Differences and similarities in men's and women's sexual self-schemas.

In the last 10 years, researchers have developed separate rating scales for men and women that purport to measure a person's thoughts about their sexual self or sexual self-schema. While these scales are promising, there are some problems intrinsic to these measures. This study addresses these concerns and tests a combined measure for use among both men and women, while retaining internal consistency and a factor structure similar to the women's scale. Having both men and women complete the same scale permitted a comparison of how men and women think about their sexual selves. In this study, gender differences netted a small effect, with men scoring lower on both the loving / warmth and reserved/conservative factors compared to women. Results are discussed in terms of future research with this new scale and the meaning of gender differences along these dimensions.

Does gender influence how we think about our sexual selves? From Freud's early connections between anatomy and sexuality to Daryl Bern's (2000) "exotic becomes erotic" theory, gender has been consistently implicated in theories of sexuality. Gender role scripts no doubt influence sexual behavior, yet one's own sense of gendered self-concept only weakly predicts sexual behavior (e.g., Garcia, 1999; Garcia & Carrigan, 1998; Whitley, 1988). It seems gender roles predict sexual behavior, but individuals may not always internalize social roles into their self-concepts. To what extent do men and women internalize social roles into their sexual sense of self?

One answer to this question comes from work in the last decade on sexual self-schemas. Building on Markus's (1977) early work that established the self-schema — a cognitive representation of who we think we are — Andersen and her colleagues defined sexual self-schema as "cognitive generalizations about sexual aspects of oneself (p. 1079). Consistent with the typical approach to schemas, the sexual self-schema shapes, and is shaped by, experience. Thus, a person who has little sexual experience may come to see their self as asexual, and this self view will influence their future sexual choices.

To assess sexual self-schemas, Andersen and Cyranowski (1994) designed a scale to covertly measure a woman's sense of her sexual self. They had university students and community members rate trait adjectives (e.g., passionate) on how well the words could be used to describe a "sexual woman." After identifying the 26 best adjectives using standard scale construction methods, they asked university students to rate themselves on a 7 point scale (0 to 6, from "not at all descriptive of me" to "very much descriptive of me") for each adjective. A factor analysis of the ratings revealed three potential dimensions to women's sexual self: passionate/ romantic, open/direct, and embarrassed/conservative.

Andersen and Cyranowski (1994) combined the passion/romantic and open/direct factors to create a "positive" score; scores on the embarrassed/conservative factor represented a "negative" score. Overall, women who scored high on the positive dimension, as measured by other established instruments, tended to be emotionally romantic and passionate, liberal in sexual attitudes, and free of inhibitions. They also reported higher levels of arousability and willingness to engage in casual sex, and they tended to be motivated towards sexually intimate relationships. In contrast, women who scored high on the negative scale tended to be emotionally reserved and inhibited in their sexuality and generally conservative about sexual matters. Andersen and Cyranowski proposed that there were four categories of women: those who had positive, negative, co-schematic (i.e., having both negative and positive dimensions), and aschematic self-concepts (i.e., having neither positive nor negative dimensions) (see Figure 1).

In a second paper Andersen, Cyranowski, and Espindle (1999) sought to circumscribe men's sexual self-schema. As with the women's scale, they used standard scale construction methods to identify 27 words that represented "a sexual man." In a factor analysis, university student participants' self-ratings on these terms suggested three factors in men's sexual selves: passionate/loving, powerful/aggressive, and open-minded/liberal. As with the women, Andersen and colleagues found that men who scored high on the three factors (i.e., schematics) compared to those who scored low (i.e., aschematics), considered themselves more sexual, more sexually arousable, and more willing to engage in casual sex. Schematic men, compared to aschematic men, reported a wider range of sexual activities, more lifetime sexual partners, and more brief sexual encounters. Schematic men were more likely to be involved in a relationship, and they estimated more sexual partners in the future than aschematic men.

Early studies with these scales demonstrated both concurrent and predictive validity for the adjective lists. For example, women with vaginismus expressed less positive sexual self-schemas than women with no pain (Reissing, Binik, Khalife, Cohen, & Amsel, 2003). Studies on predictive validity claimed that sexual self-schema scores predicted sexual changes after breast cancer surgery (Yurek, Farrar, & Andersen, 2000) and predicted sexual behavior and responsiveness among women with gynecologic cancer (Andersen, Woods, & Copeland, 1997; Andersen, Woods, & Cyranowski, 1994). However, there have been mixed results relating sexual self-schema scores to women's satisfaction with their body, even though sexual self-schema scores do predict physical attractiveness in women (Wiederman & Hurst, 1997). These findings have been largely confirmed for college-aged women, but correspondence between self-schema and sexual behavior may not be as strong for women over 30 years of age, where self-schemas appeared to be less correlated with actual sexual behavior (Volsky Rushton, 2003).

