88 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS We mentioned the absolute judgment or rating scale methods. It should be remembered that no judgment is ever strictly absolute but is always com- parative in the sense that one's experience invariably influences the so- called absolute judgment. As a consequence, at least the most recent ob- server experiences should be controlled or taken into account. A rank or- der indication with more than five items may cause trouble in interpre- tation for the reason that interest wanes with the long task of arranging in order, and with lack of interest goes the sense of responsibility for best effort. Before turning to the use of the methods we have just mentioned in evaluating cosmetics, we will examine in broadest terms how they may be applied to "difference" determinations preference study, and quality anal- ysis difference. The difference methods may be divided into two groups, one of which includes a designated standard for reference and the other requires that the observer examine all stimuli from some particular point of view dictated by the instruction. In certain cases, one approach is more sensitive, while in others, the reverse is true. The preference studies, by the same token, can be divided into comparative and absolute methods ranging from the simple "which do you prefer" to the absolute rating scale of nine points applied to a single product. In many instances, we find com- parative methods impractical and must resort to the absolute. When both can be used, we have observed excellent agreement or correlation. Finally, the analysis of quality may be approached directly or indirectly, by classification or comparison, by description, or by general behavior study. Let's begin with a study of shaving cream to illustrate the propositional approac h which precedes the test work. When possible, the univariate procedure is most valuable in order that the relative importance of each variable can be assessed as it is individually studied. As Ev^ru^zIOS or Ss^viso C•s^•as (^ Ss•ISS or PROPOSITIONS) General Propositions 1. Differences Exist: The composite of properties of shaving creams produce measurable differences in their effects on shaving. 2. Quality Uniformity Makes Measurement Valid: The variations in the quality of a shaving cream produced by any given manufacturer is less than the differences between the creams of various manufacturers. 3. The Several Effects Produced Will Be of Measurable Importance: The differences in the whisker conditioning abilities of the several creams will produce a number of effects, each of which can be measured and as- signed weights in relation to its importance to shavers. 4. Some Desirable Properties: Other things equal, a good cream will

ANALYSIS OF PSYCHOLOGICAL PROPERTIES OF COSMETICS 89 promote shorter total shaving time, closer shaves of greater permanence, less discomfort during and after shaving, less trouble for the shaver and more stable quality than a poor shaving cream. From such a schematic series, the experimenter can readily list his pa- rameters, designing in advance his test matrix. Also, he knows in advance how many subjects, observers, shaving and measuring instruments will be required. Since the standardization of procedure and statistical treatment of results is effected before beginning work, the postexperimental regrets are reduced. In addition to the laboratory work, the field testing is out- lined in advance. If it is necessary to learn how long a given package of shaving cream lasts when used at home, whether or not it is confused with other toilet goods, the different uses to which it is put, and the like the supplementary field explorations are outlined at the time of starting the more rapid and thorough laboratory work. FRo• MaNy We have found it essential to approach each problem from many different angles. For example, it is highly desirable to standardize and control all physical variables with the greatest precision. As a possible experimental procedure, one may elect to have each shaver prepare his face on one side with Cream L and on the other side with Cream J, and report which is su- perior in regard to the several criteria. Alternately, a trained barber would split the faces of the panel of subjects along an imaginary, vertical line, prepare each, and shave them. Possibly a third group would use Cream L on one day and J on another. The psychologist makes certain that the little clues of identification, along with distractions or other influences, are minimized. Then, and only then, can we confidently restrict ourselves to the reactions of small groups of consumers, analyze the results, and extra- polate to the total population. SELECT PANELS OF TRAINED OBSERVERS While the consumer work is in progress or before it starts, our practice is to subject the products under observation to the controlled scrutiny of panels of trained observers. From these groups come the valuable analytic data which completes the picture. Often the responses of the trained observers are directly predictive of the results to be found when larger groups of untrained or naive observers are called in. When one lipstick was compared with another on the basis of physical data only, the two products were essentially equivalent in per- manence or indelibility, smear resistance, and hue. Trained observers employed in a flavor analysis which was conducted under reduced illum- ination found that one was sweeter and more floral, while the other was dominantly fruity. The two could be differentiated visually in the stick