SKIN CONDITION MEASURED BY SONIC VELOCITY 15 Table IV Variance in Consumer Responses Predicted by Physical/Visual Measures QUESTION 1--Please tell us how much you DIS- LIKED or LIKED the product you tried for us overall by circling a number inside the appropriate box. (9-pt. scale, 9 = Like the most possible). QUESTION 3--Now, considering ONLY the EF- FECTIVENESS of the product you tried for us, please tell us how much you DISLIKED or LIKED this product. (Same scale as #1). Measure AE3 D28 AE28 AD28 AD3 D3 E3 E28 Variance Predicted* 13% ' 12% 10% 8% 4%- 3% 2% QUESTION 2--Now, considering ONLY the way the product you tried for us PRE- VENTED DR Y SKIN, please tell us how much you DISLIKED or LIKED this product. (Same scale as #1). Measures Variance Explained* D28 14% 1 AE3 11%] 1 AD28 11% / / ZXE28 •ø•/ / AD3 D3 4% ]/ E3 2% [ E28 2% _i QUESTION 4--Please LOOK AT and FEEL your HANDS carefully. OVERALL, how would you rate the skin con- dition of your HANDS? (7-pt. scale, 7 = Very poor). Measures Variance Explained* Measures Variance Explained* D28 13% '] D28 20% ] lo%/I AD28 8%1 AD28 10%1 ! AE3 8%! AE28 8%[ [ D3 7%[ AD3 6%_]1/ AE28 6% D3 3% 1] AD3 6%] E3 3% E3 E28 2% • E28 4% *Equal to r • times 100%. Brackets enclose correlation coefficients not significantly different at c• .05. Symbols for Measures.' E3 = day 3 elastic modlus E28 = day 28 elastic modulus D3 = total overall dorsal derm grade on day 3 D28 = total overall dorsal derre. grade on day 28 AEx = change in elasticity from day x to day 0. ADx = change in derm. grade from day x to day 0. An examination of the dermatologist's visual grading and calculated elasticity measures in Table IV reveals that the percent of variance in the consumer responses to product-related questions (questions 1 to 3) predicted by physical/visual measures, is small, at most explaining 14% of the variance. Further, for the product-related questions 1 to 3, the change in elasticity after three days is not significantly different (c• .05) from the visual grade after 28 days of product use as a predictor of response. In each of these cases, changes in the elastic moduli are significantly better predictors of panelists' responses than the absolute value of the moduli on a given treatment day. This observation reinforces the point that with the sonic velocity method, changes in the calculated elasticity modulus, where each panelist is used as his own control, are more relevant measures of changes in the outer layers of the skin than the absolute value of the calculated modulus. For the single question (//4) where the panelists were asked to rate the visual condition

16 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS of their skin, the dermatologist's visual grade on the same day as the questionnaire was completed (day 28) is significantly better at estimating the variance of the response. In this question, the expert judge and the study participants have been asked to grade the same attribute on the same day. Still, only one fifth of the variance is explained. This observation implies that a significant amount of variance in the panelists' responses may be attributed to panelist-to-panelist variations in the use of the scales. Unlike the expert, the respondents had no training or common standard scale against which to judge their answers. In this light one should not be surprised at the low correlation between panelists' responses and the physical/visual measures each panelist's personal interpretation of the questions and the scales resulted in extremely high noise in the responses. The multiple correlation method (38) was used to arrive at a linear combination of the two measures which better predicts the panelists' responses to each of the questions. Table V sets out the results of this calculation. For the product-related questions (1 Table V Optimum Combination of Visual/Sonic Elasticity Measures Predictive of Consumer Responses Z(question) = fi2 Z(AE3) + fi3 Z(D28) Beta Values for the Multiple Regression Equation Correlation % of Variation Question// fi2 fi3 Coefficient Explained 1 -- .27 -- .27 .44 19% 2 -- .22 -- .29 .42 18% 3 -- .23 -- .30 .44 19% 4 + .15 + .40 .47 22% The questions and the measurement symbols are set out in Table IV. Z(i) is the reduced and standardized response for the variable i. fi2 and fi3 are the optimum weighting factors calculated as per ref. 38. through 3), a factor made up of approximately equal mixtures of the change in elasticity after three days of treatment and the final visual grade, significantly improves the explained variance in the panelists' responses compared to the variance explained by each factor alone. This result establishes that elasticity and visual data are both important elements of consumer product benefit recognition, and is consistent with the low level of correlation between these measures. Therefore, visual and elasticity data both describe important, different, aspects of skin condition. For question 4 where the panelists provided a day 28 self-evaluated visual grade, addition of sonic elasticity information does not significantly improve the predicted variance. With these combinations of visual and elasticity data, 20% of the variance could be explained for each question. This statistical analysis has established the following. Age and initial visual skin condition are factors underlying the absolute value of the calculated elastic modulus however, together they explain only a small fraction of the variance in this physical property of the skin. Of the elasticity measures, the change in calculated elastic modulus is the only relevant measure of product-induced elasticity alterations and is independent of age. It is a novel measure of skin condition and predictor of product effects, assessing skin properties not normally detected with conventional visual examination. The change in the derived skin elastic modulus after three days of