Dataset Information

Message Design Choices Don't Make Much Difference to Persuasiveness and Can't Be Counted On-Not Even When Moderating Conditions Are Specified.

ABSTRACT: Persuaders face many message design choices: narrative or non-narrative format, gain-framed or loss-framed appeals, one-sided or two-sided messages, and so on. But a review of 1,149 studies of 30 such message variations reveals that, although there are statistically significant differences in persuasiveness between message forms, it doesn't make much difference to persuasiveness which option is chosen (as evidenced by small mean effect sizes, that is, small differences in persuasiveness: median mean rs of about 0.10); moreover, choosing the on-average-more-effective option does not consistently confer a persuasive advantage (as evidenced by 95% prediction intervals that include both positive and negative values). Strikingly, these results obtain even when multiple moderating conditions are specified. Implications for persuasive message research and practice are discussed.

SUBMITTER: O'Keefe DJ

PROVIDER: S-EPMC8275937 | biostudies-literature | 2021

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Message Design Choices Don't Make Much Difference to Persuasiveness and Can't Be Counted On-Not Even When Moderating Conditions Are Specified.

O'Keefe Daniel J DJ Hoeken Hans H

Frontiers in psychology 20210629

Persuaders face many message design choices: narrative or non-narrative format, gain-framed or loss-framed appeals, one-sided or two-sided messages, and so on. But a review of 1,149 studies of 30 such message variations reveals that, although there are statistically significant differences in persuasiveness between message forms, it doesn't make much difference to persuasiveness which option is chosen (as evidenced by small mean effect sizes, that is, small differences in persuasiveness: median ...[more]

PMID: 34267703

Similar Datasets

Project description:BACKGROUND:Although evaluation studies confirm the strong potential of men's electronic health (eHealth) programs, there have been calls to more fully understand acceptability, engagement, and behavior change to guide future work. Relatedly, mapping of behavior changes using health promotion theories including the transtheoretical model (or stages of change) has been recommended to build a translatable empirical base to advance design and evaluation considerations for men's eHealth programs. OBJECTIVE:This study aimed to use a benchmark sample as a reference group to map the recent and intended health behavior changes in Canadian men who use the Don't Change Much (DCM) eHealth program. The hypothesis being tested was that increased exposure to DCM would be positively associated with men's recent and intended health behavior changes. METHODS:DCM users (n=863) were sampled for demographic data and self-reported recent and intended health behavior changes. Respondents also reported their usage (frequency and duration) for each of the 3 DCM components (web, newsletter, and social media) and were allocated to limited exposure (257/863, 29.8%), low exposure (431/863, 49.9%), and high exposure (175/863, 20.3%) subgroups. A benchmark sample (n=2000), comprising respondents who had not accessed DCM provided a reference group. Bivariate analysis of recent and intended health behavior changes and DCM exposure levels were used to compute the strength of association between the independent variables (exposure levels) and the 10 categorical dependent variables (recent and intended health behavior changes). Binary logistic regression models were computed for each of the 10 recent and intended health behavior changes. Linear regression was used to model the association between the number of recent and intended changes and the level of exposure to DCM. RESULTS:Compared with the benchmark reference group, DCM high-exposure respondents had significantly increased odds for 9 of the 10 health behavior changes, with the largest effect size observed for Changed diet or Improved eating habits (odds ratio [OR] 5.628, 95% CI 3.932-8.055). High-exposure respondents also had significantly increased odds for 9 intended health changes, with the largest effect sizes observed for Reduce stress level (OR 4.282, 95% CI 3.086-5.941). Moderate effect size (goodness of fit) was observed for increased total number of recent (F12,2850=25.52; P.001; adjusted R2=.093) and intended health behavior changes (F12,2850=36.30; P.001; adjusted R2=.129) among high-exposure respondents. CONCLUSIONS:DCM respondents contrasted the predominately precontemplative benchmark sample mapping across the contemplative, preparation, and action stages of the transtheoretical health behavior change model. Almost 10% of variation in the recent and 13% of variation in the intended health behavior changes can be explained by DCM exposure and demographic factors, indicating the acceptability of this men's eHealth resource.

Dataset Information

Message Design Choices Don't Make Much Difference to Persuasiveness and Can't Be Counted On-Not Even When Moderating Conditions Are Specified.

Publications

Message Design Choices Don't Make Much Difference to Persuasiveness and Can't Be Counted On-Not Even When Moderating Conditions Are Specified.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets