Analysis of Eye Tracking Experiment

May 26, 2009

Analysis of Eye Tracking Experiment

Part 1 - Dishwasher Detergents

1.1  Analysis Setup

For each consumer, a “fixations-only” video was generated containing one frame per fixation. These videos were manually coded for fixations on the following volumes-of-interest (VOIs), with their abbreviations provided in parentheses (see Figure 1):

·  Cascade Liquid (CasLiq)

·  Cascade Powder (CasPow)

·  Cascade Tablets (CasTab)

·  Electrasol Liquid (ElcLiq)

·  Electrasol Powder (ElcPow)

·  Electrasol Tablets (ElcTab)

·  Palmolive Liquid (PalLiq)

·  Other Liquid (OthLiq)

·  Other Powder (OthPow)

·  Other Tables (OthTab)

·  Other (Other) – anything not labeled in Figure 1 (not dishwasher detergents).

Figure 1. VOIs for dishwasher detergent task. This photo shows new Cascade packaging. VOIs were grouped the same when the old packaging was displayed.

1.2  Results

The following figures show the results for the dishwasher detergent task. Consumers were grouped into the two blocks for these results. These blocks consist of (1) consumers who shopped while the new Cascade packaging was displayed (New); and (2) consumers who shopped while the old Cascade packaging was displayed (Old). For the New block, data from nine consumers were used. For the Old block, data from eight consumers were used. Data from other consumers were thrown out due to the videos cutting out or bad calibrations.

Figure 2. Percentage of consumers who fixated each VOI when shopping for dishwasher detergent.

Figure 2 shows the percentage of consumers, from each block, who fixated each VOI. Blue bars represent consumers who shopped while new Cascade packaging was being displayed on the shelves and red bars represent consumers who shopped while old Cascade packaging was displayed on the shelves. For example, about 90% of consumers who shopped while the new packaging was being displayed fixated Cascade liquid products and 100% of the consumers who shopped while the old packaging was being displayed fixated these products.

Figure 3. The fraction of consumers in each block who fixated Cascade Liquid products first (1), second (2), etc.

Figure 3 shows the order in which Cascade Liquid products were fixated. This plot was produced by ranking all VOIs fixated by a given consumer in the order in which they were fixated and then counting the occurrences of each “order” across consumers within each block and dividing by the total number of consumers in each block. In other words, 2 out of 9 consumers (22%) who shopped while the new Cascade packaging was being displayed fixated a Cascade Liquid product first and 5 out of 6 consumers (83%) who shopped while the old packaging was being displayed fixated a Cascade liquid product before any other dishwasher detergent product. Fractions of consumers who fixated in each order are shown because a different number of consumers were used for each block. Data from 2 consumers who shopped while old Cascade packaging was displayed are not shown in Figures 3 to 5 because these consumers spent more than 30 seconds looking at dishwasher detergents and the Fixations-only videos were prepared to start 30 seconds prior to final product selection. Note, the fractions of consumers who fixated a given VOI (Cascade Liquid, Powder, or Tablets) may not add to 1 if not all consumers fixated that VOI (e.g., only 8 of the 9 consumers in the New block fixated a Cascade Liquid product so the fractions of consumers who fixated this product 1st, 2nd, 3rd, 4th and 5th only add to 8/9 or 0.89. The horizontal axis goes to 10 because for all consumers, the largest number of different VOIs fixated was 10. Figures 4 and 5 show this same information for Cascade Powder and Tablets, respectively.

Figure 4. The fraction of consumers in each block who fixated Cascade Powder products first (1), second (2), etc.

Figure 5. The fraction of consumers in each block who fixated Cascade Tablets products first (1), second (2), etc.

Figure 6. Mean number of fixations on each VOI.

Figure 6 shows the mean number of fixations on each VOI with error bars showing the standard error. For example, consumers made an average (across consumers, within the Old packaging block) of about 20 fixations on Cascade Liquid products; an average of about seven fixations were made on Cascade Liquid products by consumers in the New packaging block.

Figure 7. Mean total gaze duration on each VOI.

Figure 7 shows the mean total gaze duration on each VOI with error bars showing the standard error. The total time spent fixating each VOI was computed for each consumer and the mean across consumers within each block is shown in this figure. For example, consumers on average spent 4.8 seconds looking at Cascade Liquid products when shopping with old packaging being displayed and an average of about 1.4 seconds looking at Cascade Liquid products when shopping with new packaging being displayed.

The Excel workbook (DD_Analysis.xls) used to generate these results is attached.