Go Negative for Clicks: Negative Sentiment in Environmental Advocacy Emails Is Associated with Increased Public Engagement

ABSTRACT

In light of environmental challenges, environmental advocacy organizations are seeking optimal ways to increase public engagement. There is debate over whether positive or negative sentiment in advocacy messages is more effective. We investigate this by examining how positive and negative sentiment in advocacy messages sent by the Environmental Defense Fund (EDF) through email was associated with the frequency with which recipients opened the email and clicked on a link within the email. We used a computer-assisted content analysis to determine the sentiment of email campaign messages sent by the EDF between November 2011 and February 2020, and then examined how public engagement varied depending on the sentiment of the email messages. Overall, we find that negative sentiment in environmental advocacy emails was associated with increased engagement.

KEYWORDS:

• Environmental advocacy
• sentiment
• public engagement
• online communication
• computer-assisted content analysis

Acknowledgements

We thank the Environmental Defense Fund for providing the campaign texts and engagement data used in the analysis for this study. Any opinions, findings, and conclusions or recommendations expressed here are those of the authors and do not necessarily reflect the views of the Environmental Defense Fund.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Some campaigns are sent out multiple times and some email addresses will have received multiple emails from the EDF.

2 Please refer to Young and Soroka (2012) for details of the positive and negative words included in the LSD. Broadly, the LSD includes a combination of what Soroka (2014) refers to as qualitative and quantitative components of sentiment; e.g. for negativity, it captures both affective components such as “fear” and “anger” and objectively negative quantities such as “endangered” species. For further conceptualization of sentiment in political texts, see e.g. Lengauer et al. (2012).

3 These results also hold when either negative or positive words are excluded from the models.