealed by leaving the scale intact. For example,
are there differences between the measures’ predictive ability for people who answered ‘rarely’
versus people who answered ‘almost daily’ to an item? Also, Storms (1980) summed all
responses, giving equal weight (importance) to all of the items regardless of their ‘intensity.’
Might masturbating while fantasizing about someone have a different impact on a measure of
sexual orientation than cuddling with someone?
Additionally, the assertion that fantasies alone construct sexual orientation has never been
shown to be valid. Given the correlational nature of the research, rather than fantasies driving
sexual orientation, it is just as plausible that sexual orientation drives fantasies. Further, research
has shown that there are at least three additional components that are important in measuring
sexual orientation- self-identity, behavior, and affect (Laumann, Gagnon, Michael, & Michaels,
1994; Mustanski, Kuper & Greene, 2014), however, Storms believed these to be inconsequential
noise that was a product of “social labeling” (p. 784). In terms of self-identity, which Storms did
measure in his study, it is important to note that, although Storms tested the alignment of self-
identified sexual orientation to erotic fantasies, he made no differentiation in, nor further
exploration of, individuals whose self-identity was in conflict with their fantasies (Udis-Kessler,
2013). Finally, Storms’ further argument that erotic fantasies were a byproduct of “learned
scripts” (p. 784) meant that Storms believed that sexual orientation was a choice based on one’s
experiences and exposures in youth rather than innate to an individual from birth.
15
Sexual Identity Model
Another model is Shively and De Cecco’s (1977) Sexual Identity Model. In most
research on sexual orientation, sexual self-identity is understood to be a single component of
sexual orientation. Instead, Shively and De Cecco see sexual orientation as a component of
sexual identity. In their model, sexual orientation is composed of two factors, physical attraction
and affectional attraction. Like Storms (1980), Shively and De Cecco propose that each factor is
composed of attraction toward same sex and cross-sex partners measured independently from
one another on a 5-point scale from not at all heterosexual to very heterosexual and not at all
homosexual to very homosexual, allowing for an individual to score high or low on each. In
addition to sexual orientation, they posit that sexual identity is also comprised of biological sex,
gender identity, and social role identity. Biological sex is how you are conscripted by the world
according to your biological make-up or genitalia (i.e. male, female, intersexed). Gender identity
is your personal sense of being a man, woman, or other identity (e.g., transgender, demiboy,
genderqueer, etc). Social sex role is the degree of masculinity and femininity, as expressed
through things like one’s appearance, behavior, personality, speech, etc. Again, like sexual
orientation, the items comprising one’s social sex roles are measured independently from one
another allowing for a two-dimensional construct where someone can be high or low on
masculinity and high or low on femininity. According to the authors, sexual identity is a process
whereby any of the elements can be congruent or incongruent from one another. As an example
of incongruency between two of the model’s factors, the authors use the case of a biological
male who has a highly feminine social sex role. They argue that such a person might become, in
their words, a transvestite (a man who dresses as a woman), as a way to reestablish congruency.
Although the argument that sexual identity is a product of sexual orientation and
16
not the other way around is intriguing and worth further empirical exploration, several
limitations have kept this model from being widely used in the 35 years since it was proposed.
First, like the Storms (1980) model, Shively and De Cecco’s model of sexual identity seems to
be highly impacted by the time period in which it was conceived. In the zeitgeist of the late
1970s there was a large uptick in research on masculinity and femininity which resulted in
several models still in use today (Bem, 1974; Spence & Helmrich, 1980) whose main
contributions were to see masculinity and femininity as separate constructs. Despite the
contribution of envisioning sexual orientation (or sexual identity) as two-dimensional, the
manifest belief that masculinity and femininity are predictive of each is an untested assumption.
When one looks at the variables that comprise social sex role, for example, appearance, behavior,
speech, etc., it may become immediately apparent that these are culturally subjective. For
example, a gay male may be associated with a feminine social sex role trait in one culture (e.g.,
United States) but not in another (e.g., New Guinea) (Ross, 1983). Further, not only are social
sex roles culturally dependent, they are also temporally tied. For example, what we would
identify as masculine or feminine or its association to something like sexual orientation changes
over time (e.g., pederasty in ancient Greece, men wearing wigs in the 18th century, women
wearing pants today). Additionally, Bullough (1976) has argued that the meanings ascribed to
social sex roles is a result of how positively or negatively a society views women. Likewise, the
idea of what makes two constructs congruent or incongruent is tied to culture and time.
