Data Shows That Self-Reference Does Not Get Followers





If you like this post, or any of my work, please, nominate me for a Shorty Award.

Following up on my last post using TweetPsych Data, I looked at a metric opposing social behavior: self-reference. This time the dataset is well over 60,000 Twitter accounts.

What I found here is pretty clear, accounts that have more followers do not tend to talk about themselves much. Want more followers? Stop talking about yourself.

If you like this post, or any of my work, please, nominate me for a Shorty Award.

If you liked this post, don't forget to subscribe to my RSS feed or my email newsletter so you never miss the science.

{ 24 comments }

Chris January 28, 2010 at 8:33 pm

It’s a interesting and intuitive indicator, Dan. Good work in simplifying the concept.

On the same point, you may want to consider whether using 38% of the words (34 of 89 words) in the body of the post to ask for a Shorty nomination may have the same negative effect on your readership?

Joe Hall January 28, 2010 at 4:29 pm

“What I found here is pretty clear, accounts that have more followers do not tend to talk about themselves much. Want more followers? Stop talking about yourself.”

Dude, come on, correlation is not causation. Its pretty interesting data, but your just making an assumption here.

Joe Hall January 28, 2010 at 4:29 pm

“What I found here is pretty clear, accounts that have more followers do not tend to talk about themselves much. Want more followers? Stop talking about yourself.”

Dude, come on, correlation is not causation. Its pretty interesting data, but your just making an assumption here.

Joe Hall January 28, 2010 at 4:29 pm

“What I found here is pretty clear, accounts that have more followers do not tend to talk about themselves much. Want more followers? Stop talking about yourself.”

Dude, come on, correlation is not causation. Its pretty interesting data, but your just making an assumption here.

Joe Hall January 28, 2010 at 4:29 pm

“What I found here is pretty clear, accounts that have more followers do not tend to talk about themselves much. Want more followers? Stop talking about yourself.”

Dude, come on, correlation is not causation. Its pretty interesting data, but your just making an assumption here.

Joe Hall January 28, 2010 at 4:29 pm

“What I found here is pretty clear, accounts that have more followers do not tend to talk about themselves much. Want more followers? Stop talking about yourself.”

Dude, come on, correlation is not causation. Its pretty interesting data, but your just making an assumption here.

John Thawley January 29, 2010 at 6:57 pm

This also assumes that “more followers” is the core mission. Is it? Does “more followers” translate to better results? What about relevant followers? Doesn’t that mean more?

Just data… not necessarily a meaningful conclusion.

Joe McCarthy January 29, 2010 at 8:17 pm

At the risk of being self-referential, I just posted a blog entry about the commoditization of Twitter followers, in which I referenced a relevant study that differentiated between “Meformers” and “Informers” (and their respective median numbers of followers) … I also include a couple of links to posts on this blog. Anyhow, I'll include the relevant excerpt from the study below:

In a paper to be presented at CSCW 2010, Is it Really About Me? Message Content in Social Awareness Streams, Mor Naaman (@informor) and his colleagues analyzed the tweetstreams of 350 randomly selected users, and distinguish between Meformers – Twitter users who tend to share information about themselves, e.g., “tired and upset” – and Informers – users who share information on other people, places and things, typically including a URL – and report that Informers tend to have more friends [= followees] (Median=131) and followers (Median=112) than Meformers (Median=61, Median=42). I do not believe they included any celebrities in their dataset, but suspect some celebrities would represent outliers for the Meformer category.

Joe McCarthy January 29, 2010 at 8:17 pm

At the risk of being self-referential, I just posted a blog entry about the commoditization of Twitter followers, in which I referenced a relevant study that differentiated between “Meformers” and “Informers” (and their respective median numbers of followers) … I also include a couple of links to posts on this blog. Anyhow, I'll include the relevant excerpt from the study below:

In a paper to be presented at CSCW 2010, Is it Really About Me? Message Content in Social Awareness Streams, Mor Naaman (@informor) and his colleagues analyzed the tweetstreams of 350 randomly selected users, and distinguish between Meformers – Twitter users who tend to share information about themselves, e.g., “tired and upset” – and Informers – users who share information on other people, places and things, typically including a URL – and report that Informers tend to have more friends [= followees] (Median=131) and followers (Median=112) than Meformers (Median=61, Median=42). I do not believe they included any celebrities in their dataset, but suspect some celebrities would represent outliers for the Meformer category.

