소비자 선택모형(consumer choice models) 연구들과 소비자 심리학(consumer psychology) 연구들은 제품의 속성 유사성이 대안 선택에 영향을 미친다는 것을 보여주지만, 그 효과에 대해서는 상반된 결과를 주장한다. 소비자 선택 모형 연구들은 한 대안이 다른 대안들과 속성이 유사할수록, 대안의 선택 확률은 낮아진다는 것을 보여준다. 반면에, 소비자 심리학 연구들은 대안의 속성 유사성이 제품에 대한 선호도를 높인다고 제시한다. 즉, 기존의 문헌은 속성 유사성이 대안의 선택확률에 긍정적, 부정적 영향을 모두 끼칠 수 있다는 것을 제시하고 있다. 본 연구는 이러한 두 개의 다른 연구 흐름을 통합하여, 유사성이 소비자 선택에 미치는 긍정적/부정적 영향은 대안에 대한 소비자의 친숙도에 의해 결정된다는 것을 보여준다. 본 연구는 소비자들이 제품에 친숙할 경우에는 속성 유사성이 대안 선택에 부정적인 영향을 미치나, 소비자들이 제품에 친숙하지 않을 경우에는 속성 유사성이 대안 선택에 긍정적인 영향을 미칠 수 있다는 가설을 제시하였다. 본 연구에서는 소비자의 대안 선택에 관한 이산선택(discrete conjoint choice)자료를 다변량로짓(multinomial logit)모형을 사용하여 가설검증을 실시하였으며, 연구의 결과는 본 연구의 가설을 지지하였다.
The current research examines the influence of the attribute similarity among competing brands on consumer`s choice. Attribute similarity is defined as the extent to which a brand has similar with other brands in terms of the attribute levels. Both consumer choice models and consumer behavior studies show that brand preference is affected by attribute similarity. However, the two streams of research emphasize different aspects of the attribute similarity effects. Empirical consumer choice models (e.g., nested logit models and logit models allowing for brand similarity) indicate that the probability that choice probability of a brand decreases as the brand is more similar to others in the choice set due to increased competition and substitutability with other similar brands, resulting in a negative effect on choice (e.g., Manrai 1995). On the contrary, consumer behavior theories suggest that a brand that is more similar to others is preferred to the brands that are less similar (e.g., Loken and Ward 1990). When a brand is more similar to others in attribute levels, the brand is perceived as being a typical brand, and people tend to prefer typical brands to less typical brands. This research attempts to integrate the two streams of research and presents that the influence of attribute similarity on choice is moderated by product familiarity which is defined as the extent to which a set of products are more familiar and easy to evaluate. In general, attribute similarity increases competition and substitutability, resulting in decreased choice probability. However, it is expected that influence will be more positive when products are less familiar, because attribute similarity is more likely to be used as an extrinsic cue. On the other hand, when the product is more familiar, the negative influence of attribute similarity on choice will be more prominent since attribute similarity does not play a role as a decision heuristic. Thus, it is predicted that when products are more familiar, consumers tend to prefer a brand that is more similar to others, whereas when products are less familiar, consumers prefer a more unique alternative. We test the moderating effect of product familiarity on the influence of similarity on choice by analyzing a set of discrete conjoint choice data. The data set includes consumer choice of a film camera out of three familiar products (i.e., 35mm cameras) and three less familiar products (i.e., APS cameras). The data were considered to be appropriate for the study because 35mm cameras are more familiar to consumers than are APS cameras, which were only newly introduced when the main study was conducted. A pretest also supported our conjecture regarding product familiarity showing that 35mm cameras were more familiar than were APS cameras. Thus, it is predicted that the effect on choice of attribute similarity is moderated by the type of camera: Attribute similarity will exert a negative effect on the selection of the 35mm camera, whereas attribute similarity will exert a positive influence on the selection of the APS camera. We tested the influence of attribute similarity by four different models that differ in the manner in which attribute similarity affects choice: The base model, the common attribute model, the interaction model, and the separate similarity model. The Base model assumes that the utility of brand is determined only by the level of attributes and the price. Thus, the model does not incorporate the influence of attribute similarity on choice. The common similarity model assumes that the attribute similarity of a brand with other alternatives in the choice set also influences choice. The Interaction Model includes the interaction between product type and attribute similarity that captures the differential influence of similarity on choice. It is predicted that the interaction term coefficient will be significantly positive, indicating that the influence of attribute similarity on choice is more positive for the APS cameras. The Separate Similarity Model estimates the similarity effect separately for 35mm and APS cameras. According to our theoretical assumptions, it has been predicted that the coefficient is negative for the 35mm camera but is positive for the APS camera. The estimated parameters supported our hypothesis. The base model indicated that all of the coefficients were significant and the directions were as predicted. The common similarity model indicated that the similarity index was negatively significant, revealing the overall negative influence of attribute similarity on choice. The results of the interaction model showed that the similarity effect was negative and the interaction coefficient was positive. These findings suggest that the influence of attribute similarity on choice was more positive for the APS cameras than for the 35mm cameras, thus supporting hypothesis 1. The separate similarity model shows the influence of similarity in greater detail. For 35mm cameras a brand that is more similar to competing brands have a lower probability to be selected than a brand with more unique attributes. For APS cameras, on the other hand, similarity with other brands increases choice probability. Consequently, the results in general show that for 35mm cameras that are more familiar, attribute similarity has a negative effect on choice, whereas attribute similarity has a positive influence for choice of APS cameras.
The findings present converging evidence showing that product familiarity moderates the influence of attribute similarity on choice. When the product is less (more) familiar, consumers are more likely to choose one of the alternatives that are more similar (dissimilar) to others. These finding may bridge the gap between empirical choice modeling and consumer psychology research streams that have focused mostly on either negative or positive influence of attribute similarity on choice.