Fig 1.
Generalized overview of survey design.
Fig 2.
Demographics of survey participants (N = 1,444).
(A) Main scientific disciplines. Note that disciplines correspond to bioRxiv submission categories, and participants could select multiple disciplines. (B) Country of residence. The top 30 countries by total participants are shown. (C) Gender. (D) Institution type. (E) Career status.
Fig 3.
Publishing behaviour of survey participants over the past 5 years.
(A) Number of articles published by survey participants in peer-reviewed scientific journals in the past 5 years as a corresponding author. (B) Number of articles published by survey participants in peer-reviewed scientific journals in the past 5 years as a co-author. (C) Proportion of articles in (A) that were posted as preprints. (D) Proportion of articles in (B) that were posted as preprints. Authors who reported that they published 0 articles in the past 5 years as a corresponding author or as a co-author were not included in (C) or (D), respectively.
Fig 4.
Decision-making for posted preprints.
Results show answers to survey questions on a 5-point Likert scale, divided by articles that authors published as a corresponding author (left) and articles that authors published as a co-author (right).
Fig 5.
Motivations for posting preprints. Results show answers to survey questions on a 5-point Likert scale, divided by articles that authors published as a corresponding author (left) and articles that authors published as a co-author (right).
Fig 6.
Benefits of posting preprints on article impact.
Results show answers to survey questions on a 5-point Likert scale, divided by articles that authors published as a corresponding author (left) and articles that authors published as a co-author (right).
Table 1.
Results of ordinal logistic regression for survey questions in Fig 4 (corresponding authorships only; N = 1146).
Values are presented as exponentiated odds-ratios, with the range in parentheses representing 95% confidence intervals. Significant results at the 95% level are indicated in bold and with “*”.
Table 2.
Results of ordinal logistic regression for survey questions in Fig 5 (corresponding authorships only; N = 1146).
Values are presented as exponentiated odds-ratios, with the range in parentheses representing 95% confidence intervals. Significant results at the 95% level are indicated in bold and with “*”.
Table 3.
Results of ordinal logistic regression for survey questions in Fig 6 (corresponding authorships only; N = 1146).
Values are presented as exponentiated odds-ratios, with the range in parentheses representing 95% confidence intervals. Significant results at the 95% level are indicated in bold and with “*”.
Table 4.
Classification of free-text responses, in response to the question “Were there any additional reasons that motivated you to deposit these articles as preprints?”.
Categories are ordered alphabetically (except for “Other”). Percentages for corresponding and co-authorships refer to the proportion of all responses containing the relevant category; responses could contain multiple categories (N = 253 comments on corresponding authorships, 38 comments on co-authorships).
Fig 7.
Reasons that survey participants did not post articles as preprints.
Results show answers to survey questions on a 5-point Likert scale, divided by articles that authors published as a corresponding author (left) and articles that authors published as a co-author (right).
Table 5.
Results of ordinal logistic regression for survey questions in Fig 7 (corresponding authorships only; N = 1021).
Values are presented as exponentiated odds-ratios, with the range in parentheses representing 95% confidence intervals. Significant results at the 95% level are indicated in bold and with “*”.
Table 6.
Classification of free-text responses, in response to the question “Were there any additional reasons that caused you to not deposit these articles as preprints?”.
Categories are ordered alphabetically (except for “Other”). Percentages for corresponding and co-authorships refer to the proportion of all responses containing the relevant category; responses could contain multiple categories (N = 180 comments on corresponding authorships, 137 comments on co-authorships).
Fig 8.
Differences between articles posted as preprints versus those not posted as preprints.
Results show answers to survey questions on a 5-point Likert scale, divided by articles that authors published as a corresponding author (left) and articles that authors published as a co-author (right).
Table 7.
Results of ordinal logistic regression for survey questions in Fig 8 (corresponding authorships only; N = 953).
Values are presented as exponentiated odds-ratios, with the range in parentheses representing 95% confidence intervals. Significant results at the 95% level are indicated in bold and with “*”.
Table 8.
Classification of free-text responses, in response to the question “Were there any other differences between the articles you chose to deposit as preprints versus those you did not deposit, or differences in your expectations for how the articles would be received in your community?”.
Categories are ordered alphabetically (except for “Other”). Percentages for corresponding and co-authorships refer to the proportion of all responses containing the relevant category; responses could contain multiple categories (N = 394 comments on corresponding authorships, 93 comments on co-authorships).