This is the first part of our three studies. Three kinds of multivariate approaches are methodologically possible to evaluate the pure temperamental factors free from social desirability factors in personality questinnaires. The first approach would be to obtain the factor-estimate scores by the multiple regression method(called the complete estimation method by Harman 1967)using the factor structure matrix of the predictive variables on the wanted pure temperamental factors which had been obtained for our special purpose by the partial correlation analysis of higher order. This is a special kind of "group principal factor analysis" combined with an ordinary oblique factor analysis of the residual correlation matrix out of which the social desirability factors had been partialed. The initial correlation matrix consists of the intercorrelations among the original personality questionnaire scores(12 YG scales in standard situation)plus the same number of scale scores in a valuation situation which are scored from the items of the same scales with the same keys. The second one is to construct the pure temperamental factor scales by items selection according to the principle of factor-trueness proposed by Cattell and Tsujioka, where the items will be selected by suppressing the social desirability factors and at the same time by aiming the true directions of the pure temperamental factors in the total factor space. The third one is to directly orthogonalize the original number of personality questionnaire scales to those social desirability factors. The present study is the first one of these approaches. In the authors' previous studies(1975a, 1975b)three social desirability factors : Social Desirability Factor for Emotionality(SD-E), Social Desirability Factor for Introversion-Extraversion(SD-I)and Personal Desirability Factor on Reflectiveness(PD-R); seven temperamental factors: Emotional Instability(^1E), Dominance(^1D), Impulsiveness(^1I), Unreflectiveness(^1R), Frustrativeness(CO), Aggressiveness(AG)and Fantasticalness(DO), named towards high score in YG scales, were established as orthogonal factors between two domains while as oblique factors within each domain(either temperamental factor space or social desirability factor space). Ten pairs of the factor-estimates of 3 social desirability and 7 temperamental factors were calculated from 300 college(200 male and 100 female)students by Formula No.3 using 12 YG scores only or those plus 12 valuation judgement scores. The correlation matrix(partly shown in TABLE 1)among 10 pairs of factor-estimates was analysed by the principal component analysis and then 10 principal components were rotated to Varimax solution as shown in TABLE 2. The results in Tables 1 and 2 reveal that each pair of the factor-estimates of the 7 temperamental factors behaves as if the pair is almost the same measures in the total factor space. However each pair of 3 social desirability factors behaves differently from one to another. In order to recover the reduction of the factor saturation of each social desirability factor, the writers decided to construct three scales for measuring three social desirability factors which can be selected from the items in the valuation judgements by the principle of factor-trueness by suppressing the temperamental factors in this case. Finally a new experiment by the same method for checking the effectiveness of excluding the social desirability factors was suggested to be applied to the data which will be obtained in a more fakable situation.