Effective Social Media Engagement for Nonprofits: What Matters?

We employ public management relationship theory to examine how nonprofits can effectively engage social media stakeholders in two-way communication. Though many nonprofit organizations have a social media presence, there is variance in how well organizations use social media to engage stakeholders. Simply having a social media presence is not enough to engage stakeholders. We examine Facebook posts of a stratified random sample of youth development organizations to determine what predicts stakeholder engagement. We find the type of Facebook post is a significant predictor of stakeholder engagement. Longer posts also significantly predict increased stakeholder engagement. At the organizational level, having many posts is a significant negative predictor of stakeholder engagement, indicating that users may feel bombarded and are less likely to engage. Increased organizational spending on advertising as a proportion of total budget is positively associated with stakeholder engagement.


Introduction
How can nonprofit organizations engage social media stakeholders?Engagement involves twoway communication between nonprofit organizations and social media users.While many nonprofit organizations have a social media presence, owning a social media account does not equate to two-way engagement with stakeholder groups (Maxwell & Carboni, 2014).Nonprofits can include social media as part of a larger communications strategy to engage stakeholders.Although the public relations literature has long emphasized targeted communications strategies for organizations, scholars have only recently begun to examine the targeted use of social media for stakeholder engagement (Bortree & Selzter, 2009;Maxwell & Carboni, 2014;Waters, 2008;Waters et al., 2009;Westcott, 2007).Social media provides a communication platform for nonprofits to reach large numbers of stakeholders quickly, efficiently, and publicly.Understanding effective social media engagement is especially important for nonprofits, as the number of nonprofits using social media has exploded in recent years (M+ R NTEN, 2012).
However, many nonprofits do not strategically employ social media for two-way communication with users (Bortree and Seltzer, 2009;Brock and Buteau, 2012;Foundation Center, 2014;Lovejoy, Waters & Saxton, 2012;Waters et al., 2009;Xifa & Grau, 2010).Nonprofits potentially face multiple challenges to effective use of social media for stakeholder engagement, including lack of staff, knowledge or resources to manage social media presence, privacy concerns of clients, and grantor restrictions on information that can be shared.For example, a recent study notes a network of organizations failed to adopt social media usage as a result of organizational policy (Maxwell & Carboni, 2014).A recent survey found that 74% of nonprofits using social media employ it as a megaphone for one-way engagement rather than a platform for strategic two-way communications.Additionally, 60% of respondent organizations were unsure why they needed to evaluate the effectiveness of social media presence (Sharma, 2014).
We use Hierarchical Linear Modeling (HLM) to examine Facebook post data of youth development organizations to determine which Facebook posts generate the most engagement from stakeholders.We use Relationship Management Theory (RMT) as a theoretical base to highlight the interaction with which nonprofits engage their stakeholders.RMT theory emphasizes communication as a tool to build relationships with stakeholders as a means to achieve organizational goals (Ledingham, 2003).Stakeholder engagement through social media is a communication strategy used to develop and maintain relationships with stakeholders that may contribute to achieving organizational goals.We find that Facebook post type is a significant predictor of stakeholder engagement.Specifically, photos are significantly more likely to result in stakeholder engagement than other types of status updates such as videos, external links, or plain text.Longer posts also significantly predict increased stakeholder engagement.At the organizational level, having many posts is a significant negative predictor of stakeholder engagement, indicating that users may feel bombarded and are less likely to engage.Also at the organizational level, increased spending on advertising as a proportion of total budget is positively associated with stakeholder engagement.

Literature Review and Hypothesis Development
Relationship Management Theory (RMT) recognizes that developing relationships with stakeholders is critical to the well-being of organizations and that communication is a strategic tool to build and maintain relationships (Dozier et al., 1995;Ledingham, 2003).Aspects of relationships, especially communication between focal organizations and stakeholders, are the appropriate units of analysis in this theory.Communication with stakeholders should be dialogic, or two-way engagement, rather than simple one-way information sharing, and should be tailored to specific stakeholder groups (Ledingham, 2003;Ledingham & Brunig, 2000).In this study, we focus on nonprofit organization communication with Facebook users to determine which types of posts by nonprofit organizations are most likely to lead to Facebook user engagement with organizational posts.
Social media are natural platforms used to encourage two-way communication between organizations and social media users in ways not previously possible (Lovejoy & Saxton, 2012;Saxton, Guo & Brown, 2007).Social media provide organizations with opportunities to engage in simultaneous dialogic communication with many stakeholders in a public space.The way communications are managed by nonprofit organizations may affect stakeholder attitudes, perceptions, knowledge and behavior toward organizations, which can result in more or less engagement by stakeholders (Ledingham and Brunig, 1998).Stakeholder engagement is an indication of the relationship between the nonprofit and targeted stakeholder group (Facebook users) (Broom et al., 2000).Facebook posts are a communication strategy used by nonprofits to engage stakeholders.One way to gauge whether organizations are effectively engaging with stakeholder groups via social media is to determine whether exchanges are two-way with targeted groups responding to the communications of the nonprofit organization.As social media use among organizations becomes more prevalent, organizations compete for the attention of users in the social media space.User engagement with an organization's social media communications signals that organization has broken through to "win" the competition for user attention.
We examine Facebook social media communication strategies to determine which strategies are likely to result in two-way engagement of stakeholders.Facebook user engagement is an underdeveloped area of scholarly and practical inquiry, even though nonprofits organizations' use of social media has exploded in recent years.Facebook users are distinct from other stakeholder groups in that they have chosen to engage with the nonprofit organization through a social media platform.While these users may interact with the organizations in other ways, we only consider the "Facebook user" aspect of their interaction with the nonprofit in this paper.Although nonprofits generally still rely primarily on email communication, Facebook is the most popular social media platform for nonprofits (Auger, 2013;M+R NTEN, 2012).Despite increased Facebook presence, research has established that nonprofits do not quite know how to use social media and risk losing users with too many messages (M+R NTEN, 2012).
How an organization chooses to communicate is a strategic decision about how to share information and engage stakeholders (Ledingham, 2003).Information should be useful to stakeholders (Taylor, Kent & White, 2001).Usefulness of social media often depends on how information is being shared (Crespo, 2007).In Facebook, the type of post is likely to influence whether users engage with the post by liking, commenting on, or sharing the post.Facebook offers four types of posts: status updates, links to external sites, multimedia posts with photo, and multimedia posts with video.Status updates include only text and can be experienced by being read.Multimedia posts with photo or video may also contain text.Photos and videos are experienced by being viewed.Posts with links include web links to external websites.We predict that as posts grow more dynamic, they will result in more engagement.By dynamic, we mean that users experience the post in multiple ways.Status updates are read, while multimedia posts with photos and video, and posts with external links require more participation to be fully understood.Those posts that require more participation are likely to result in more engagement.This relationship is stated in the following hypothesis: H1: More dynamic posts will result in more total post engagement by social media stakeholders.
Length of post is also likely to be important.We predict that longer posts will result in more engagement because there will be more information for stakeholders to experience and respond to, making it likely that more users will relate to the post.This relationship is stated in the following hypothesis: H2: Longer posts will result in more total post engagement by social media stakeholders.
In addition to characteristics about the individual post influencing total engagement for that post, we propose that organizational characteristics will influence stakeholder engagement with a post.Stakeholder engagement is a dynamic process made up of individual transactions that form a relationship between the organization and stakeholder group (Ledingham, 2003).Careful management of transactions may increase engagement.Additionally, organizational factors such as dependence on contributions, capacity to communicate with external stakeholders, and organizational size may influence the total engagement received by a particular post.
Prior use of dialogic strategies may lead to greater engagement (Bortree and Selzter, 2009;Sweetser and Lariscy, 2008;Taylor, Kent & White, 2001).We predict that total Facebook activity will influence stakeholder engagement with a particular Facebook post.As total Facebook activity goes up, total engagement for a post will also go up.This relationship is stated in the following hypothesis: H3: As total Facebook activity increases, total post engagement for individual posts by social media stakeholders will also increase.
In a social media platform, communication with stakeholders can be viewed by many people; it is public.Community support is tied to developing dialogic communication with stakeholder groups (Ledingham and Brunig, 1998).Following this, we predict that total stakeholder engagement for individual posts will be related to total stakeholder engagement for all posts by the organization.As organizations receive more support from Facebook stakeholders, they will raise their organizational profile, and stakeholders will be more likely to engage specific posts.This relationship is stated in the following hypothesis: H4: As Facebook stakeholder engagement for an organization increases, the total post engagement for individual posts by social media stakeholders will also increase.
Organizational factors unrelated to Facebook activity are also likely to influence total engagement with an individual post.Following a resource-dependence perspective (Pfeffer and Salancik, 1978), we propose that as nonprofit organizations become more dependent on contributions as a source of revenue, they are more likely to focus on creating posts that generate stakeholder engagement.Where organizations are dependent on stakeholders for resources, are more likely to be strategic about communications to ensure continued stakeholder investment (Ledingham, 2003).This relationship is stated in the following hypothesis: H5: As nonprofit organizations become more dependent on contribution revenue, the total post engagement for individual posts by social media stakeholders is likely to increase.
Additionally, organizations with dedicated public relations or public relations like staff are more likely to engage in targeted social media that meets organizational goals (Curtis et al., 2010).We predict that as organizations commit more resources to external stakeholder engagement, their Facebook posts will receive more engagement.This relationship is stated in the following hypothesis: H6: As nonprofit organizations invest more resources in communicating with external stakeholders, the total post engagement for individual posts by social media stakeholders will increase.

