The authors regret to report that we found several errors in the values presented in the text. The first seven of the errors listed below arose from the failure to reflect the minor corrections discovered during the revision process, and the last two were a simple calculation error. We want to assure readers that all figures in the text have been thoroughly reviewed and correctly adjusted below, rendering them unaffected by the errors reported.
• In the second paragraph of the “Social desirability bias and survey mode effects” section, “0.391 [0.291, 0.491]” should read “0.395 [0.302, 0.488].”
• In the next paragraph, “0.078 [−0.070, 0.226]” and “p = 0.301” should read “0.083 [−0.055, 0.221]” and “p = 0.241,” respectively.
• In the last paragraph of the “Heterogeneity in susceptibility to social desirability bias” section, “0.475 [0.193, 0.756]” should read “0.450 [0.178, 0.721].” Moreover, “p < 0.001 in both surveys” should read “p = 0.001 or less in both surveys.”
• In the fourth paragraph of the “Exploratory subgroup analyses” section, “about 27 and 24 percentage points, respectively” should read “about 32 and 22 percentage points, respectively.”
• In the last paragraph of the same section, “about 40 percentage points” should read “about 43 percentage points.”
• In the last paragraph of the “Sensitivity analysis for non-strategic misreporting” section, “0.436 [0.261, 0.610], 0.404 [0.214, 0.593], 0.336 [0.116, 0.557], and 0.242 [−0.023, 0.507]” should read “0.436 [0.262, 0.610], 0.404 [0.215, 0.593], 0.336 [0.116, 0.557], and 0.242 [−0.022, 0.506].”
• In the seventh paragraph of the “Discussion” section, “In the face-to-face survey, the magnitude of SDB was 47.5 percentage points” should read “In the face-to-face survey, the difference in the magnitude of SDB was 45.0 percentage points.”
• In the tenth paragraph of the same section, “82 percent” should read “83 percent.”
• In Note 27, “3.0%” should read “3.1%,” and “22.1%” should read “22.2%.”
We also found a duplication of information in Note 11, which mirrors the content present in the main text. The correct Note is as follows: “Although we could not implement blocked randomization based on age and education in the face-to-face survey due to practical limitations, because blocked randomization only affects the estimation efficiency, the lack of blocking in the face-to-face survey should not distort the comparison of survey modes.”
Furthermore, we found minor errors in our program code, which slightly affected Figure A.6 in Online Appendix B.3. We uploaded the appendix file with a corrected figure to a repository where we published the replication materials (https://doi.org/10.7910/DVN/RQWIZL).
Once again, these errors do not affect the conclusions of the article. We apologize for these errors.
