Everyone wants to create engaging content and learn if people are engaged. But how is engagement measured accurately? Nicky Keatch’s research project, part of his academic focus on social psychology, aimed to explore the intricate nature of engagement. This study sought to better understand how people emotionally, cognitively, and behaviorally interact with movie trailers. Importantly, it includes both self-reported and implicit measures. Read more about his findings in this blog.Â
What is Engagement?
Engagement plays a vital role in how individuals interact with different types of content, whether it’s advertisements, movie trailers, or educational videos. Most video marketers find engagement the most important metric. In this study, we utilized a model based on Ben-Eliyahu’s (2018) framework and the work of a previous Noldus intern (Dayenne Sarkol), which divides engagement into three main dimensions:
- Cognitive Engagement: relating to how well viewers concentrate on the trailers and maintain their attention.
- Affective Engagement: relating to the emotional responses the content elicits. In our context this means the level of interest, positive attitude, involvement and motivation.
- Behavioral Engagement: relating to the actions and decisions viewers take as a result of their engagement. This includes recommendations, brand related actions (e.g. re-posting the content on social media) and interacting with the content/product.
The previous internship found that a real-time custom expression of engagement was too complex. Therefore, in this project we focus on the separate components. The three dimensions are explored in relation to facial expression responses towards trailers. In order to investigate this an implicit measure of engagement was created by splitting the trailers and asking the audience if they would like to continue or stop watching a specific comedy trailer.
How the Study Was Conducted
The study involved 75 participants, recruited via Prolific and personal invitation of the researchers. Participants watched three movie trailers from different genres: My Spy: The Eternal City (action-comedy), Fly Me to the Moon (romantic-drama), and Paddington in Peru (family-comedy). Each trailer was divided into two parts. Participants had the option to either stay watching or switch to another trailer after the first part. Their decisions provide an implicit behavioral proxy for engagement. Moreover, FaceReader Online measured implicit cognitive and affective engagement through facial expressions like happiness, attention, and sadness. Meanwhile, self-report of each of the engagement components measured engagement explicitly. This complementary approach of combining implicit and explicit measures of engagement aimed to create a more comprehensive and accurate picture of engagement.
Self-reported engagement is higher when people keep watching
As expected, self-reported data on engagement revealed higher levels of engagement for participants who decided to continue watching rather than switch trailers (switch decisions were roughly around 50%). Specifically, behavior and affective factors were significantly higher in those who continued watching the trailer. However, the cognitive component (e.g. concentration) was only marginally significant.
It was expected that this pattern would be similar across trailers, however, this was not the case.
- People who choose to watch the next part of the Paddington trailer scored higher on all engagement components.
- Those choosing to keep watching the Fly Me to the Moon trailer only scored higher on affective engagement.
- There was no difference in explicit engagement between the group that switched and stayed in the My Spy trailer. Thus not giving a clear indication why people choose not to watch the second part.
Facial expressions show converging and diverging patterns
Facial expressions are implicit ways of measuring reactions toward stimuli, since they occur automatically and often uncontrolled. FaceReader Online gives an analysis of facial expressions, such as happiness, which is produced by smiling and laughing (important for affective engagement). In addition, it also gives an indication of a more cognitive component: by estimating the visible attention given to the screen.
In the implicit response different results for the trailers also arose.
- An increase in cognitive engagement was shown by higher mean attention only in the Paddington trailer. People deciding to watch the second part score on average 15% higher on Attention.
- Converging with the explicit response, the clearest affective engagement difference is in the Fly me to the Moon trailer: there is a 3 times higher mean happiness in the group that continued to watch.
- Also similar as in the explicit response, there was no difference in mean expressions between the My Spy groups. But the My Spy trailer shows an additional relevant emotional dynamic that may clarify their decision (see below). There is a joke that people appear to find funny (compared to those who stay), but does not lead them to watch the second part of the trailer.
The advantage of analyzing facial expressions is this ability to capture immediate, real-time responses. Interestingly, all trailers show larger/more happiness peaks when people continue to watch the trailer.
How to benefit from implicit and explicit engagement measurement
These findings align with the notion that engagement is a multidimensional construct. It also underscores the difficulty in measuring engagement and the importance of context, since each trailer seemed to elicit engagement differently. These findings have implications not only for theoretical models of engagement but also for practical applications in content creation and marketing strategies.
The use of implicit and explicit responses ensures a more accurate and nuanced understanding of engagement. This ultimately enables more effective prediction of viewer behavior and improving content creation strategies in fields like entertainment, advertising, and education. This blog shows good examples of how to create engaging videos, and by using FaceReader Online the effectiveness can then be tested with implicit and explicit measurements.