What are we really seeing?
I’m not going to explain the scientific method to someone that describes himself as “The Social Media Scientist.” In fact, I have enormous respect for Dan Zarrella and Hubspot, and I know he understands it. So why he writes an article like Twitter Accounts with a Profile Picture Have 10 Times More Followers Than Those Without beats the hell out of me.
Bottom line is, you can’t notice a correlation between two things and then assert one is the cause of the other without eliminating other possibilities.
Here is where he asserts causation: Effect of Profile Picture on Followers [emphasis mine].
Oh, and here, too: “…if you want to get followers on Twitter, it’s a good idea to upload a picture of yourself.”
And here are 5 likely alternative explanations for the correlation Dan noticed.
- Number of tweets. Maybe those without pictures tend to tweet less, and it is the latter variable that is causing their lack of followers.
- Spammy content. Maybe those without pictures tend to tweet spam more often, and it is the latter variable that is causing their lack of followers.
- Age: Maybe those without pictures tend to be newer accounts, and it is the latter variable that is causing their lack of followers.
- Location: Maybe those without pictures tend to live in locations where Twitter use is less common, and it is the latter variable that is causing their lack of followers.
- Lack of effort: Maybe those without pictures tend to put less effort into acquiring followers, and it is this latter variable that is causing their lack of followers.
I could go on and on. But that’s where you come in! Extra points for funny hypotheses left in comments.
As I’ve said before, Dan noticed something interesting here just by sharing the correlation he found. That alone was worthy of a blog post and perhaps a larger conversation. I don’t understand why he had to then jump to a conclusion and taint the larger effort. I really don’t. Based on our conversation (click “show conversation” after jump), he doesn’t seem to think there’s a problem with what he did. Cognitive dissonance? Who knows?
Remember, I’m writing this because I’m a fan, and posts like this might help influencers like Dan get better—and by extension, our study of the social media universe can improve. But I won’t say posts like this will do much of anything. That, after all, would be assuming causation 
.
{ 6 comments… read them below or add one }
Ian – Although I am virtually certain that the initial fact as it’s asserted here is true (those with pictures have way more followers), I agree wholeheartedly with you. You can’t prove any causation. Instead, I would be willing to wager that your causes are more likely than the photo alone, with the possible exception of #4 on your list.
There’s no easy way to determine if the photo alone would make a difference, although I do tend to give less credence to those without an avatar of their own. Remember the “owl” default one that Twitter used to have? I like the multicolored birds better.
Jason Crouch´s last blog ..Easter fun for Jenny!
This kind of thinking of the evidentiary basis for belief seems to infiltrate science even at the highest level.
You’re left only with a correlation between your independent variables no matter how much you analyze the data. This is the bane of most of the social sciences. You really can’t establish causality without a controlled double blind study.
With twitter as the medium… I think such a study might be easy to put together. Ultimately though… I’m wondering if the set of twitter accounts without a profile picture is even statistically significant?
–bjw
By the way for a VERY thorough statistical treatment of the effect of profile picture (in this case on a dating site) check out: http://blog.okcupid.com/index.php/2010/01/20/the-4-big-myths-of-profile-pictures/
@jason- You’re right, there’s no easy way–but my point is that we shouldn’t make statements like “effect on” if we haven’t done the legwork, especially if there are many other likely explanations.
@brian- So yes, ultimately proving this kind of thing without a doubt is nearly impossible, but this is a far cry from at least *looking* at other variables besides the one we come across first, or admitting that there are alternative explanations, neither of which Dan Zarrella has done.
Ian Greenleigh´s last blog ..5 causes just as likely as the one @danzarrella seemingly picked out of a hat
I suppose there is little to say that has not already been said. Further examination would probably verify that those who lack a profile picture value Twitter less and therefore use it less, which naturally means they would have fewer followers. If Ashton Kutcher were to remove his profile pic, I doubt his following would cease. On a lesser-known figure, the lack of an authentic profile pic may cause a would-be follower to doubt the legitimacy (or usefulness) of the account.
That being said, I want to take a shot at the extra credit…
If Dan has not already launched the following study, I am sure he soon will. There is a high correlation between people who use social media and have no idea who Dan is (unfortunate, really). Had more users of Twitter known Dan, they likely would have been inclined to update their profiles to include a photo so as to give him a better head count. Unfortunately, this was not the case. Since there is a high correlation of Dan tweeting his discoveries and his discoveries getting re-tweeted or blogged, it is likely that he will be re-tweeted enough to eventually know and be known by everyone on Twitter. At that time, when we have all begun using profile pics properly, Dan will find that there is a high correlation between using a profile pic and having both a strong and weak following on Twitter.
Of course, this is only poking fun. I am fascinated by and grateful for the work Dan does.
@jason t.
“Further examination would probably verify that those who lack a profile picture value Twitter less and therefore use it less, which naturally means they would have fewer followers.”
You put it better than I ever could have.
And, regarding your proposed study…he might actually be up for it, since he eventually RTd this very critique.
Thanks for stopping by!