The Multidimensional Scale of Sexuality
Although it is a model that has received very little notice in the empirical
literature, The Multidimensional Scale of Sexuality (MSS; Berkey, Perelman-Hall, & Kurdek,
1990) bears at least a brief discussion as it was a model that attempted to specifically identity
17
bisexual orientation diversity. Agreeing with Klein that unidimensional measures of sexual
orientation are too limiting to properly address sexual complexity, the MSS, in addition to
measuring same-sex, cross-sex, and asexual orientations, attempts to account for the variability
in bisexuality by introducing six categories of bisexuality: (1) homosexual orientation prior to
exclusive heterosexual orientation; (2) heterosexual orientation prior to exclusive homosexual
orientation; (3) predominant homosexual orientation (frequent homosexual desires and/or sexual
contacts) with infrequent heterosexual desires and/or sexual contacts; (4) predominant
heterosexual orientation (frequent heterosexual desires and/or sexual contacts) with infrequent
homosexual desires and/or sexual contacts; (5) equal orientation toward members of both sexes,
where desires for, and/or sexual contacts with members of both sexes occur on a fairly regular
basis (i.e., concurrent bisexuality); and (6) equal orientation toward members of both sexes,
where exclusive homosexual orientation (or vice versa), on an on-going basis (i.e., sequential
bisexual).
In addition to using the above as self-identity categories, the authors also created one
sexual behavior item and four affect/cognitive items within each category to place respondents
into orientation categories resulting in a 45 item scale, nine orientation identities with 1
behavioral and 4 affect/cognitive items for each one. The four affect items were arousal to sexual
fantasies/dreams, sexual attraction, emotional (love) factors, and arousal to erotic material.
Participants also reported their self-selected number on the Kinsey Scale.
Results revealed that affect was more strongly associated with the ‘correct’ orientation
category than behavior. That is, someone who’s cognitive/affect measure, for example, said they
were a concurrent bisexual, were indeed more likely to self-identify as a concurrent bisexual. In
fact, affect ratings loaded correctly on all of the self-description orientations, whereas, behavior
18
only correctly correlated with the corresponding orientation category on four of the seven
orientation categories (‘asexual’ and ‘heterosexual after predominant homosexual orientation’
had been eliminated from analyses as no participant identified with either of these categories).
Further, the MSS allowed for a more revealing picture of people who said they were Kinsey 3s.
Fifty-six percent of these individuals self-identified as concurrent bisexuals and just over 30% as
sequential bisexuals. However, note that the 2s and 5s on the MSS are not the same as Storms’
(1980) bi-heterosexual and bi-homosexual bisexual types which we previously discussed.
Despite its critique of measures that treat sexual orientation as a trade-off between
heterosexuality and homosexuality, the MSS, itself, is guilty of this very thing. Notice that
people can identify as currently hetero/homo sexual with some homo/hetero sexual attractions or
behaviors (which, uniquely, allows for an individual to be considered bisexual based on their
behavior/affect even if they do not embrace the label, is a major strength of the MSS). However,
these orientation categories are a forced choice between heterosexuality and homosexuality.
There is no option for an individual to identify as an “unequal” bisexual, what Weinberg and his
colleagues (Weinberg, Williams, & Pryor, 1994) would have called bi-heterosexuals (bisexual
individuals with greater attraction to cross-sex partners) and bi-homosexuals (bisexual
individuals with greater attraction to same-sex partners). Rather than bisexuality being the flash
point of orientation, it is a byproduct of degree of heterosexuality and homosexuality. A further
critique of the MSS is that the six bisexual categories were created as opinions of the authors.
Might there be other more important sub-categories of bisexuality (such as bi-homosexual or bi-
heterosexual) which would emerge from exploratory research.
Further, Berkey et al.’s (1990) finding that affect better predicted orientation category
than behavior might be an artifact of the methodology. Behavior was measured by a single item
19
(despite the authors commenting that a measure “should never include only one item,” p. 83)
while affect/cognition was calculated by collapsing the four ‘affect’ items into a single variable
(the mean of the four). First, if you look at the four items, we might dispute that they all measure
the same thing. Is arousal to same-sex erotic material equal to having been in love with someone
of the same sex? We also believe that arousal could just as easily have been included under the
behavioral category (if, for example, arousal led to masturbation). Therefore, the greater
predictive ability of the affect measure might simply have been that it included the mean of four
items which, themselves, might have been a mix of behavioral and affective items.
