156 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS Burning/stinging Very high burn/sting High burn/sting Moderately high burn/sting Moderate burn/sting Slight burn/sting Very slight burn/sting No burn/sting Right __ Very high burn/sting __ High burn/sting __ Moderately high burn/sting __ Moderate burn/sting __ Slight burn/sting __ Very slight burn/sting __ No burn/sting OTHER DISCOMFORT (please describe): ANALYSIS The non-parametric Wilcoxon Sign Rank test for paired data was used to compare the irritation potential between the two products for each attribute within each study (9). The response variable consisted of the mean of the sum of the scores for all evaluations over the five-day period. The means and standard deviations for each of the four studies are reported as descriptive statistics. The absolute value of the difference between treat- ment scores for each subject was used to assign the ranks used in the analysis. In cases where a significant difference was determined, treatment with the product with the lower mean did match the Wilcoxon statistic and thus induced less of the specified irritation. Pearson correlation coefficients were calculated to determine the degree of correlation between all attributes for each test material in the four studies. For the purpose of discussion, coefficients between 0.65 and 0.70 are considered to indicate slight correla- tion of attributes. Coefficients greater than 0.70 and less than 0.85 are considered to indicate strong correlations among the attributes compared. RESULTS In the third study (Table III), comparing two synthetic complexion bars, there was no Treatment with Bar A produced significantly less erythema than treatment with Bar B. No significant differences were determined between the bars in terms of the subjective irritation attributes of dryness, tightness, itching, and burning/stinging. When as- sessing the relationship among the five attributes for Bar A in this study, there was a high degree of positive correlation between the attributes of tightness and dryness. Burning was slightly correlated to dryness, tightness, and itching. For Bar B, tightness was highly correlated positively to dryness and slightly correlated to burning. Results from the second study (Table II), comparing two complexion bars, indicated significant differences between the bars for both the clinical and subjective irritation

SUBJECTIVE IRRITATION BY SOAPS 157 Table I Facial Irritation Study 1: All-Family Bars Accumulated mean sum -+ S.D. Bar A Bar B p-Value Erythema 4.47 + 5.28 20.00 + 16.57 Dryness 7.0 + 4,86 8.35 + 5.21 Tightness 7,18 -+ 4.65 7.53 -+ 5.24 Itching 1.76 -+ 3.34 1.47 -+ 2.85 Burning/stinging 3.24 -+ 3.33 3.82 -+ 4.13 0.002 0. 190 O.54O 0.370 O.64O Pearson correlation coefficients,' Bar B Dryness Tightness Itching Burning/ s?inging Dryness Tightness Itching Burning/stinging • 0.19 0.07 -0.21 0.01 0.88 0.002 0.89 '• 0.19 0.54 0.53 0.05 0.68 0.65 0.17 0.36 0.54 0.47 0.70 O.66 0.32 Erythema Dryness Tightness Itching Bar A Erythema Dryness Tightness Itching responses. There was no correlation indicated among the clinically scored erythema or subjective irritation attributes for Bar C in this study. The Pearson coefficients describing the relationship among the attributes for treatment with Bar D only, however, indicated some degree of positive correlation between all attributes except between erythema and both tightness and itching. Erythema was slightly correlated to both dryness and burning. Dryness moderately correlated to all other attributes, and tightness was highly correlated to both itching and burning. In the third study (Table III), comparing two synthetic complexion bars, there was no significant difference in the induced clinical erythema or in the subjective irritation attributes of dryness, tightness, itching, or burning/stinging between the two bars. For Bar E as well as Bar F, there was a moderate-to-strong positive correlation among all of the subjective irritation attributes, but no significant relationship between these and erythema. Washing with the synthetic detergent bar resulted in significantly less induced ery- thema than washing with an all-family bar in the fourth study (Table IV). No signifi- cant difference was perceived between the bars in terms of any of the subjective irrita- tion attributes. There was no significant correlation noted among any of the attributes for either bar. DISCUSSION Little work to evaluate the phenomenon of subjective irritation has been reported (2).