Joe McCarthy January 29, 2010 at 8:17 pm

At the risk of being self-referential, I just posted a blog entry about the commoditization of Twitter followers, in which I referenced a relevant study that differentiated between “Meformers” and “Informers” (and their respective median numbers of followers) … I also include a couple of links to posts on this blog. Anyhow, I'll include the relevant excerpt from the study below:

In a paper to be presented at CSCW 2010, Is it Really About Me? Message Content in Social Awareness Streams, Mor Naaman (@informor) and his colleagues analyzed the tweetstreams of 350 randomly selected users, and distinguish between Meformers – Twitter users who tend to share information about themselves, e.g., “tired and upset” – and Informers – users who share information on other people, places and things, typically including a URL – and report that Informers tend to have more friends [= followees] (Median=131) and followers (Median=112) than Meformers (Median=61, Median=42). I do not believe they included any celebrities in their dataset, but suspect some celebrities would represent outliers for the Meformer category.

Joe McCarthy January 29, 2010 at 8:17 pm

At the risk of being self-referential, I just posted a blog entry about the commoditization of Twitter followers, in which I referenced a relevant study that differentiated between “Meformers” and “Informers” (and their respective median numbers of followers) … I also include a couple of links to posts on this blog. Anyhow, I'll include the relevant excerpt from the study below:

In a paper to be presented at CSCW 2010, Is it Really About Me? Message Content in Social Awareness Streams, Mor Naaman (@informor) and his colleagues analyzed the tweetstreams of 350 randomly selected users, and distinguish between Meformers – Twitter users who tend to share information about themselves, e.g., “tired and upset” – and Informers – users who share information on other people, places and things, typically including a URL – and report that Informers tend to have more friends [= followees] (Median=131) and followers (Median=112) than Meformers (Median=61, Median=42). I do not believe they included any celebrities in their dataset, but suspect some celebrities would represent outliers for the Meformer category.

Joe McCarthy January 29, 2010 at 8:17 pm

At the risk of being self-referential, I just posted a blog entry about the commoditization of Twitter followers, in which I referenced a relevant study that differentiated between “Meformers” and “Informers” (and their respective median numbers of followers) … I also include a couple of links to posts on this blog. Anyhow, I'll include the relevant excerpt from the study below:

In a paper to be presented at CSCW 2010, Is it Really About Me? Message Content in Social Awareness Streams, Mor Naaman (@informor) and his colleagues analyzed the tweetstreams of 350 randomly selected users, and distinguish between Meformers – Twitter users who tend to share information about themselves, e.g., “tired and upset” – and Informers – users who share information on other people, places and things, typically including a URL – and report that Informers tend to have more friends [= followees] (Median=131) and followers (Median=112) than Meformers (Median=61, Median=42). I do not believe they included any celebrities in their dataset, but suspect some celebrities would represent outliers for the Meformer category.

Jorge Mir January 29, 2010 at 9:51 pm

Yeah. That sounds about right. Talk about yourself on facebook, I guess. Every time I mention anything about my personal life I lose followers.

Jorge Mir January 29, 2010 at 9:51 pm

Yeah. That sounds about right. Talk about yourself on facebook, I guess. Every time I mention anything about my personal life I lose followers.

Jorge Mir January 29, 2010 at 9:51 pm

Yeah. That sounds about right. Talk about yourself on facebook, I guess. Every time I mention anything about my personal life I lose followers.