So, while the MSS is to be lauded for taking a “bisexuality first” approach to measuring
sexual orientation, much of the wording of the categories reinforced the unidimensional view of
bisexuality being a trade-off between heterosexuality and homosexuality. However, the MSS,
has some strengths that at least warrant its place in the discussion of building a model measure of
sexual orientation. It included measures of self-identity, behavior, and affect to categorize
individuals and, of all the other models, it is the only one that addresses the possibility of
identity, behavior and affect incongruence.
Savin-Williams Assessment of Sexual Orientation
The Savin-Williams Assessment of Sexual Orientation (Savin-Williams, 2010) is
a measure of sexual orientation rather than a model but warrants mentioning as it offers
measurement on seven or nine points, depending on the version used, to try to tweak out the
nuances of those who are not exclusively heterosexual or homosexual. The categories on the
measure are:
Exclusively Heterosexual/Straight, only sexually attracted to cross-sex partners
(Heterosexual/Straight, nearly always sexually attracted to cross-sex partners and rarely sexually attracted to same-sex
partners)
20
Mostly Heterosexual/Straight, mostly sexually attracted to cross-sex partners and occasionally sexually attracted to same-sex
partners
Bisexual Leaning Heterosexual/Straight, primarily sexually attracted to cross-sex partners and definitely sexually attracted
to same-sex partners
Bisexual, more or less equally sexually attracted to same-sex partners and cross-sex partners
Bisexual Leaning Homosexual/Lesbian, primarily sexually attracted to same-sex partners and definitely sexually attracted to
cross-sex partners
Mostly Homosexual/Lesbian, mostly sexually attracted to same-sex partners and occasionally sexually attracted to cross-sex
partners
(Homosexual/Lesbian, nearly always sexually attracted to same-sex partners and rarely sexually attracted to cross-sex
partners)
Exclusively Homosexual/Lesbian, only sexually attract to same-sex partners
(Note the items on the 7-point and 9-point versions are identical with the inclusion of the two
items in parentheses in the latter measure.)
This scale allows for the measurement of the bisexual types that Weinberg et al. (1994)
and Storms (1980) identified and sees sexuality on a “spectrum” (Savin-Williams, 2014).
However, it, once again, uses a unidimensional measure pitting bisexuality as the trade-off
between heterosexuality and homosexuality and measures sexual orientation using only
behavior.
Sexual Configurations Theory
Finally, one of the most recent attempts to create a model of sexual orientation is van
Anders’ densely packed, Sexual Configurations Theory (SCT; 2015). This theory has yet to be
tested or applied to any research of which we are aware. In essence, van Anders wanted to create
a theory that encompassed sexuality in all its diverse manifestations. Importantly, she criticized,
as she called it, the “unitary lust conceptualization of sexual orientation” (van Anders, 2015, p.
1178) of most models of sexual orientation. van Anders argues that sexual behavior focused
theories of sexual orientation are too narrow and limited to assess all the complexities he
envisioned sexual orientation involved. She was particularly interested in including
conceptualizations of love. The SCT is an attempt to capture once and for all, all the
21
complexities and nuances of sexual orientation. The theory assesses aspects of an individual’s
identity, orientation (interests, attractions, fantasies), and status (behaviors, activities). But,
unlike other measures of sexual orientation, SCT also includes the assessment of gender/sex, a
term van Anders uses to assess the target(s) of an individual’s orientation and defines it as
“whole people/identities and/or aspects of women, men, and people that relate to identity and/or
cannot really be sourced specifically to sex or gender” (p. 1181). van Anders envisions this term
to include social labels such as woman, man, transgender woman, transgender man, ciswoman,
cisman, genderqueer, and intersex. Essentially, she argues that people can be attracted to the sex
of an individual (biological), their gender (social construction of their sense of masculinity or
femininity), or combinations of the two. The theory posits that unpacking gender/sex could result
in questions like: “When I am intimately interested in being with women, am I interested in
people who identify as women? People who have vulvas and/or vaginas? People who are
recognized as female?” (p. 1181). To measure gender/sex the theory uses the terms eroticism
(bodily sexual pleasure, orgasm) and nurturance (warm loving feelings and closeness).
Additional constructs of SCT are partner number (the number of partners one has or would like
to have related to sexuality, eroticism, and nurturance) and other parameters like partner age
attraction (older, younger).
