Effects of nostalgic messages on ad persuasiveness: a meta-analysis

Abstract

This meta-analysis examined the effectiveness of nostalgic messages on ad persuasiveness. Based on 23 articles (31 studies, 39 pairs of comparisons, N = 5814), this study found that nostalgic messages (compared to non-nostalgic messages) have a positive, small effect on persuasion (r = .19, SE = .02, 95% CI [.15, .24], p < .001). Additionally, the effect of nostalgic messages tended to be stronger if (1) the advertised product was high in hedonic values and (2) a link between the brand/product and personal memories was present. Implications, future directions and limitations are also discussed.

Keywords:

• Nostalgia
• meta-analysis
• persuasion
• advertising

Disclosure statement

The authors declare that there is no conflict of interest.

Funding

This study was supported by the 2019 Summer Research Fellowship at the California State University, San Bernardino.

Notes on Contributors

Ying Cheng (Ph.D., Michigan State University) is an Assistant Professor in the Department of Management at the California State University, San Bernardino. Her research areas focus on persuasion and message features in the contexts of health, advertising, and crowdfunding. Additionally, she also studies the interplay between technologies and health, including how technologies can impact and promote people’s health behaviors and well-being. Her work has been published in journals such as Journal of Health Psychology, Health Communication, Journal of Health Communication, American Behavioral Scientist, and Computers in Human Behavior.  

Xiaodi Yan (Ph.D., Michigan State University) is an Assistant Professor in the Guangming School of Journalism and Communication at China University of Political Science and Law. She studies message effects in various contexts including intergroup relations, stigma, health and risks. She is also interested in the role of emotion and culture in these processes.

Data availability statement

The data of this meta-analysis were provided in the Tables and the supplementary materials.

Notes

1 When the information about the sample size assigned to the experiment vs. control condition was unavailable, the researchers first contacted the first author to solicit the information; if no response was received, this study assumed that participants were equally distributed across different conditions.

2 Following the procedure delineated by Borenstein et al. (2009, 225–232), a combined effect size was computed as the mean of each dependent outcome reported in a study:

$\overline{Y}=\frac{1}{m}{\sum }_{j=1}^m{Y}_j$

where

j = the j^th outcome in a study

m = the number of outcomes reported in a study

Y = the effect size.

Additionally, the variance of the mean of multiple outcomes in a study was computed as:

$\mathit{var}\left(\frac{1}{m}{\sum }_{j=1}^m{Y}_j\right)={\left(\frac{1}{m}\right)}^2\mathrm{ }({\sum }_{j=1}^m{V}_j+\sum _{i\ne j}\left(r_{ij}\sqrt{V_i}\mathrm{ }\sqrt{V_j}\right))$

where

i, j = the i^th, j^th outcome in a study

m = the number of outcomes reported in a study

V = the variance

r = the correlation between $Y_i$and $Y_j$

Notably, because most studies included in this meta-analysis did not report the correlation between different persuasion outcomes, we set the correlation as 1. This assumption can be justified for two reasons. First, studies have report strong correlations between attitude toward the ad, attitude toward the brand, and purchase intention (Spears and Singh 2004). Second, setting correlations between multiple outcomes as 1 allows us to assess the overall effect size in a more conservative way as this approach can overestimate the variance but underestimate the precision.

3 Three studies targeted promoting charity organizations; as such they cannot be coded on product type.

4 References marked with an asterisk (*) indicate studies included in the current meta-analysis.