Color Influence

Report

Subject:

Color (Psychological aspects)

Personality (Research)

Personality assessment (Research)

Authors:

Lange, Rense

Rentfrow, Jason

Pub Date:

12/01/2007

Publication:

Name: North American Journal of Psychology Publisher: North American Journal of Psychology Audience: AcademicFormat: Magazine/Journal Subject: Education; Psychology and mental healthCopyright: COPYRIGHT 2007 North American Journal of Psychology ISSN: 1527-7143

Issue:

Date: Dec, 2007 Source Volume: 9 Source Issue: 3

Topic:

Event Code: 310 Science & research Canadian Subject Form: Colour

SUMMARY AND DISCUSSION

As was anticipated, the present findings clearly indicate that people's color preferences, as assessed via the Dewey Color System Test, indeed provide meaningful information about their personalities, interpersonal styles, and behaviors. In particular, the data indicated that this test predicted with considerable precision all six of the Basic Interest Scales of Strong's Interest Inventory (Harmon, et al., 1994), and nearly all of the sixteen Primary Factors of Cattell's 16PF (Russel & Karol, 2002). As such, the present findings show far greater consistency than do those obtained in earlier research inspired by Luscher's (1971) work (cf., French & Alexander, 1972; Picco & Dzindolet, 1994; Seefeldt, 1979; Stimpson & Stimpson, 1979; Stone, 2001, 2003).

As was already found earlier by Seefeld (1973), men prefer yellow far less often than do women. Despite the finding of several statistically significant differences between men and women, our neural net approach proved of little use in inferring respondents' color preferences from their sex or vice-versa. It appears that the statistically significant associations are simply too weak to yield clearly identifiable effects. Note that this situation may well explain the inconsistent pattern of sex differences observed in earlier research (cf., Seefeldt, 1979 vs. Stimpson & Stimpson, 1979). By contrast, the findings provided strong support for the many-to-one hypothesis. Specifically, even when maximally different respondent configurations are constructed via the cluster analysis, highly similar color preferences continue to prevail. Note that our test of this hypothesis essentially relies on non-rejection of [H.sub.0]. However, the observed [lambda] statistics were so low as to be meaningless, and there is little reason to believe that these low values are the result of sampling biases.

A number of additional issues remain. For instance, it is not clear at this point whether our findings regarding personality must be attributed to the differences between the colors contained in the Dewey Color System Test and Luscher's color test, whether neural nets simply provide a superior method of analysis relative to the simpler statistical methods employed in previous research, or whether the particular test and analysis combination used here is responsible. Also, while neural nets form a holistic oriented approach in which choices are combined into patterns that simultaneously encompass respondents' likes and dislikes, neural nets have the disadvantage of being a black box only, i.e., they must be used "as is." In particular, neural net's weights are difficult to interpret directly (Gallant, 1994), and short of actually running the neural net even knowing all its weights provides little guidance in identifying which patterns of color preferences are associated with which particular personality traits.

Note that we have essentially bridged the relation between two types of tests, i.e., color preferences and the personality related variables assessed by the Strong and the 16PF. However, we have not yet established that the Dewey

...