Chase Brumfield January 30, 2010 at 6:48 pm

Dan, Looking at your data… I’m wondering something. The less followers they have doesn’t necessarily mean that these accounts “lost members” or have low numbers because of using social language. Perhaps many of these accounts are just new accounts. I’d like to see a side by side of this graph and also the age of these accounts (might not be possible) I’d also be very interested to know if some of the larger accounts started out using self-reference to build a loyal following and then those members took care of the rest by word of mouth… and the account didn’t need to self-reference anymore. However, I do think your post makes a very good point. There’s no reason to follow someone on twitter if they’re not providing something useful to you. This doesn’t however mean that self-reference is a bad thing. If you’re self referencing yet adding something useful to the consumer (aka humor/a coupon code/a sale) then I don’t see how that could lose followers. However, if you’re simply telling them what you had for lunch that day… well then obviously your not going to keep many. Any thoughts on this? P.S. I agree with the post below… always got to remember correlation is never causation

milian January 30, 2010 at 6:05 pm

latest music and photo gallery.
bagla , hindi and english songs album download.

Harry Houdini February 1, 2010 at 12:38 pm

If you talk rubbish, then of course you will always loose followers, whatever it is you write about. The best selling autobiographies are about interesting people, who have interesting things to say. Farcebook, twitter, etc are no different.

Tommy Toy February 1, 2010 at 4:05 pm

ZARELLE, there is definitely something wrong with tweetpsych. How do you explain celebrity posts who post about themselves. “i’m at starbucks”, “i just got laid”, “just finished wrapping the movie”, etc. yet some of these celebrities have huge numbers of followers.

In short, people want to know about you. Call that self-reference. Okay, I am guilty. I am not afraid to give an opinion, if you don’t like my opinion, at least offer a better solution. challenge me. if you want to drop me as a follower cause your silly ego or feelings were hurt, good bye.

Sorry, but i disagree with your findings.

lordmatt February 1, 2010 at 6:38 pm

I'd like to play with the raw data because aside from a strong trend at one end of the scale the least lines regression I'm guessing would be fairly flat. I'm only judging that your chart is actually a scatter plot and going best fit by eye (on a screen no less) but you're best fit doesn't look so comfortable to me. That said I'm not disputing your conclusions.

Mary Adams February 2, 2010 at 5:25 pm

Could you please clarify the data? On the left-hand axis the data range from .65% to 1.3%. What does social language mean?

alephnaughtpix February 2, 2010 at 2:18 pm

This is very interesting, although I think you might be jumping too easily to the conclusion “Want more followers? Stop talking about yourself.” on the assumption that people are following or unfollowing based on what you say. However it could be the other way around- you could be talking less about yourself as a *result* of having more followers- and therefore engaging in conversation with a greater number of people.

It's probably a bit of both, but it would be interesting if there was a way to determine how much is one direction, and how much the other.

Carri Bugbee February 4, 2010 at 5:01 am

I think this all depends upon the type of account you're running and why people are following you. To be truly helpful, I think this data would need to be adjusted for sentiment, though that's still a bit of voodoo.

Having ramped up and/or managed 40+ Twitter accounts in a wide variety of business categories, I have a sense of when you can get away with talking about yourself and when you can't.

Anecdotally, I think that individual tweeters who are self-deprecating or talk about mishaps tend to engender support. Everyone likes an underdog and we all want to be helpful! But nobody likes a boaster.

One of my biggest pet peeves is people who retweet other people’s retweets of their own tweets (for example, if I tweeted: RT @tweeter RT @CarriBugbee blah, blah, blah). I know many others who find this just as egregious, yet I still see so-called “experts” doing this to give themselves props.

For businesses, talking about deals and promotions has become a generally accepted practice, as long as these tweets are interspersed with other useful information. Many people will only follow a brand for deals or customer service (the Razorfish study released in October 2009 confirmed this), so they expect brands to talk about themselves.

@CarriBugbee
Social Profiles: http://www.CarriBugbee.com

Raj February 6, 2010 at 7:59 am

hey thats a wonderful tip here..

had never realized that this could be the tip top twitter users might be using

thnks :)

{ 8 trackbacks }