In each area of the SCT, an individual’s strength of a parameter can range from 0%-100%
and each variable can be branched and co-incident to account for how a specific individual
configures on the three main areas of identity, orientation, and status (e.g. a bisexually-identified
individual who is attracted to same-sex and cross-sex individuals but only engages sexually with
women, or desiring to flirt with women but only interested in penetrative sex with men) and
within each branch there can be exceptions to further delineate the three areas. As it tries to
22
incorporate all the possible configurations of sexual orientation, the theory sinks into
abstruseness. Any theory which tries to be ‘all things to all people’ winds up describing behavior
rather than predicting it. If every exception receives its own branch or every combination of
sex/gender attraction is a separate category, the theory loses its utility as a scientific tool. You
wind up with innumerable sexual orientations, many of which have so many branches of
‘difference’ that only a very small number of people fit into each one. For example, van Anders
asks why can’t a lesbian can be a woman who is also attracted to men? Other than violating the
definition itself, as an exception branch, it would require its own category and new definition. As
it stands now, the research utility of the SCT remains to be demonstrated. As it stands now, the
theory’s true utility is in illuminating, yet again, the diverse, unwieldy, complexity of sexuality.
Conclusion
General critiques of the models and measures.
The monosexual bias in model measures. A limitation of many of the described models
is the use of prescribed monosexualcentric labels as response options (e.g., response choices such
as “not at all heterosexual,” “mostly homosexual,” etc.). Using the terms homosexual and
heterosexual as measure response categories makes an assumption that one’s sexual orientation,
or any like factor, can be identified by fitting one of these two categories or, by default, being
identified as something ‘other’ if it does not. This means that bisexuality is and can only be an
off-shoot of the two (i.e., one is bisexual as a result of not fitting either monosexual category)
rather than its own sexual orientation label (which, by the way, would allow for the inclusion of
attractions to bisexual individuals as a response choice on measures). Further, this
conceptualization of sexual orientation precludes the reverse, that heterosexuality and
homosexuality may emerge as the extremities of a bisexual orientation, whereby bisexuality
23
becomes the focal point in the measurement of sexuality variables rather than the byproduct.
This conceptualization starts with bisexuality as the ‘inclusive’ orientation and as measure
responses exclude sexual or affectional options, reduces one to a monosexual orientation and
from there, one would be labeled heterosexual, gay, or lesbian (see Figure 5).
INSERT FIGURE 2.5 HERE
Data driven versus researcher driven perspective on identifying sexual orientation
categories. Finally, a critique of most of the models and measures we reviewed is that the
researcher or researchers in each case imposed their own categories of sexual orientation on the
participants, trying to fit individuals into their definitions of sexual orientation, rather than the
other way around. In other words, in most cases sexual orientation categories were derived a
priori, before participants answered a single question. Further, researchers assumed what
variables made up an individual’s sexual orientation in terms of behavior, affect, and self-
identity. So, while we can say which of the variables included in the models are most predictive
of sexual orientation, we have no way of knowing if important variables have been overlooked.
We would encourage future research to gather open-ended responses from a broad swath of the
populous on what factors they identify as contributing to sexual orientation and the weight they
give each of these factors. From these discussion we could create data driven models of sexual
orientation.
All of the models and measures discussed have various flaws and limitations. However,
as our understanding of sexual orientation expands, we should be able to better synthesize and
refine the models to a point where some revised version of one of the current models or a new
model rising from an old one, will be able to capture a significant amount of the variability in
what it is to define someone as bisexual. One of the problems with almost all existing models is
24
that they see bisexuality as either a tradeoff or a spinoff of either same-sex or cross-sex
orientations. I suggest that if we flipped the premise, putting bisexuality as the starting point of
sexual orientation, we might open ourselves to new conceptual “a-has” in our quest to measure
and define, not just bisexuality, but sexual orientation in general.
25
References
Baldwin, A., Schick, V. A., Dodge, B., van Der Pol, B., Herbenick, D., Sanders, S. A. &
Fortenberry, J. D. (2016). Variation in sexual identification among behaviorally bisexual
women in the Midwestern united states: Challenging the established methods for
collecting data on sexual identity and orientation. Archives of Sexual Behavior, 46(5),
1337-1348. doi:10.1007/s10508-016-0817-0
Bem, S. L. (1974). The measurement of psychological androgyny. Journal of Consulting and
Clinical Psychology, 42, 155–62.
Berkey, B. R., Perelman-Hall, T., & Kurdek, L. A. (1990). The multidimensional scale of
sexuality. Journal of Homosexuality, 19(4), 67-87. doi:10.1300/J082v19n04_05
Bullough, V. L. (1976). Sexual variance in society and history. Los Angeles, CA: Wiley & Sons.
