The E-Score summarizes the performance of a Stim into a few headline numbers.
To do this it’s important to compare the relative performance of the Stim to others we have tested, and consider how variable these metrics are. For this purpose, we use benchmark results to calibrate the E-Score.
The E-Score has a single “overall” value, and 3 components. The overall value is an average of the components. Each component has equal weighting but components have different distributions, meaning some are more variable than others. We normalize these components based on benchmark results from thousands of participants viewing hundreds of Stims. From time to time these benchmarks are updated (on request, we can provide historical E-Scores for any Stim for your reference).
The components are:
- Emotion – derived from peak engagement as a fraction of population
- Eyes – derived from the frequency of visual disengagement events
- Effects – derived from changes in implicit association
If you did not include Implicit association testing, the E-Score can be taken as the average of the Emotion and Attention components only.
The EmotionScore is the peak frequency of aggregate emotional reaction over all 6 AUs and over the whole session time. This is compared to our baselines for high quality and low quality media when measured with typical sample populations. The motivation for this is that all emotional reactions are better than none at all, the latter being far more forgettable and essentially having limited impact.
The EyesScore is a measure of aggregated distraction frequency – how often participants disengage and look away from the stimulus entirely. This happens naturally, regardless of content, but the rate increases when participants are bored or not emotionally engaged. Therefore, we score based on the rate of look-away events compared to baselines for stimulating and boring content and a library of historical examples. We found that distraction events are the best indicator of how visual attention relates to emotional engagement. Distraction events also peak at the least interesting moments in a video, which is why we also provide this metric as a vector over time.
The effects score can be measured explicitly or implicitly. Explicit measurement is via a survey question; implict via an implicit reaction test. The results of both are expressed as a percentage of the population, which we convert into a score.