Conversations and Conversationalists in Social Media

Josh Bernoff recently wrote that Forrester Research has added "Conversationalists" to its Social Technographics typology of social media users, which had previously included the occasionally overlapping categories of creators, critics, collectors, joiners, spectators and inactives. He and his colleagues define conversationalists as people who post status updates on Twitter or other social networking sites at least weekly, and estimate that 33% of U.S. online adults fall into this category.

Josh and his colleagues offer a rationale for why this type of social media user is sufficiently distinct from other other patterns of use to merit the addition of the first new rung in the "ladder" in 2.5 years:

Conversationalists reflects two changes. First, it includes not just
Twitter members, but also people who update social network status to
converse (since this activity in Facebook is actually more prevalent
than tweeting). And second, we include only people who update at least
weekly, since anything less than this isn't much of a conversation.

I agree that users who post regular status updates represent an important new category to include on
the ladder. I'm curious, though, as to why Forrester estimates that the number of U.S. online adults who update their status on Facebook or Twitter is
33%, while a recent Pew Internet report (October 2009) on "Twitter and Status Updating" estimates that "some 19% of internet users now say they use Twitter or another service to share updates about themselves, or to see updates about others". That is, among all U.S. adults who use the Internet, Forrester estimates that 33% post updates weekly, while Pew estimates that 19% have posted or viewed updates ever … suggesting [to me] that fewer than 19% – in the Pew sample – have actually posted updates, and fewer still post updates at least weekly.

I found myself wondering about whether the significant discrepancies in these estimates are due to definitions, methodologies,
sample period or something else? According to the image caption below the ladder, the Forrester sample was collected in Q4 2009, but I cannot find any further details (without paying $499) on the Forrester page Introducing the New Social Technographics study. An article about the study in the Wall Street Journal reports that 10,000 people were included in the survey … but doesn't say anything about the methodology (e.g., how they were selected or how the posting behavior was ascertained). The page describing the methodology used for the Pew status update study states that "the results in this report are based on data from telephone interviews
conducted by Princeton Survey Research International between August 18
to September 14, 2009, among a sample of 2,253 adults, 18 and older." If the Pew study is based on self-reporting while the Forrester study is based on monitoring actual online use, that may help explain some of the discrepancy.

I'd speculated about a discrepancy between Forrester and Pew estimates a few years ago with respect to participation in the blogosphere (reading, writing and commenting).
Specifically, both research organizations were looking at the
proportion of Gen Y'ers (ages 18-26) who read blogs. Interestingly (to
me), in that case, the Pew estimate (41%) was significantly higher
than the Forrester estimate (24%). So I don't mean to imply that either
organization is inflating (or deflating) their numbers, but I am
interested to learn what factors might explain these discrepancies. I
posted a comment on Josh's blog post seeking clarification, but there has been no response (yet) and he has responded in comment below and on in a comment on his blog: the discrepancy may be due to the framing of questions on the respective surveys (more below).

FWIW, aside from the numbers, I think that Forrester's category of posting status updates is a more informative than Pew's category of posting or viewing
status updates (in Forrester terms, I imagine that people who view but
do not post updates would be labeled "spectators"). I also believe that
narrowing the scope to those who post weekly is far more useful than to
include anyone who has ever posted a status update – I've visited a number of Facebook or Twitter homepages that have only one status update, and I wouldn't consider those users "conversationalists".

And speaking of "converstionalist", while I think it's important to include a category of people who regularly post status updates, I wonder if "status updaters" – or "updaters" – would be more accurate (Molly, who also posted a comment on Josh's blog, also suggested "updaters"). To me,
"conversationalists" refers to the people who use @username directed
messages on Twitter, and/or post comments on Facebook status updates (or other FB content) and/or use wall-to-wall messages, to converse with
other users.
While I suspect that the majority of Facebook users converse regularly with other Facebook users – Facebook's own statistics report that 10% of Facebook users post status updates daily and the average user posts 25 comments on FB content per month – I'm not sure what proportion of these conversations take place via status updates. That said, if 10% of FB users post status updates daily, I would not be at all surprised if, say, 33% of FB users post updates at least weekly. 

Even if we restrict our consideration of conversations to status updates, the nature of these conversations are different in Twitter and Facebook. In Facebook, only explicitly designated friends can view or comment on status updates (or, at least, that was the case when the Forrester and Pew studies were done … before Facebook's most recent privacy setting changes). In Twitter, status updates are out there for all the world to see, and anyone can direct a message to anyone else (though the intended recipient of such a message may not read or respond to it). While Twitter allows users the option of "protected tweets", in which case only a user's followers can view their status updates, I have encountered very few Twitter users who opt for this restriction. In any case, generally speaking, conversations on Facebook have typically taken place within a restricted network, whereas Twitter conversations have typically taken place out in the open, and so there are some significant differences in the nature of the conversations – and conversationalists – on each social media platform … even though the differences between each platform seem to be declining over time. In a more thorough analysis of these differences, danah boyd has shared some thoughts on Twitter vs. Facebook status updates.

I've been doing some research into the ways that online Twitter conversations influence and are influenced by presence and interactions in the physical world. While there is definitely a dark side to the use of digital backchannels in shared physical spaces, there are plenty of positive examples of people using Twitter to converse and connect with friends and consequential strangers in real life (my favorite collection of examples: Twittertales, an ebook containing short stories compiled by "Conversation Agent" Valeria Maltoni).

Others have also been researching conversations on Twitter, using the more restricted definition of conversations (@username directed messages on Twitter). A number of recent studies suggest that

• conversational tweets – and tweeters – are still in the minority
• the definition of a conversation is still very much an open issue
• a methodology for sampling and counting conversations and conversationalists is still an open issue

Java, et al., [2007] found 12.5% of tweets and 21% of tweeters in their general dataset of 1.3 million tweets and 76,177 tweeters were involved in a conversation; Huberman, et al. [2009] found that 25.4% of all tweets were directed, and Honeycutt & Herring [2009] found that 30% of their 1,472 tweets were directed, of which 31% – or 9% of all tweets – received a public reply within 30 minutes. boyd, et al. [2010] found that 36% of a random selection of 720,000 tweets included a reference to a user (“@user”), and 86% of these began with such a reference and thus 31% were inferred to be a directed message. Shamma, et al. [2009] found 17.8% directed messages in their dataset of 3,238 tweets about a recent U.S. presidential debate – also collected by searching for hashtags – of which 10.23% were reciprocated (i.e., were involved in a conversation).

There are clearly conversations – and conversationalists – in social media, and I'm glad a growing number of people and organizations are looking at [some of] the practices of using social media to engage in conversations. I hope we will converge on definitions and methods for measuring conversations as we gain more experience in analyzing – and using – social media in these ways. I know there is a workshop on Microblogging coming up at the CHI 2010 conference this April, co-organized by Julia H Grace (@jewelia), Dejin Zhao (@djzhao) and danah boyd (@zephoria), so perhaps we'll see some progress on this convergence in the near future.