The protocol adheres to the "Generic Protocol Food Choice Simulator," which complies with the Netherlands Code of Conduct for Scientific Practice and has been approved by the Social Sciences Ethics Committee of Wageningen University.
1. Setting Up the Virtual Store Equipment
Figure 1: The virtual store setup. The virtual store uses one PC equipped with three 42 inch LCD screens that render 180° visibility. A separate PC is added to accommodate the data management program. This PC enables a research coordinator to monitor the progress and to start new virtual environments without interrupting participants. Please click here to view a larger version of this figure.
2. Building Virtual Stores for Experiments
Figure 2: The virtual shop editor and examples of products in the product library. The editor has a drag-and-drop interface to allow researchers to easily select products from the library and directly place them on the shelves. In addition, a pop-up window can be used to either add or edit a product by clicking on a product in the library. Please click here to view a larger version of this figure.
3. Preparing the Data Management Program to Record Data
4. Participant Selection Criteria
5. Preparation for the Experiment
6. Running a Practice Test
7. Running the Main Test
Figure 3: An example of the observation window that signals the recording of data. When the data management program is recording data, the "Status data plugin" window and the "Status event plugin" show a green mark. Also, time should be elapsing and the number of samples should be growing. Please click here to view a larger version of this figure.
Figure 4: The visualization window displayed in the data management program. The orange bar represents the entire shopping time, since the participant entered the store until he/she pressed "Esc" to indicate the end of the shopping trip. The green bar denotes the time spent on the examined products. These outputs can be converted into tables that are easy to use in combination with SPSS or other statistical programs. Please click here to view a larger version of this figure.
8. Export the Data
Figure 5: Data profile filter scheme for exporting shopping-related behavior. The data profile filter allows researchers to select and export the data of interest. For example, this scheme opts for shopping-related behaviors (e.g., shopping duration, number of products examined, number of product purchased, and number of products returned). Please click here to view a larger version of this figure.
Figure 6: Data profile filter scheme for exporting movement-related behavior. This scheme filters the movement-related behaviors (e.g., moving speed and moving time) that occur when participants move in the store (speed >0.100 m/s). The behaviors and times when participants stand still are filtered out. Please click here to view a larger version of this figure.
The virtual store displayed using a PC with three 42-in LCD screens has been applied to examine the effects of supermarket layout on consumer shopping behavior (e.g., total shopping time, movement duration and speed, total number of products examined, and total number of products purchased) and perceived shopping experience. The virtual store enables the researcher to flexibly modify the attributes of store shelves (i.e., shelf length and shelf orientation) and to examine these effects in a laboratory setting.
As an example, results from the store layout study are provided. In the study, supermarket stores were built using 4 different layouts, in which shelf length (short versus long shelves) and shelf orientation (paralleled arrangement versus unparalleled arrangement) were varied. These stores are depicted in Figure 7.
Figure 7: Pictures of four store layouts in the store layout experiment. The layouts differ in shelf length and shelf orientation: 1) store with long and parallel shelves, 2) store with short and parallel shelves, 3) store with long and non-parallel shelves, and 4) store with short and non-parallel shelves. Please click here to view a larger version of this figure.
The study was performed in accordance with the "Generic Protocol Food Choice Simulator" and approved by the Social Sciences Ethics Committee of Wageningen University. All participants signed an informed consent form prior to participating in the experiments. In the present example, participants (n = 241, 71% female) were divided into four groups; each group visited one of four store layouts. Participants were trained on how to use the virtual store in a practice session. Next, they completed a shopping motivation manipulation task that asked them to recall shopping trips with either hedonic or utilitarian shopping motivations. Subsequently, the participants started the main test, in which they were requested to shop for a dinner using a shopping list. Participants were asked to imagine that they were shopping with either hedonic or utilitarian motivation (the same motivation as in the previous recall task was assigned). The shopping list consisted of fixed-choice (8 pre-determined types of products) and free-choice products (undetermined products from the fruit and vegetable category). The free-choice products were used to test the effects of store layout on the number of products purchased. Once the participants finished shopping, they filled in a computer-based questionnaire to evaluate their shopping experiences, perceptions about the store, and willingness to revisit the store.
