Understanding multi-channel research shoppers: an analysis of Internet and physical channels

Abstract

The purpose of this study is to analyze how channel characteristics influence the channel attitudes of multi-channel research shoppers. In addition, the study also considers consumers’ complex shopping behaviors that involve using one channel for search and another for purchase simultaneously. Survey data were collected from 191 consumers with recent (i.e., in the last 3 months) multi-channel experiences. Qualified respondents had either searched in physical retail stores and purchased on the Internet or searched on the Internet and subsequently purchased in physical retail stores. The research hypotheses were examined using multiple regression analysis. The results show that customer perceptions of channel characteristics drive channel choice attitudes. Importantly, we find the significance of channel characteristics varies between the Internet and the physical channel. The research findings deepen our understanding of shopping behaviors in the multi-channel context. Practitioners may use the findings to manage the characteristics of the complementary channels while also designing marketing programs to create positive consumer attitudes toward search and purchase.

Appendices

Appendix 1. Multiple regression equations

To test hypotheses, this study formed four regression models as follows:

Model one

The goal of the model is to test Hypotheses 1a and 2a. 

$$SA_{physical} = \beta_{0} + \beta_{1} IA_{physical} + \beta_{2} SE_{physical} + \varepsilon$$

(1)

SA search attitude, IA information availability, SE search effort, β ₀ intercept, β _i regression coefficient, ε error term.

Model two

The goal of the model is to test Hypotheses 1b and 2b. 

$$SA_{Internet} = \beta_{0} + \beta_{1} IA_{Internet} + \beta_{2} SE_{Internet} + \varepsilon$$

(2)

SA search attitude, IA information availability, SE search effort, β ₀ intercept, β _i regression coefficient, ε error term.

Model three

The goal of the model is to test Hypotheses 3a, 4a, 5a, 6a, 7a, and 8a. 

$$\begin{aligned} PA_{physical} &= \beta_{0} + \beta_{1} PC_{physical} + \beta_{2} SQ_{physical} \\ &\quad + \beta_{3} PE_{physical} + \beta_{4} PR_{physical} + \beta_{5} IA_{Internet} + \beta_{6} SE_{Internet} + \varepsilon \\ \end{aligned}$$

(3)

PA purchase attitude, PC purchase convenience, SQ service quality, PE purchase effort, PR purchase risk, IA information availability, SE search effort, β ₀ intercept, β _i regression coefficient; ε error term.

Model four

The goal of the model is to test Hypotheses 3b, 4b, 5b, 6b, 7b, and 8b. 

$$\begin{aligned} PA_{Internet} &= \beta_{0} + \beta_{1} PC_{Internet} + \beta_{2} SQ_{Internet} + \beta_{3} PE_{Internet} \\ &\quad + \beta_{4} PR_{Internet} + \beta_{5} IA_{physical} + \beta_{6} SE_{physical} + \varepsilon \\ \end{aligned}$$

(4)

PA purchase attitude, PC purchase convenience, SQ service quality, PE purchase effort, PR purchase risk, IA information availability, SE search effort, β ₀ intercept, β _i regression coefficient, ε error term.

Appendix 2. The Chow test (Chow 1960; Doran 1989)

The hypotheses of the Chow test are:

H ₀

The two regression models have the same regression coefficients and intercepts.

H ₁

The two regression models have not the same regression coefficients and intercepts.

The hypotheses can be tested by the F ratio:

$$F = \frac{{\left( {SSE - SSE_{1} - SSE_{2} } \right)/K}}{{\left( {SSE_{1} + SSE_{2} } \right)/\left( {N - 2K} \right)}}$$

(5)

SSE: sum of the squared errors from the whole sample; SSE _i : sum of the squared errors from the ith sub-sample; K: the total number of estimated regression parameters (i.e. the number of independent variables plus one intercept); N: the total sample number.

The F ratio will be distributed as F(K, N-2 K).