WHY WOMEN USE MAKEUP 131 at decreasing the load of their negatively perceived attractiveness and/or global self- image), while those of class B subjects refer rather to a "seduction" functional profile (i.e., aiming at supporting and promoting a positive self-image). If we analyze more precisely these two classes, we observe a statistically significant division of the "seduc- tion" class (i.e., class B) into three different subclasses (Table II), depending on the relative valuation of the natural and made-up face of each subject. The same analysis for class A did not reveal sustainable subclasses. In the first subclass (subclass B 1) of class B, the made-up face is considered to be more valuable than the natural face and less variability is introduced in the makeup structure. For example, these subjects claim "Without makeup, I'm insipid" (81 %), "I don't like my face without makeup" (69%), or "When I'm made-up naturally, I feel myself' (100%). For the third subclass (subclass B3) and in comparison to subclass Bl, the "natural" face is rather perceived as more valuable than the made-up face. For instance, these subjects typically agree with "I use makeup to control the image of myself" (30%) or "In the evening, I make myself up to allure" (100%). Concerning the intermediate subclass (subclass B2), "natural" and "made-up" faces are considered as valuable as each other relative to the subject's personal and professional life. For example, women of this subclass claim "I always make myself up when I am in contact with people" (70%), Table II Analysis of the Makeup Questionnaire for Class B (n = 4 9) Sentences When I'm made-up, I feel glamourous With makeup, I want to smile I always make myself up when I am in contact with people When I'm made-up, I feel natural When I'm made-up naturally, I feel myself I'm not afraid to change my makeup process to remain faithful to myself I make myself up in a more intensive way in the evening A woman who does not make herself up, it is a pity In the evening, I make myself up to allure Without makeup, I'm insipid A woman who does not make herself up is a woman who overrates intellectual dimension to the detriment of appearance I use makeup to control the image of myself When I'm made-up, I want to allure I don't like my face without makeup Without makeup, I don't like myself When I'm made-up, I feel sensual Percentage of agreement for specific sentences Subclass Bl Subclass B2 Subclass B3 (n = 16) (n = 23) (n = 10) 5 6B 61B SOB 44AB 44AB 6QB 5 6A 7QAB soA 6A 4 SB 4 QAB lQQB S7AB 60A 13A 26A 20A SSAB 91B lOOB 5 6AB 3 5A 60AB 44A 7QAB lOOB SlB 39A soAB QA 4A 4 0AB 75B 61AB 30A 75B 61AB SOB 69B 35A 4 0AB SOB 17A 4 0AB 69AB 7 4AB 90B Significance S (p = 0.0 4 ) NS S (p = 0.03) S (p = 0.0 4 ) S (p = 0.0 5 ) S (p 0.01) S (p 0.01) NS S (p = 0.01) S (p = 0.03) S (p = 0.01) NS NS S (p = 0.0 4 ) S (p = 0.0 4 ) S (p = 0.01) When there is a significant difference among classes, we indicate by letters (A, B) the class to which it belongs. If the classes have the same letter, they are comparable if not, they are significantly different.

132 JOURNAL OF COSMETIC SCIENCE but they also enjoy their face without makeup ("Without makeup, I don't like myself" 17%). Finally, Table III summarizes the classifications detailed above. ORGANIZATION OF PSYCHOMETRIC VARIABLES IN SUBJECTS As a first result, the ANOV A revealed, interestingly, that psychological variables ob- tained from psychometric self-questionnaires were not statistically related to aging (Table IV). Thereafter, a principal components analysis (PCA) was carried out in order to further investigate the relations between measured psychometric parameters, and to reduce data dimensionality. The results performed on the nine psychometric variables retained (i.e., the STAI score, the extroversion score, the neuroticism score, the RAS score, the SEI global score, and the four SEI sub-scores) are shown in the correlation circle represented in Figure 1 a. The PCA identified three main factors that explain 71.8% of the total variance. The first two factors account for 62.6% of the initial variability of the data, while factor 3 accounts for 9.1 % of the total variance. Their projection in the factors' space shows that the axis corresponding to factor 1 (accounting for 51.6% of the total variance) is essentially related to the SEI general score, the RAS score, and the ST AI T-anxiety score, which is inverse! y correlated to the SEI general score. The axis corresponding to factor 2 represents 11.1 % of the remaining variance and is related to the EPI extroversion and neuroticism scores. The projection of the psycho- metric variables in the other factors' space, represented in Figure 1 b, shows that the axis corresponding to factor 3 is only related to the SEI home-parents score and independent of the others psychometric variables. These results show clearly that five psychometric variables out of the nine are able to describe satisfactorily the differences between the volunteers. Note that the projection of the psychometric variables in the factors' spaces (Fl vs F2 and Fl vs F3) confirms that the variable "age" is independent of all these variables. Concerning the psychometric classification of our subjects, significant differ- ences between classes (class A and class B) were observed for the five psychometric variables (see Table V for a statistical summary). For general self-esteem, the mean(± SD) score was 12.4 (±3.9) for the volunteers of class A (n = 21) and 20.6 (±2.4) for the volunteers of class B (n = 49). These results reveal that the female subjects of class B express a better self-esteem than those of class A. Concerning the mean of the STAI T-anxiety scores, data obtained in class A were significantly higher than those of class B, with respective scores of 46.6 (±8.1) and 35.6 (±5.5). The volunteers of class A can therefore be labeled as "anxious," opposite those of class B. Concerning the RAS score, participants of class A (-5.2 ± 16.2) were labeled as Table III Classification of Volunteers According to Their Use of the Makeup Process (n = 70) Main background for makeup Self-perceived value of natural face and make-up face Class A (n = 21) Camouflage Natural image made-up image Subclass Bl (n = 16) Natural image made-up image Class B (n = 49) Seduction Subclass B2 (n = 23) Natural image = made-up image Subclass B3 (n = 10) Natural image made-up image