The data management program recorded shopping behavior (e.g., total shopping time, moving speed, and total number of products purchased). Afterwards, variables were exported from the data management program to 3 separated tables: Table 1, Table 2, and Table 3. Table 1 presents the total shopping time, the total number of products examined, and the total number of products purchased by each participant. Table 2 presents the total movement duration (i.e., shopping time) that was selected from a filter of speeds higher than 0.001 m/s. Table 3 presents the moving speed that can subsequently be used to calculate the walking distance (walking distance (m) = average moving speed (m/s) x total moving time (s)).
Table 1: Examples of shopping-related behavioral data from each participant (i.e., total shopping time, total number of products examined, total number of products purchased, and total number of products returned), exported from the data management program. All shopping-related behavioral data from each participant should be organized in one row before transferring it to SPSS or other statistical programs. This exported data will be stored to the file called "Behavioral data" in the export folder of the data management program. Please click here to view a larger version of this table.
Table 2: Examples of movement-related data (i.e., the moving speed and the walking position of each participant), exported from the data management program. The movement-related data is selected when participants moved with speeds higher than 0.100 m/s. This selection filters out all data that occurred when participants stood still. All movement-related data from each participant should be organized in one row before being transferred to SPSS or other statistical programs. This exported data will be stored to a file called "Numerical data" in the export folder of the data management program. Please click here to view a larger version of this table.
Table 3: Examples of movement duration (indicated in the shopping duration column), exported from the data management program. The movement duration is retrieved from the behavioral data table that filters out the time during which participants did not move (speed <0.100 m/s). This duration is shorter than the total shopping duration. The exported data will be stored to a file called "Behavioral data" in the export folder of the data management program. Please click here to view a larger version of this table.
Once the data was exported, univariate ANOVA was applied to analyze the effects of shelf length and shelf orientation on in-store shopping behavior. The effects of store layout can be presented in various forms, such as bar charts and tables.
Figure 8 displays the total number of products examined and the total number of products purchased in the supermarkets with different store layouts. The results from the virtual store confirmed that store layout attributes, specifically the interaction of shelf length and shelf orientation, influenced the number of products examined (F (1,237) = 4.66, p < .05, ηp² = .02) and the number of products purchased (F (1,237) = 3.47, p = .06, ηp² = .01). The findings showed that when shelves were placed in parallel, the length of the shelves did not affect the number of products examined (Mshort ± SDshort = 16.12 ± 5.37, Mlong ± SDlong = 17.12 ± 5.99, F (1,237) = 0.81, p = .37, ηp² = .00), nor the number of products purchased (Mshort ± SDshort = 12.00 ± 2.77, Mlong ± SDlong = 12.22 ± 2.37, F (1,237) = 0.24, p = .63, ηp² = .00). In contrast, when the orientation of the shelves was unparalleled, shorter shelf lengths stimulated a higher number of products examined (Mshort ± SDshort = 17.62 ± 6.48, Mlong ± SDlong = 15.23 ± 6.45, F (1,237) = 4.65, p < .05, ηp² = .02) and purchased than longer shelf lengths (Mshort ± SDshort = 12.30 ± 2.15, Mlong ± SDlong = 11.35 ± 2.37, F (1,237) = 4.61, p < .05, ηp² = .02).
Figure 8: The total number of products examined (left) and the total number of products purchased (right) in a supermarket with different store layouts (short versus long shelves placed in a paralleled or in an unparalleled orientation). The total number of products examined (packages or items) increased every time the participants clicked on a product. This number differs from the total number of products purchased (packages or items), by which the number of products in the purchase basket was recorded. Participants were allowed to return any selected products. p <0.10+, p <0.05*, p <0.01**, p <0.001*** Please click here to view a larger version of this figure.
In addition to product choice behaviors, the virtual store can also record time and movement-related behaviors, such as, the shopping time and the walking distance. Figure 9 and Figure 10 show the effects of shelf attributes on the shopping time and walking distance of participants, respectively.
Figure 9: Total shopping time (s) participants spent in the supermarket with different shelf lengths and shelf orientations. The total shopping time accounts for the time participants spent between entering the store and leaving the store. The data management program also allows researchers to filter out the time that participants spent in a specific area. p <0.10+, p <0.05*, p <0.01**, p <0.001*** Please click here to view a larger version of this figure.