Data and Methods
We analyze Facebook data from a stratified random sample of 150 youth serving nonprofit organizations based on National Taxonomy of Exempt Entities (NTEE) code.The (NTEE) is the IRS classification system used to organize nonprofit organizations in the United States (Urban Institute, 2014).Originating from the National Center for Charitable Statistics (NCCS) at the Urban Institute, NTEE separates youth development organizations from a taxonomy of ten overarching categories.These include mentoring, youth centers, agricultural programs such as 4-H, advocacy organizations, and Boys and Girls clubs (Guidestar, 2014).Guidestar (2014) lists nearly 37,000 youth development organizations in its most recent directory, which classifies youth development under Human Services.If a separate search is employed for youth overall, the directory produces close to 49,000 organizations.Of the tens of thousands of youth organizations, the missions involve a range of services, but generally focus on leadership, sports, emotional growth, after school, and intervention strategies for at-risk youth, and employment services (Guidestar, 2014).
Due to labor intense social media data collection, we chose a random sample of 50 organizations from three revenue classes of youth serving nonprofit organizations for a total of 150 organizations.This sample size is appropriate for our population.Revenue classes range from $500,000 to $1 million, $1 million to $2.5 million, and $2.5 million to $5 million.We chose these revenue classes because organizations with more than $500,000 in revenue are more likely to have resources and capacity to strategically engage stakeholders via social media.Of the 150 organizations we sampled, 116 organizations had an active Facebook presence during the period studied.

Data Sources
We examine Facebook activity data and IRS Form 990 data for individual organizations.Facebook data are gathered over a two-week period using Simply Measured.Nonprofit data come from Guidestar's individual organizational IRS Form 990 reports, taken from the prior year.Simply Measured is a social media analytics platform developed for organizations to assess their social media efforts including engagement of social media users.This program allows for detailed content analyses of Facebook engagement data.Engagement is defined as likes + shares + posts (Simply Measured, 2014).Reports for individual organizations were downloaded in July and August of 2014.Each report covers two weeks' worth of Facebook post activity for each organization.Guidestar is a commercially available program, offering financial data and 990 forms for tax-exempt organizations in the U.S. The IRS Form 990 provides data collected from the IRS related to nonprofit organizations' finances, missions, and programs.

Modeling Strategy
To analyze the data, we employ a hierarchical linear modeling (HLM) strategy.HLMs are multilevel regression models designed to evaluate cross-level interactions in nested data.This strategy is widely used by researchers in multiple fields to assess effects of independent variables across nested cases (Raudenbush and Bryk, 2002).HLM should be used if data are clustered into statistically-meaningful groups, meaning that data are not independent within a cluster, and regression relationships vary by group.Classic examples of nested data include students in schools and children in families.In this paper, we nest individual Facebook posts within organizations.A likelihood ratio test indicates individual post data is meaningfully clustered into organizations.

Variables
Dependent Variable-In HLM, dependent variables are at the lowest level of data (Level 1), which is the individual Facebook post.The dependent variable for our model is total engagement per post, which includes the number of likes, comments, and shares for each post.It is a continuous variable.Because this variable is not normally distributed, we use the natural log of total engagement per post in our model.Data comes from Simply Measured Reports.Descriptive statistics can be found in Table 1.
Level 1 Independent Variables-Level one independent variables relate to individual Facebook posts and include the type of post and length of post.Data for both variables come from Simply Measured reports.Type of post is a categorical variable with the following categories: status updates, links to other content, multimedia photo, or multimedia video.In our model, the reference category is status update.This variable relates to Hypothesis 1. Length of post is a continuous variable that refers to the number of characters in a post.Because this variable is not normally distributed, we use the natural log of length of post in our model.This variable relates to Hypothesis 2. Descriptive statistics can be found in Table 1.
Level 2 Independent Variables-Individual post data are clustered into organizations.Level two independent variables are measured at the organizational level.Variables are overall Facebook activity during the period studied; stakeholder engagement with other Facebook posts by the organization; contributions as a proportion of total annual income; and advertising as a proportion of total annual expenses.Data for overall Facebook activity and engagement come from Simply Measured Reports.Contribution income and advertising expense data come from IRS Form 990 reports.
Overall Facebook activity is a continuous variable that includes the count of all posts during the time period studied.This variable relates to Hypothesis 3. Engagement from other Facebook posts is a continuous variable that includes the total engagement with all other posts by the organization during the period studied.This variable relates to Hypothesis 4. Contributions as a proportion of total income is a proxy variable for how dependent the organization is on stakeholders.As contributions increase, dependence also increases.This variable relates to Hypothesis 5. Advertising as a proportion of total expenses is a proxy variable for how many resources the organization commits to appealing to external stakeholders.It relates to Hypothesis 6. Descriptive statistics for all variables can be found in Table 1.
Controls-We control for organizational size.The total assets in an organization is used as a proxy for size.In this data set, organizational income and assets were highly correlated.Models with income as a proxy for size rather than assets produced statistically similar results.Data are derived from IRS Form 990 reports for individual organizations.Descriptive statistics are found in Table 1.

Results
We found strong support for three of our hypotheses and partial support for a fourth hypothesis.Two of our hypotheses were not supported.Results for each hypothesis are discussed below.
Results for our multilevel model can be found in Table 2.
There was partial support for Hypothesis one.Status posts were the reference category for this model.We predicted that more dynamic posts (photos, videos, and web links) would result in more total post engagement.Multimedia posts with photos had significantly more engagement than status posts.Multimedia posts with videos had no statistical difference in engagement than status posts.Posts with web links had significantly less engagement than status posts.
These results indicate that photos are more likely than other categories of posts to result in Facebook user engagement.
There was strong support for hypothesis two.We predicted longer posts would result in increased engagement of social media stakeholders for individual posts.This variable has a statistically significant effect on total post engagement in the expected direction indicating longer posts result in more engagement.We did not find support for hypothesis three.We predicted that increased Facebook use would lead to more engagement for individual posts.This variable was statistically significant but in the wrong direction.The model indicates that as the number of total posts by an organization increases, the total engagement for individual posts declines.
We found strong support for hypothesis four.We predicted the amount of Facebook engagement with other posts would be a predictor of Facebook engagement with a single post.This variable has a statistically significant effect on total post engagement in the expected direction.This indicates more post engagement for other posts is associated with increased total engagement for individual posts.
We did not find support for hypothesis five.We predicted that as nonprofit organizations became more dependent on contribution revenue, total post engagement would also increase.
The variable did not have a significant effect on total post engagement.The effect was not in the expected direction.This indicates there is a negative relationship between reliance on contributions and total post engagement.
We found support for hypothesis 6.We predicted that as nonprofit organizations invested more in capacity to communicate with external stakeholders, total post engagement would also increase.This variable was statistically significant in the expected direction indicating that increased capacity to communicate with external stakeholders is a predictor of total Facebook engagement.

Discussion
This study sheds light on which social media communication strategies and organizational factors are likely to lead to stakeholder engagement.We viewed stakeholder engagement with Facebook posts as a sign of two-way communication and engagement of stakeholders that may relate to organization achieving its broader goals.When stakeholders are engaged in social media platforms, they may be more likely to engage with the organization in other ways.
While our study relied on data from youth serving nonprofit organizations, the results may be useful in other types of nonprofits.For example, understanding that photos are significantly more likely than other types of posts to generate support may lead nonprofits to experiment with more photos on their social media platforms.That longer posts lead to more engagement may lead nonprofits to focus on platforms like Facebook where they want to engage stakeholders rather than platforms like Twitter where post lengths are limited.Also of note is that the more messages a nonprofit posts, the less engagement individual posts will receive.This raises questions about whether stakeholders are experiencing message bombardment and disengage from organizations' attempts at dialogic communication.
Questions remain about how organizational factors influence stakeholder engagement on social media platforms.Expenditures of resources on advertising were a strong positive predictor of social media engagement, while contributions as a percentage over total revenue did not influence stakeholder engagement.It could be that a temporal order exists where investment in advertising leads to more social media stakeholder engagement, which in turn leads to more contributions.In other words, does targeted social media engagement lead to returns on investment in the form of donations?This is different from the question we posed as we focused on contributions as an antecedent to engagement related to dependence of organizations on contributions.More research is needed to answer the important question of whether stakeholder engagement through social media leads to increased contributions to organizations.

Conclusion
The importance of Facebook as a social media tool has multiple implications for nonprofit organizations.Used strategically, social media has the potential to increase the number of engaged stakeholders.Specifically, social media use has potential to increase dialogic communication with stakeholders in a low-cost, public arena.Social media consultants typically recommend that online nonprofit or public sector communication adhere to the same etiquette standards as face-to-face communication (Mergel & Greeves, 2013).Within these guidelines, varying employees assigned to differing stakeholder groups (e.g., donor relations vs a program manager) may adopt their own social media communication messagesrather than relinquishing all social media to a single staff member who is often inexperienced.Each retains authority over the content made available to the targeted audience.
However, many nonprofit organizationss employ social media for information dissemination without understanding how to effectively engage stakeholders in two-way communication.To effectively use social media for stakeholder engagement, nonprofits should introduce social media into their larger communication strategy as part of the strategic planning process.This involves devoting resources to developing social media platforms.Nonprofits should also involve staff with communications responsibilities in overseeing social media messaging as part of the larger communication strategy.
Communication strategies might include making sure organizational brand is consistent across social media platforms and that each platform is used to its fullest advantage.Doing so requires staff to work within the branding identity across all platforms.Such an approach might include the use of colors, logs, and style of writing across various forms of media (Mergel & Greeves, 2013).By adopting a brand message, employees can work under a general policy set forth by the organization, ensuring professional comments and posts.Organizational policies that operate within the strategic framework can allow for creativity while simultaneously creating a professional environment that prevents personal use or inappropriate posts.
Nonprofit organizations also have the ability to effectively gather their own Facebook, Twitter, and overall Web 2.0 analytics that inform their future decision making through online platforms like Simply Measured.All targeted approaches to social media use, such as donor cultivation, event announcements, and youth recruitment to new programs, can be measured quickly and easily, including who and what type of posts result in the intended audience responding.
Understanding where different types of posts and the use of differing media platforms (e.g., Facebook, blogging, Wikis, or Twitter) are most beneficial for an organization are both key to improving two-way communication.Peer networking with partner agencies, on the other hand, is a potentially useful benefit of social media if adopted with privacy and management concerns in mind (Maxwell & Carboni, 2014;Mergel & Greeves, 2013).
In sum, using social media effectively requires careful planning and strategy implementation on the part of organizations.Future scholarly inquiry should further emphasize how organizations can manage social media to realize their organizational goals.For example, which platforms are best for engaging donors?It should also provide insight as to how social media strategies are tied to organizational characteristics.For example, are organizational characteristics related to staffing associated with more effective social media use?Further research should also focus on characteristics of social media users, an issue we were unable to examine due to the limitations of our dataset.For example, are engaged social media users also donors?As the prevalence and sophistication of Web 2.0 technologies grows, these questions will become even more important and timely for nonprofit organization strategic planning.