Figure 8
From: Misinformation does not reduce trust in accurate search results, but warning banners may backfire
The presence of misinformation (red bars and points) had no effect on trust (Y-axis) in all of the surrounding accurate results on the page compared to when none of the results on the page contained misinformation (blue bars and points). This null effect was stable across experiments (rows) and for both clicked (left column) and non-clicked (right column) results. The Y-axis displays the mean trustworthiness ratings given to accurate results across all ranks on the page by participants who saw a misinformation result on the page (the red bars and points on the right side of each panel) compared to those who saw no misinformation result anywhere on the page (the blue bars and points on the left). Each bar in the figure represents trustworthiness ratings from 96 to 1473 participants, depending on the experiment and click category displayed. For example, the red bar in the right-hand panel in the “Experiment 1” row shows that when participants evaluated an accurate result they had not clicked after seeing misinformation in another result, they gave it an average trustworthiness rating of about 3.1 scale points, and the average for these results when there was no misinformation present (blue bar) was similar at about 3.1 scale points. For the data from all experiments (bottom row), ratings are mean centered within each experiment and the respective click category because the experiments’ overall mean trustworthiness ratings differed for both clicked and non-clicked results, so we needed to standardize them within each experiment before aggregating them. Error ranges represent 95% Confidence Intervals of the means.