Figure 10: The walking distance of participants in the supermarket with different shelf lengths and shelf orientations. The walking distance was determined by multiplying the moving time (s) with the average shopping speed (m/s). The duration of the moving time used to calculate walking distance differs from the total shopping time because the moving time is exclusively recorded during participant movement. In contrast, the total shopping time accounts for the movement time and the time spent viewing and selecting products. Thus, the total moving time can be attained by only selecting the time during which participants move faster than 0.100 m/s. p <0.10+, p <0.05*, p <0.01**, p <0.001*** Please click here to view a larger version of this figure.
In addition to the effects of shelf attributes, the current research also focuses on shopping motivations to understand their influence on in-store shopping behavior. The results reveal significant main effects of shopping motivations on all in-store behavioral variables. Consumers with a hedonic motivation searched for (i.e., clicked on) (Mhedonic ± SDhedonic = 17.97 ± 6.93) and purchased more products (Mhedonic ± SDhedonic = 12.25 ± 2.42) than consumers with a utilitarian motivation (products examined: Mutilitarian ± SDutilitarian = 15.10 ± 4.82, products purchased: Mutilitarian ± SDutilitarian = 11.69 ± 2.43, see Figure 11). They also spent more time (Mhedonic ± SDhedonic = 607.18 ± 205.07 s, Mutilitarian ± SDutilitarian = 480.94 ± 134.25 s, see Figure 12) and walked longer distances (Mhedonic ± SDhedonic = 89.87 ± 31.15 m, Mutilitarian ± SDutilitarian = 80.73 ± 34.08 m, see Figure 13). The interaction effect of shopping motivation and store shelf attributes was not significant.
Figure 11: The total number of products examined (left) and the total number of products purchased (right) by participants with utilitarian and hedonic shopping motivation. The number of products examined and purchased are presented across all store layouts. Participants were assigned to shops under either utilitarian or hedonic shopping motivation prior to a shopping task. The shopping motivation was manipulated by a motivation manipulation task and a shopping situation. p <0.10+, p <0.05*, p <0.01**, p <0.001*** Please click here to view a larger version of this figure.
Figure 12: Total shopping time (s) spent in the supermarkets by participants with utilitarian or hedonic shopping motivation. The total shopping time accounts for the entire time that participants with different shopping motivations spent in the virtual supermarket across all store layouts. p <0.10+, p <0.05*, p <0.01**, p <0.001*** Please click here to view a larger version of this figure.
Figure 13: The distance that participants with utilitarian and hedonic shopping motivation walked. This figure shows the average walking distance across all store layouts. p <0.10+, p <0.05*, p <0.01**, p <0.001*** Please click here to view a larger version of this figure.
Virtual Supermarket Software | GreenDino BV | http://www.greendino.nl/virtual-labs.html | This software consists of editor, product library and consumer interface. |
Data Management Software: Observer XT | Noldus Information Technology | http://www.noldus.com/human-behavior-research/products/the-observer-xt | This software records observational data and facilitates the exportation of researcher-specified data sets using filters |
3D SpaceNavigator | 3Dconnexion | http://www.3dconnexion.eu/index.php?id=26&redirect2=www.3dconnexion.eu | A 3D SpaceNavigator allows participants to walk and make turns in the virtual store. In addition, it can be used by participants to adjust their eye-level during a shopping trip. |
3D moddeling software (e.g. Blender or 3DS Max) | Blender Foundation / Autodesk | https://www.blender.org/ http://www.autodesk.nl/products/3ds-max/overview | In case 3D models need to be made or adjusted 3D modeling software is needed. Many objects can be found online under different licencing agreements. |
Contract Reseach | Wageningen Univeristy and Research | http://www.wur.nl/en/Expertise-Services/Research-Institutes/Economic-Research.htm | The socio-economic research institute (Wageningen Economic Research) with experience in conducting the consumer research with the virtual store. |
People's responses to products and/or choice environments are crucial to understanding in-store consumer behaviors. Currently, there are various approaches (e.g., surveys or laboratory settings) to study in-store behaviors, but the external validity of these is limited by their poor capability to resemble realistic choice environments. In addition, building a real store to meet experimental conditions while controlling for undesirable effects is costly and highly difficult. A virtual store developed by virtual reality techniques potentially transcends these limitations by offering the simulation of a 3D virtual store environment in a realistic, flexible, and cost-efficient way. In particular, a virtual store interactively allows consumers (participants) to experience and interact with objects in a tightly controlled yet realistic setting. This