Research conducted with the men's self-schema scale is much more preliminary and overall less solid. Aarestad (2000) reported that men's sexual self-schema could not predict sexual behavior, but was related to the man's experience of romantic relationships. Barner (2003) raised questions about the men's scale since positive sexual self-schemas in men were unrelated to self-reported history of sexual aggression and coercion, despite the fact that powerful/aggressive was a key dimension of a positive self-concept in the men's scale.

It may be that a range of problems were confounding results, especially with the men's scale. Both scales were initially developed as unobtrusive measures of sexual selves, so the instructions for completing the scales were very general. The hope was that raters would be more honest if they didn't know they were rating their sexual self. Participants were simply asked to "describe yourself and to "consider whether or not the term describes you." The intent was to have a covert measure of sexuality, and Andersen and colleagues were successful in this regard. But there are dangers when using such vague instructions to assess a dimension of self-concept. What dimensions of self-concept are participants considering when rating themselves on this task? It seems reasonable to assume that one would rate oneself differently if thinking about oneself as a student, employee, or lover. This is especially troubling given that self theorists have suggested that we probably do not have one self, but rather many possible selves that emerge at different times in different contexts (e.g., Markus & Nurius, 1986). If the scale is measuring some generalized concept of self, how valid is that when specifically attempting to measure how one thinks of themselves as a sexual person? In line with this reasoning, participants need to know which dimensions of self they are rating. Even though a more explicit rating of one's sexual self risks a self-presentation bias, ratings must assess how people think of themselves in a primarily sexual manner and not in a general sense.

There are further problems. The traits in the scales are strongly related to gender stereotypes. Thus, it could be that raters are considering their sense of themselves as a gendered person, which may or may not correspond with their sexual behavior. For instance, a man may see himself as aggressive in general, or even in specific contexts like work settings, but submissive in his sexual relationships. Indeed, many of the trait adjectives used in the women's and men's scales (e.g., aggressive, compassionate, domineering, independent, individualistic, sensitive, sympathetic, timid, warm) are identical or synonymous with adjectives used in common gender role questionnaires such as the Bern Sex Role Inventory (S. Bern, 1974) or the Personality Attributes Questionnaire (Spence, Helmreich, & Stapp, 1975). It may be that the Andersen scales conflate sexual and gendered selves making their scale an assessment of how much men and women rate their sexual self consistent with gender stereotypes.

Furthermore, self-ratings of this sort need to be anchored. When rating oneself as "sympathetic," without guidance, how should one arrive at a quantified judgment on a seven point scale? Sympathetic compared to whom? These ratings cannot be made in isolation without context, since everyone could anchor their ratings differently. As noted by others who conducted research on personality self-rating tasks (e.g., McCrae, Stone, Fagan, & Costa, 1998), it is imperative that researchers provide some anchors for comparison; otherwise, the participants are left to imagine the basis of comparison. At the very least, in tasks like this, participants should compare themselves to others of the same gender and age, so there is some stated and consistent basis for which they can make ratings.

Comparing the actual content of Andersen's men's and women's sexual self-schema scales suggests further concerns. Table 1 lists the items from each scale in alphabetical order. Thirteen of the adjectives in the two scales are identical; 14 are very similar if you count "warm" and "warm-hearted" as synonymous. Thus, over half of the items of both scales overlap, suggesting that sexual self-concepts for men and women are somewhat related, obviating the need for two different measures of sexual self. Andersen and her colleagues appeared to have begun with the premise that men and women understand their sexual selves differently; yet it seems their data suggest this is not entirely the case. Andersen and colleagues (1999) pointed out these similarities, noting that in both scales the passionate/romantic dimension is primary, and they shared about half of the same items. Additionally, the men's open-minded factor corresponded to women's open/direct factor. Andersen and her colleagues noted, however, that these scales seemed to function differently. For men, open-mindedness appeared related to behavioral activation and emotional involvement, whereas for women "open/direct" related to behavioral activation only. That may be, but such similarities in content challenge the rationale underlying separate scales to assess sexual self-schema for each gender. Indeed, Andersen and colleagues (1999) suggested that researchers might have both men and women fill out both the women's and men's versions of the scale, which is basically the inspiration for this article.

