How Trusted Is Your Spokesperson? Grunig’s Relationship Theory Meets Social Media

Just in time for the post-Valentine’s Day breakup… how to measure the health of your relationships.

We’re spending a lot of time these days shoveling snow off the various roofs at Shankhassick Farm. Many of them were built over a hundred years ago, and when blizzard after blizzard piles the snow up, the pressure on those old timbers mounts. Unless you can relieve the strain, the whole structure is liable to collapse.

Just like many relationships around Valentine’s Day.

We’ve all been through it. The excitement and anticipation of a holiday with your true love. Wondering what to get him or her. Wondering what they’re going to get you. The pressure mounts, feelings build up and suddenly, just like an old New England roof, the strain becomes too much and the relationship collapses.

I’m not making this stuff up. Research shows that there is a huge spike in breakups and divorces in the period right after Valentine’s Day.

So, what does this have to do with public relations? Well, the business equivalent of deep snow on the roof is anytime there is a change in management or organizational structure. Relationships with clients, vendors, and employees are tested by change. Unless you measure the health and strength of those relationships and know what might weaken them, chances are there’s a breakup in your future.

Which is why we’ve spent several years testing ways to measure complex concepts such as trust, credibility, and reputation. We’ve long recommended using Grunig’s Relationship Instrument to survey your clients and stakeholders on the strength of your relationships. But surveys can be expensive and take a long time to implement.

In recent years, social media has emerged as a viable way to assess the emotions and opinions of stakeholder groups, assuming of course those publics are consumers and social media users. In 2011, my former company, KDPaine & Partners (now Salience Insight) tested the limits of automated as well as human content analysis to measure relationships, adapting Grunig’s relationship instrument to media content analysis (“Beyond Sentiment Analysis, Can we Automate Trust Measurement?”IPRRC 2011 Proceedings, p. 597).

That study demonstrated that automated measurement of relationship concepts was inaccurate and unreliable. However, it also indicated that human content analysis for specific concepts might have potential in measuring the health of stakeholder relationships, provided that there was sufficient data, coders were highly trained, and the coding instructions were well-tested.

In late 2012, I was approached by a popular Silicon Valley entrepreneur to measure his credibility and reputation among his target stakeholders. Initially, a conventional relationship survey was recommended. But when it became clear that the audience that mattered to this individual was his followers on Twitter and Facebook, a content analysis of comments and posts about this individual became an option. Using the lessons learned from the 2011 study, I proposed human content analysis of social media to test the credibility and trustworthiness of this entrepreneur among his core stakeholder groups.

A Study that Measured a Relationship Using Social Media Content

With a goal of measuring the extent of trust and credibility expressed by the stakeholders in social media, a human-coded content analysis was undertaken. The study included 7,529 social media items that mentioned this individual over a three-month period. Posts about the individual as well as comments on his own posts were included in the analysis. Grunig’s relationship theory was used as the framework for the analysis, and a detailed set of coding instructions were developed and tested based on Grunig’s definitions of relationships.

Each item was analyzed for specific qualitative characteristics that would reveal how the author perceived the celebrity. In other words: Did they trust him? Did they feel positively about him? And did they understand his philosophy and beliefs? 91% of all items discussed some aspect of a relationship.

In order to tease out from conversations how people felt about their relationships with the celebrity, we defined a list of key words, phrases, and concepts that were associated with each concept. We used trained humans to analyze the content and categorize each item for one or more of these concepts. Examples of some of the words that were associated with each concept were:

What We Found:

• Trust was the second most frequently discussed concept, most frequently expressed via Twitter, followed by Facebook and Pinterest. Trust was seldom expressed in blog posts, LinkedIn, or on Google+. Comments on Facebook posts yielded a great deal of high-quality conversation in which trust was expressed.
• Most people discussed the celebrity in terms of a communal relationship as opposed to exchange. Only 1% of all conversations reflected the perception of an exchange relationship. (There was virtually no discussion of control mutuality, and we therefore eliminated it from the analysis.)
• Discussions of the celebrity’s social media activity, political involvement, and biography did not generate as much trust as did his philanthropy. The celebrity’s primary charity was also a significant driver of trust.
• In terms of specific events, the celebrity’s participation in conversations around major national news such as Hurricane Sandy and veteran support were key drivers of conversations that conveyed trust.
• Retweets of the celebrity’s posts about Hurricane Sandy and a Twitter fundraising campaign were responsible for more than half the mentions of trust. Trust was also frequently expressed when the celebrity distributed media, or gave a heads up about something he noticed, or events he was involved in. People trust that what the celebrity shares is worth reading.
• Only 1% (44 stories) were negative, mostly driven by politics and election discussions. There was a minor amount of negativity about the produce with which the celebrity is still frequently associated.
• The high degree of nature conversations that conveyed a communal relationship were a bit misleading, since was is no way to identify whether someone was sharing a post because they like the animal, or they like the celebrity.
• Curiously, there were few comments on the celebrity’s posts to Pinterest, as people were most interested in repinning infographics rather than discussing them.

 What We Learned:

• By using trained human coders it is possible to use social media conversations to measure levels of trust, commitment, and satisfaction with your brand, assuming that your stakeholders are actively using social networks.
• There are several key elements necessary to developing an accurate assessment of relationship health in social media:
  1. An excellent and well-tested methodology and set of reader instructions.
  2. A brand that is already active in social media and has demonstrated a healthy level of engagement. You can’t just be shouting at the Twitterverse, there needs to be conversations about your brand already occurring.
  3. A sufficient quantity of conversation to make the data valid. (You can’t do a detailed analysis and draw conclusions on 50 tweets.)
  4. A coding methodology that has “buckets” for various types of activities, so you can correlate those activities to expressions of trust, commitment, etc.