The reason this sometimes occurs is due to the way EmotionAll® Scores are calculated. The EmotionAll® Scores are based on the peak (or maximum) values of particular emotions:
- Attraction: Measures the PEAK in surprise within the first 8 seconds of viewing.
- Retention: Measures the PEAK of happiness after the first 8 seconds.
- Engagement: Measures the PEAK level of engagement reached throughout the video.
- Impact: Measures the average of the PEAK happiness value and the end happiness value
When looking at the emotions results segmented by demographic groups, sometimes those peak moments occur at different seconds for the different groups. This means that when the data is then aggregated together across all demographic groups, those peak percentages in emotion that occurred within each of the demographic splits are reduced now that the percentages are being calculated out of the larger total data set.
For example, in the image below you can see females scored an EmotionAll® of 8 (within the 85th percentile) while males scored an EmotionAll® of 9 (within the 90th percentile). When looking at the Happy traces, you can see that the peak in Happiness occurs at about the 20th second among females with 18.9% of the 137 women that were included in the study showing happiness at that second. On the other hand, the max point in Happiness occurs at about the 27th second for males with 21.9% of the 95 men that viewed this video showing happiness:
Now, looking at the overall EmotionAll® Score across all genders, you can see this video scored below each of the individual demographic groups with an EmotionAll® of 8 (within the 80th percentile). The peak in Happiness occurs at the 27th second with 16.8% of the total 232 viewers showing happiness at that moment:
In summary, the segmented peak percentages in Happiness (18.9% for females and 21.9% for males), were greater than the peak percentage in Happiness (16.8%) when the genders were looked at in combination. This same incident can occur across some or all of the emotions that go into calculating the EmotionAll® Scores, resulting in the overall score being lower than those of the individual demographic groups.