Having men and women complete a composite scale with all items would address two other problems. Andersen and Cyranowski (1994) found that women saw their sexual self more negatively than men, and they included more negatively valued items (e.g., embarrassed) in the women's scale. Indeed, there were only three negative adjectives for the men (i.e., conservative, reserved, and inexperienced). Thus, the factor analysis on the men's scale confirmed there was no negative factor in their sexuality ratings. Is this really the case, or is this a consequence of reluctance among college-aged men to characterize their sexuality in negative terms? Perhaps they have no negative conception of their sexual self. It could also be that men understand their negative sexual self in similar terms to women, so completing a common scale with negative items would answer this question. Additionally, administering all items to both sexes would permit a comparison of men's and women's responses to identical items. This would help test the premise that men and women think about their sexual selves differently.

This study tested a composite version of the sexual self-schema scales (items from both scales). The scale was modified to instruct participants to explicitly consider their "sexual self when completing the scale. Moreover, the instructions were modified to include comparison anchors for gender and age. It was expected that men and women would rate themselves differently overall on the scale. It was also hypothesized that a factor analysis would confirm a three factor solution for both men and women, similar to what Andersen and her colleagues found for women only, suggesting that in fact, men do consider negative dimensions of their sexuality.

This study drew participants from two populations: a Canadian university and an undergraduate college in the United States. In Canada, undergraduate students enrolled in social science courses at a large English-speaking university in Montréal were given the opportunity to complete the sexual self-schema questionnaire. In all, 439 students (251 women and 188 men) completed questionnaires. Participation was voluntary; however, a modest incentive was offered: participants were entered into a draw for two prizes of $100 each. In this sample, the race or ethnicity of participants was unknown, but the school is widely recognized as providing instruction to a largely cosmopolitan and ethnically diverse student body. Most students identified as heterosexual (94.1%), with 3.4% identifying as either gay or lesbian, and 2.5% claiming a bisexual identity.

In the U.S. sample, 394 undergraduate students (224 women and 170 men) enrolled in a wide range of disciplines in a college in New York City completed the questionnaire. As in the Canadian sample, participants volunteered for the study, but were placed in a draw for one prize of $100. Most participants were White (62%), but many were Hispanic (12%), African-American (9.6%), Asian (7.3%), or other/multi-racial (9.1%). The majority of students identified as heterosexual (97%), 1.5% identified as either lesbian or gay, and 2% identified as bisexual.

A modified version of the sexual self-schema scale was used. To emphasize the measurement of sexual self-concept, the questionnaire was given the title "Sexual Self-Schema Scale." Items from both the men's and women's scales were combined, and the neutral items included in the original scales were omitted. In addition, the instructions were changed to make explicit the measurement of one's sexual self-schema and to encourage comparisons to others of similar sex and age. The instructions for the scale were as follows:

Modification of the scale did not seem to affect its internal consistency: coefficient alpha for the combined scale was .82, which was consistent with previous studies on the two independent scales (Andersen & Cyranowski, 1994; Andersen et al., 1999).

In both samples, research assistants approached instructors of courses for permission to enter their classes and administer the questionnaire. Once permission was obtained, researchers read instructions for the questionnaire and distributed a consent form that included the instruction that participants would be asked questions regarding "how I feel about my self as a sexual person" so as to cue participants to rate their sexual self. Participants were then instructed to return the consent forms separate from the questionnaires; thus, all participants responded in a completely anonymous fashion. All participants were offered feedback on the results, which was sent to those who requested it upon completion of the study.

The results of this study focus on a factor analysis of the combined scale among the men and women participants and a test of gender differences in responses to the scales. Since the samples were drawn from U.S. and Canadian students, a comparison of American and Canadian students was also conducted.

