Fig 1.
The effect of celebrity on survey respondents’ willingness-to-engage with the advertisement shown to them.
Fig 2.
The effects of the celebrity featured and the respondents’ beliefs about them, reasons for clicking, beliefs about the celebrities’ reasons for appearing and their own demographic characteristics on their willingness-to-engage with a single, randomly selected advertisement.
Positive values indicate that the variable increases WTE, while negative values indicate that it decreases WTE. Mean estimates (points) are accompanied by 95% credible intervals, calculated as highest posterior density intervals (lines). Grey shading identifies groups of conceptually-related predictor variables. Within these groups predictors are presented in order from most positive to most negative effect to facilitate visual comparisons.
Fig 3.
The effect of celebrity on survey respondents’ ability to recall campaign message of the advertisement shown to them.
Fig 4.
The effects of advertisement-specific characteristics (Fig 4A).
The celebrity featured and individual-specific respondent characteristics (Fig 4B) on the advertisement chosen when all four were presented alongside one another. The parameter estimates are presented on the scale of the linear predictor. Positive values indicate that the variable increases the probability that the celebrity is chosen relative to the baseline category (Crawford Allan), while negative values indicate that it decreases the probability that he is chosen. Mean estimates (points) are accompanied by 95% credible intervals, calculated as highest posterior density intervals (lines). Grey shading is used to highlight groups of conceptually-related predictor variables. Within these groups, the sets of parameters associated with each predictor variable are ordered by the set mean from most positive to most negative effect to facilitate visual comparisons.