Fig 1.
Visualization of the hypotheses and observed opinion shifts during Study 1.
The horizontal axis of Fig 1 shows initial opinion distance, which—following specifications of positive and negative influence in recent formal modeling work—is the core independent variable of our analyses. The vertical axis charts the main dependent variable, the direction and magnitude of opinion change towards the source after the first stimulus. Positive values indicate that the participant’s opinion shifted towards the source. Negative values represent an opinion shift away from the source. The magnitude measures in both cases how much the initial opinion dissimilarity was reduced or increased by the opinion shift in absolute terms. The two grey solid lines define the maximal positive and negative opinion shift that is possible given the initial opinion distance.
Table 1.
Fixed Effects Estimates (Top) and Variance-Covariance Estimates (Bottom) for Models of the Predictors of Opinion Shifts after the First Stimulus in Study 1.
Fig 2.
Relationship between initial opinion distance and attraction measured after the first stimulus in Study 1.
Table 2.
Fixed Effects Estimates (Top) and Variance-Covariance Estimates (Bottom) for Models of the Predictors of Attraction Ratings after the First Stimulus in Study 1.
Table 3.
Relative Frequency of Positive, Negative, and No Opinion Shifts Observed in Study 2 after the First Stimulus and after All Stimuli.
Table 4.
Fixed Effects Estimates (Top) and Variance-Covariance Estimates (Bottom) for Models of the Predictors of the First Opinion Shift in Study 2.
Fig 3.
Visualization of observed opinion shifts in the control treatment and the disliking treatment of Study 2.
In the right panel, the dashed line shows model prediction for the average initial attraction.
Fig 4.
Average opinion shift after the first stimulus in the disliking treatment of Study 2, broken down by liking vs. disliking and small (below the mean of all) vs. large (above the mean of all) initial distances.
The number of cases is displayed on top of the bars. Error bars show 95% confidence intervals.
Table 5.
Fixed Effects Estimates (Top) and Variance-Covariance Estimates (Bottom) for Models of the Predictors of Attraction Ratings after the First Stimulus in Study 2.