Epstein, R. & Robertson, R. E. (2014). How to measure sexual orientation range and why
it's worth measuring. Journal of Bisexuality, 14(3-4), 391-403.
doi:10.1080/15299716.2014.933378
Fisher, Davis, Yarber, & Davis, S. (2013). Handbook of sexuality-related measures. NY: Sage
Publications
Galupo, M. P., Mitchell, R. C., Grynkiewicz, A. L., & Davis, K. S. (2014). Sexual minority
reflections on the Kinsey Scale and the Klein Sexual Orientation Grid: Conceptualization
and measurement. Journal of Bisexuality, 14, 404–432.
doi:10.1080/15299716.2014.929553
Hanson, C. E., & Evans, A. (1985). Bisexuality reconsidered: an idea in pursuit of a definition.
26
Journal of Homosexuality, 11(1-2), 1-6.
Kinsey, A. C., Pomeroy, W. B., & Martin, C. E. (1948). Sexual behavior in the human male.
Philadelphia, PA: Saunders.
Kinsey, A. C., Pomeroy, W. B., Martin, C. E., & Gebhard, P. H. (1953). Sexual behavior in the
human female. Philadelphia, PA: Saunders.
Klein, F. (1993). The bisexual option (2nd ed.). New York, NY: Haworth Press.
Klein, F., Sepekoff, B., & Wolf, T. (1985). Sexual orientation: A multi-variable dynamic
process. In F. Klein & T. Wolf (Eds.), Two lives to lead: Bisexuality on men and women
(pp. 35-49). NY: Harrington Park Press.
Laumann, E. O., Gagnon, J. H., Michael, R. T, & Michaels, S. (1994). The social organization of
sexuality: Sexual practices in the United States. Chicago, IL: University of Chicago
Press.
Masters, W. H. & Johnson, V. E. (1979). Homosexuality in perspective. Boston, MA: Little,
Brown.
Mustanski, B., Kuper, L, & Greene, G. J. (2014). Development of sexual orientation and identity.
In D. L. Tolman & L. M. Diamond (Eds), APA handbook of sexuality and psychology:
Volume 1. Person-based approaches. American Psychological Association.
doi:10.1037/14193-019
Ross, M. W. (1983). Homosexuality and social sex roles: A re-evaluation. Journal of
Homosexuality, 9(1), 1-6.
27
Savin-Williams, R. C. (2010). Sexual Orientation Label (7-point),
http://www.human.cornell.edu/hd/sexgender/research.cfm
Savin-Williams, R. C. (2014). An exploratory study of the categorical versus spectrum
nature of sexual orientation. Journal of Sex Research, 51(4), 446-453.
doi:10.1080/00224499.2013.871691
Sell, R. L. (1997). Defining and measuring sexual orientation: A review. Archives of Sexual
Behavior, 26(6), 643-658.
Shively, M. G. & de Cecco, J. P. (1977). Components of sexual identity. Journal of
Homosexuality, 3, 41-48.
Spence, J. T., & Helmreich, R. L. (1980). Masculine instrumentality and feminine
expressiveness: Their relationships with sex role attitudes and behaviors. Psychology of
Women Quarterly, 5(2), 147-163.
Storms, M. D. (1980) Theories of sexual orientation. Journal of Personality and Social
Psychology, 38(5), 783-792.
Swan, D. J., & Habibi, S. (in press). When is a bisexual really bisexual? Testing the “one and
done” rule of male same-sex behaviour. Psychology of Sexualities Review.
Udis-Kessler, A. (2013) Genealogy of the concept of bisexuality. In M. Storr (Ed.), The critical
bisexuality reader. NY: Routledge.
van Anders, S. M. (2015). Beyond sexual orientation: Integrating gender/sex and diverse
28
sexualities via sexual configurations theory. Archives of Sexual Behavior, 44, 1177-1213.
doi:10.10.1007/s10508-015-0490-8
Weinberg, M. S., Williams, C. J., & Pryor, D. W. (1994). Dual attraction: Understanding
bisexuality. NY: Oxford University Press.
Weinrich, J. D. (2014) Multidimensional measurement of sexual orientation: Present. Journal of
Bisexuality, 14(3-4), 373-390, doi:10.1080/15299716.2014.951138
Weinrich, J.D., Klein, F., McCutchan, J. A., & Grant, I. (2014). Cluster analysis of the Klein
Sexual Orientation Grid in clinical and nonclinical samples: When bisexuality is not
bisexuality? Journal of Bisexuality, 14(3-4), 349–372.
doi:10.1080/15299716.2014.938398
Whalen, R. E., Geary, D. C., & Johnson, F. (1990). Models of sexuality. In D. P. McWhirter, &
J. M. Reinisch (Eds), Homosexuality/heterosexuality concepts of sexual orientation (pp.
61-70). NY: Oxford University Press.
