Political Capital in the 21st Century: Presidential Grassroots Lobbying Organizations in the Obama Administration

Abstract

This article advances the study of presidential leadership by analyzing the circumstances under which Organizing for America/Organizing for Action (OFA) mobilized core supporters of President Barack Obama for policy purposes using digital tools and other resources developed for the president’s election campaign. I argue that the policy mobilization efforts of such presidential grassroots lobbying organizations (PGLOs) are constrained by electoral opportunity costs and innate factors which lead PGLOs to be most effective at maintaining the support of presidential copartisans in Congress. These constraints are used to develop several hypotheses about the timing and frequency of OFA grassroots lobbying mobilization efforts, which are tested using a dataset of 245 OFA mobilization requests to supporters sent via e-mail from January 2009 through January 2017. The resulting analysis finds support for several of these hypotheses, as well as the proposition that PGLO grassroots lobbying efforts carry electoral costs. Furthermore, it supports the claim that OFA leadership were most willing to pay the costs of lobbying mobilization when President Obama’s copartisans set the agenda in Congress under unified government and OFA mobilization of Democratic “super-activists” could have most easily kept presidential priorities under consideration. Rather than judging them on their ability to mobilize the mass public and coerce opposition legislators to side with the President on key votes, I argue that these findings support a reevaluation of OFA and similar groups' effectiveness on the basis of their ability to unify and motivate the president's co-partisan allies in Congress.

Acknowledgments

The financial support through Agence Nationale de la Recherche–Labex Institute for Advanced Study in Toulouse is gratefully acknowledged. Helpful comments from Dan Alexander, David Primo, Lawrence Rothenberg, Karine van der Straeten, and Hye Young You, as well as the participants in the seminar series at the Center for the Study of Democratic Institutions at Vanderbilt University are gratefully acknowledged.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1 “Obama’s Army of E-Mail Backers Gives Him Clout to Sway Congress,” Bloomberg, December 1, 2008. http://www.bloomberg.com/apps/news?pid=newsarchive/&sid=aEVXKOC3s8.k

2 E-mails were obtained by creating an account on www.barackobama.com. There were no known terms of service for subscription to the OFA e-mail list. OFA blog data were obtained through the Internet Archive, a third-party website (Organizing for America 2009).

3 The Organizing for America e-mail list was transferred to the Obama re-election organization in April 2011, and in January 2013, the list was transferred from the Obama campaign organization to Organizing for Action. From January 2009 until April 2011, a public blog on the OFA website posted every e-mail sent by Organizing for America domain e-mail addresses which were not sent by local leaders. Nearly all local leaders’ e-mails were associated with the 2010 or 2012 primary and general elections and none engaged in grassroots lobbying. These messages were formatted as “forwarded letters” rather than blog posts, and from January 20, 2010 until April 2011, every received grassroots lobbying e-mail was posted on this blog. Forty-five grassroots lobbying events were included in the dataset using e-mails also posted as OFA blog posts. Combining e-mail and blog observations resulted in a total of 245 grassroots lobbying e-mails sent via e-mail by OFA during Barack Obama’s two terms in office.

4 OFA frequently sent e-mails which did not include grassroots lobbying requests. This is not terribly surprising, as OFA engaged in policy-based actions beyond federal legislative policy making, occasionally focusing on policy implementation or state-level issues. More importantly, much as election campaigns work to lay foundations for later GOTV operations, OFA frequently sent e-mails requesting small donations and signing broadly-worded petitions which plausibly leveraged the foot-in-the-door technique to boost participation in later grassroots lobbying requests (Rogers 2012). Analysis of these other e-mails would shed light on how OFA worked to maintain relationships with supporters, but such a worthwhile analysis lies beyond the scope of this project, which is focused on OFA’s self-described emphasis on federal lawmaking via grassroots lobbying.

5 The perfect crossposting of OFA lobbying e-mails on the OFA public blog during a significant portion of the first term as well as qualitative evidence that OFA found that broadly sent general messages were more effective than targeted, specific messages, provide reason to believe that messages were consistently administered across the OFA mailing list once tailored through testing on much smaller subsamples (Kreiss 2016).

6 Lagging this variable by a week produces no significant changes to regression estimates.

7 The three weeks in the dataset ending in 2017 used the value from 2016 as that was the final year in which InternetLiveStats.com generated Internet usage estimates.

8 The Poisson model has a ratio of residual deviance to degrees of freedom of 1.12, suggesting that the data may be overdispersed and violate the assumptions of the Poisson distribution. Cameron and Trivedi’s overdispersion test, as well as a likelihood ratio test of the negative binomial and Poisson models, indicate that the data are overdispersed at the p = 0.05 level (Cameron and Trivedi 1990).

9 Sixty-three percent of the dataset’s observations passed without an OFA grassroots lobbying e-mail. This might lead to curiosity about a zero-inflated negative binomial model, but there is no theoretical expectation that grassroots lobbying e-mails could not be sent during certain weeks. If the president values policy, even when opportunity costs are high, PGLO mobilization may still be worthwhile. The predictors of the count component of a zero-inflated negative binomial model are assumed to be the same as the standard negative binomial, and all zero counts are assumed to have the same probability of belonging to the zero component. Such a model does not produce an appreciable improvement in model fit over a negative binomial model, and it is not discussed.