Andersen and Cyranowski (1994) found three factors in the ratings of women's sexual self-schema, yet Andersen and colleagues (1999) found two factors in the ratings of men's sexual self-schema. Did combining the items from the two scales alter the factor structure of the trait set? Since previous research reported three factors in the women's responses, and it was believed that men's responses would also cluster similarly, a three-factor solution was forced in a principal components analysis (using oblimin rotation with Kaiser normalization). The factor analysis converged on factors equivalent to Andersen's romantic, embarrassed, and power factors for both men and women (see Table 2). In this solution, the three factors could be more accurately labeled: loving/warm, reserved/conservative, and direct/outspoken. These three factors explained 37.9% of the variance. Five adjectives did not load strongly on any factor (factor loadings < .40) and are listed at the bottom of this table. These items were dropped from the scale for subsequent analyses. Thus, the version of the scale that should be used for future analyses is shown in Appendix A.

Proponents of Andersen's work might further argue that a two factor solution is more appropriate for men, but a three factor solution more suitable for women. Further factor analyses were conducted on the men alone, women alone, and men and women combined, using both two and three factor solutions. In this case, the maximum likelihood factor extraction was used to estimate the goodness-of-fit for each of the models. In each case, the three factor solution accounted for 38% of variance, but the two factor models accounted for 31% variance. The goodness-of-fit statistic was significant (p < .0001) and very large (χ² ranging from 1,959 to 5,152) for all 6 models, likely due to the large sample size, yielding inconclusive comparisons.

Although a three factor solution was superior to a two factor for this sample, the three factor solution was unsatisfactory on several accounts. For one thing, a solution that explains less than 40% of the variance in sexual self ratings is poor. Andersen's previous reports failed to mention the variance accounted by either their two and three factor solutions for men and women respectively. Andersen claimed using eigenvalues as the basis for selecting the number of factors. Using this widely accepted strategy with the current data indicated a nine factor solution for this heterogeneous sample (Canadian and U.S., men and women combined). Table 3 lists the factor loadings of an exploratory principal components analysis with a nine factor solution (using oblimin rotation with Kaiser normalization). This solution was stronger than either of the two or three factor models: the nine factor solution accounted for 58.8% of the variance in participant ratings. This model also highlighted the breadth of people's conceptualizations of their sexual selves along nine dimensions, a range of positive and negative traits referencing behaviors, emotionality, and power, arguably a more sophisticated approach.

Average total scores for each of the subscales for all subjects, along with correlations between each of the scales are presented in Table 4. The subscales were correlated but independent structures, the correlations between them were significant but small. The coefficient alpha estimates of reliability for each subscale were all reasonable.

A main premise underlying the separate assessment of sexual self-schemas among the genders was that men and women characterized their sexual selves differently. A comparison of men and women on all adjectives in a MANOVA between-subjects design tested this assumption. The main effect test of gender differences across all adjectives was significant, F(40, 792) - 4.88, p = .0001, η² = .198, a small effect (η² < .2; Cohen, 1988). Univariate tests of gender differences across each adjective revealed small significant gender differences for about half of the adjectives. Table 5 lists the means and univariate tests of difference for the items that led to significantly different responding among men and women; non-significant results were not reported. All the effect sizes were small.

Using the item associations identified in the preceding factor analysis, a further test of gender differences was conducted. Subscale scores were calculated for each of the three factors (loving/warm, reserved/conservative, and direct/outspoken) simply by adding values for the items that loaded on each factor (reversing the scores for "unromantic" since it loaded negatively on the loving/warm dimension). A MANOVA test of gender difference was significant for all three sub-scales combined, F(3, 829) = 17.42, p = .0001, η² = .059, a very small effect. Means for each subscale and gender are reported in Table 6. Univariate tests of differences found that most of the gender differences were due to the fact that women scored higher than men on the loving/warm warm factor, F( 1,831) = 41.38, p = .0001, η² = .047, and the reserved/conservative factor, F(1, 831) =15.55, p = .0001, η² = .018, but men and women scored almost equal on the direct/outspoken factor.

One further method of comparing genders involved classifying participants into one of four groups — aschematic, positive, negative, and co-schematic — and comparing the proportion of men and women in each group. To determine membership in each group, participants were classified as either high or low on the negative subscale (reserved/conservative) and the positive subscales (loving/warm + direct/outspoken) using a median split. The median for the negative scores was 25 and the median for the positive scores was 119. Those who rated themselves below the median on both positive and negative dimensions were classified as aschematic; above the median on both were classified as co-schematic; above the median on positive dimensions but below the median on negative dimensions were classified as positive; and those who rated themselves as above the median on negative the dimension but below the median on positive dimensions were classified as negative.