paper presents the key elements of using a desktop virtual store to study in-store consumer behavior. Descriptions of the protocol steps to: 1) build the experimental store, 2) prepare the data management program, 3) run the virtual store experiment, and 4) organize and export data from the data management program are presented. The virtual store enables participants to navigate through the store, choose a product from alternatives, and select or return products. Moreover, consumer-related shopping behaviors (e.g., shopping time, walking speed, and number and type of products examined and bought) can also be collected. The protocol is illustrated with an example of a store layout experiment showing that shelf length and shelf orientation influence shopping- and movement-related behaviors. This demonstrates that the use of a virtual store facilitates the study of consumer responses. The virtual store can be especially helpful when examining factors that are costly or difficult to change in real life (e.g., overall store layout), products that are not presently available in the market, and routinized behaviors in familiar environments.
People's responses to products and/or choice environments are crucial to understanding in-store consumer behaviors. Currently, there are various approaches (e.g., surveys or laboratory settings) to study in-store behaviors, but the external validity of these is limited by their poor capability to resemble realistic choice environments. In addition, building a real store to meet experimental conditions while controlling for undesirable effects is costly and highly difficult. A virtual store developed by virtual reality techniques potentially transcends these limitations by offering the simulation of a 3D virtual store environment in a realistic, flexible, and cost-efficient way. In particular, a virtual store interactively allows consumers (participants) to experience and interact with objects in a tightly controlled yet realistic setting. This paper presents the key elements of using a desktop virtual store to study in-store consumer behavior. Descriptions of the protocol steps to: 1) build the experimental store, 2) prepare the data management program, 3) run the virtual store experiment, and 4) organize and export data from the data management program are presented. The virtual store enables participants to navigate through the store, choose a product from alternatives, and select or return products. Moreover, consumer-related shopping behaviors (e.g., shopping time, walking speed, and number and type of products examined and bought) can also be collected. The protocol is illustrated with an example of a store layout experiment showing that shelf length and shelf orientation influence shopping- and movement-related behaviors. This demonstrates that the use of a virtual store facilitates the study of consumer responses. The virtual store can be especially helpful when examining factors that are costly or difficult to change in real life (e.g., overall store layout), products that are not presently available in the market, and routinized behaviors in familiar environments.
People's responses to products and/or choice environments are crucial to understanding in-store consumer behaviors. Currently, there are various approaches (e.g., surveys or laboratory settings) to study in-store behaviors, but the external validity of these is limited by their poor capability to resemble realistic choice environments. In addition, building a real store to meet experimental conditions while controlling for undesirable effects is costly and highly difficult. A virtual store developed by virtual reality techniques potentially transcends these limitations by offering the simulation of a 3D virtual store environment in a realistic, flexible, and cost-efficient way. In particular, a virtual store interactively allows consumers (participants) to experience and interact with objects in a tightly controlled yet realistic setting. This paper presents the key elements of using a desktop virtual store to study in-store consumer behavior. Descriptions of the protocol steps to: 1) build the experimental store, 2) prepare the data management program, 3) run the virtual store experiment, and 4) organize and export data from the data management program are presented. The virtual store enables participants to navigate through the store, choose a product from alternatives, and select or return products. Moreover, consumer-related shopping behaviors (e.g., shopping time, walking speed, and number and type of products examined and bought) can also be collected. The protocol is illustrated with an example of a store layout experiment showing that shelf length and shelf orientation influence shopping- and movement-related behaviors. This demonstrates that the use of a virtual store facilitates the study of consumer responses. The virtual store can be especially helpful when examining factors that are costly or difficult to change in real life (e.g., overall store layout), products that are not presently available in the market, and routinized behaviors in familiar environments.