The proportion of men and women categorized into each of the four schematic groups is presented in Table 7. Overall, about a fourth of all participants fell into each category. A chi-square analysis found that the proportions of men and women in each category were unexpected, χ²(3, N = 833) = 11.6, p < .0001. Men were more likely to rate themselves as aschematic and less likely to rate themselves as co-schematic than expected. Contrary to Andersen and colleagues' earlier work, 25% of the men in this study rated their self as negative suggesting that many men view their sexual self negatively. Also unexpectedly, almost a third (31%) of the men rated themselves as aschematic, having no outstanding negative or positive characteristics. This contrasts with women's reporting: fewer women rated themselves as aschematic and more as co-schematic than expected. Thus, women were more likely to see their sexual selves as having both positive and negative dimensions than men, but men were more likely (than chance) to see their sexual self as either negative overall or low on both positive and negative dimensions.

A MANOVA comparison of the three subscales across country of origin yielded a significant main effect, F(3, 829) = 6.43, p = .0001, η² = .023. This difference was a medium effect size, suggesting that overall Americans rated themselves differently than Canadians. The means for each of the subscales across countries can be found in Table 8. A test of univariate differences on these scales found a significant difference only for the direct/outspoken subscale, F(1, 831) = 11.32, p = .001, η² = .013, suggesting that American students rated themselves slightly higher on direct/outspoken variables, but this was a very small effect.

Combining the adjectives from the men's and women's sexual self-schema scales into one scale illustrated that both men and women relied on the same basic dimensions when judging their sexual selves: loving and warm, reserved and conservative, and direct and outspoken. This finding countered Andersen's earlier work that was premised on the idea that men and women evaluated themselves using different dimensions. Andersen and colleagues proposed an empirically derived factor model, but in this study, their three factor solution was basically confirmed for both men and women. Specifically, this study supported the notion that men and women have both negative and positive dimensions to their sexual self. So the results of this study were consistent with the existing theory about sexual self-schemas: a three factor model is the most preferable factor model for this set of traits in a mixed sample. However, a nine factor solution suggested an even stronger model of sexual selves. There must be more than three dimensions to a person's sexual self-schema, and a scale using a wider range of items would lead to a different pattern of results.

Administering the same items to both genders also permitted a test of whether men and women rated their sexual selves similarly. This study found that women and men considered the same broad dimensions, yet when comparing differences in how men and women rated themselves on specific adjectives and across the three main dimensions, a pattern emerged. Men rated themselves slightly, but significantly, lower on the loving/warmth dimension and reserved/conservative factors. Thus, men see themselves as marginally less loving and warm and less reserved/conservative. It is difficult to know why these results were obtained. It may be that men may in fact be less loving/warm, perhaps because they conform to gender role expectations. Another explanation is that men may be less willing than women to express these emotions, yet have the same basic emotional experiences as women (Jansz, 2000), especially warmth and love (Alexander & Wood, 2000). A social psychological approach to gender and emotion suggests that these differences were not, presumably, because of any innate difference between the sexes (consistent with a sociobiological approach), but rather due to the internalization of gender role and expectations in this particular context. Participants could have used the masculine stereotype as "unemotional" and "sexually outgoing" as a heuristic to guide their ratings on the scale, and even, perhaps, their behavior.

There are limitations to the findings in this study. The sample of participants was an English-speaking, mostly urban cosmopolitan sample of university students in Montréal and New York City. Montréalers, in particular, are renowned for their egalitarian and liberal attitudes towards sexuality; the same may be said of residents of New York even though this sample was drawn from a college in New York's only Republican borough. Of course, older respondents or respondents who are less educated or live in rural areas might also respond differently.

Another limitation is more conceptual. It may not be that men considered themselves less loving/warm, for example, rather it could be that men connote different meanings to the words "loving" and "warm" in comparison to women. However, this is unlikely since Plaud, Gaither, and Weller (1998) had men and women rate 400 adjectives for their sexual connotations and found very few gender differences in the sexual meanings of words. Furthermore, the current study is based on items very similar to items used to assess gender conformity, thus this measure may be simply measuring a person's sense of congruence with the stereotype of man or woman as a sexual person.

There are also a few problems that are more fundamental. In an attempt to keep the assessments of sexual selves covert, Andersen and her colleagues used personality trait adjectives in their scales that were not obviously related to sexual functioning. The traits that they identified as tapping into the sexual self schema were mostly psychological trait-related adjectives. However, there are other dimensions that might be relevant when thinking about one's sexual self (e.g., heterosexual, monogamous, etc.), and this method does not permit these considerations. A solution to both these limitations would be to develop a more comprehensive measure, combining the psychological dimensions in this scale with other qualities related to a person's sexual self-conceptions such importance, strength, and range or direction of sexual desire, as well as feelings of self-efficacy in the sexual context.

This study has cleared the way for some interesting and important research. The first step should be to establish basic reliability and validity for the composite sexual self-schema scale in other populations, perhaps using a more comprehensive assessment of sexual self-schemas. It may also be profitable to use the scale in a cross-sectional or longitudinal study exploring whether sexual self-schema change over the course of one's life and how. It is clear that self-schemas influence experiences, and vice versa, so changing circumstances may mean changes in self-schema. Further research can also now move towards addressing the relationship between sexual self-schema and other qualities associated with sexual relations. For example, it might be useful to explore the relationship between gender roles and sexual self-schema. Since many gender and sexuality theorists are moving away from simply considering cognitive representations of self towards a more dynamic and socially based approach (i.e., sexual scripts), it would be interesting to explore the relationship between one's sexual self-schema and gender role scripts. Is there a disjunction between how we think of our selves and how we behave? Under what circumstances do these differences occur? Are they more prominent for women, who because of social restrictions in our society, find that they may not feel free to act in accordance with how they see themselves? More generally, are there changes in sexual self-concept in response to changes in context?

It is also tempting to speculate on the degree to which our sexual self-schema is relationally based. That is, how much of our sexual self-schema is derived by the sexual relationships we have? Moreover, are there particular combinations of sexual self-schemas related to especially healthy or troublesome sexual relationships? For instance, what happens when aschematics pair-up with those who have positive schemas? The sexual self-schema scale developed here could help to answer these questions.

This research was supported by a grant from the Foundation for the Scientific Study of Sexuality. The author would also like to thank the following for their research assistance: Sarah Bloomfield, Veronica Asgary, Minji Kang, Karen Pasture, and Yadira Del Rio.

DIAGRAM: Figure 1. Schema Categories.

Instructions: Below is a listing of adjectives people often use to describe themselves as a sexual person.

• For each word, consider whether or not the term describes how you feel about yourself as a sexual person compared to others of your same gender and age.
• Rate each term on a 7-point scale ranging from 0 = not at all descriptive of me to 6 = very much descriptive of me to indicate how accurately the adjectives describe you.
• There are no right or wrong answers.
• Please be thoughtful and honest.
• If you have any difficulties, please use this question as a guideline: To what extent does the term ----- describe me?
Not at all Very much descriptive descriptive of Me of Me Romantic 0 1 2 3 4 5 6 Soft-hearted 0 1 2 3 4 5 6 Passionate 0 1 2 3 4 5 6 Powerful 0 1 2 3 4 5 6 Warm 0 1 2 3 4 5 6 Outspoken 0 1 2 3 4 5 6 Loving 0 1 2 3 4 5 6 Spontaneous 0 1 2 3 4 5 6 Timid 0 1 2 3 4 5 6 Independent 0 1 2 3 4 5 6 Feeling 0 1 2 3 4 5 6 Sympathetic 0 1 2 3 4 5 6 Domineering 0 1 2 3 4 5 6 Arousable 0 1 2 3 4 5 6 Stimulating 0 1 2 3 4 5 6 Revealing 0 1 2 3 4 5 6 Aggressive 0 1 2 3 4 5 6 Direct 0 1 2 3 4 5 6 Warm-hearted 0 1 2 3 4 5 6 Frank 0 1 2 3 4 5 6 Exciting 0 1 2 3 4 5 6 Experienced 0 1 2 3 4 5 6 Sensitive 0 1 2 3 4 5 6 Uninhibited 0 1 2 3 4 5 6 Reserved 0 1 2 3 4 5 6 Embarrassed 0 1 2 3 4 5 6 Conservative 0 1 2 3 4 5 6 Unromantic 0 1 2 3 4 5 6 Compassionate 0 1 2 3 4 5 6 Cautious 0 1 2 3 4 5 6 Self-conscious 0 1 2 3 4 5 6 Straightforward 0 1 2 3 4 5 6 Inexperienced 0 1 2 3 4 5 6 Prudent 0 1 2 3 4 5 6 Individualistic 0 1 2 3 4 5 6 Sensual 0 1 2 3 4 5 6