LinkedIn profiles and statistical inference
LinkedIn tells me that:
Members with a photo get 11 times more profile views.
</img>
I wonder how misleading that statement is.
I don’t think it’s unfair to conclude that the impression LinkedIn wants to give a casual reader when they say “Members with a photo get 11 times more profile views”, is that “if you put a photo on your profile, you will get 11 times more profile views.” Especially when they put that statement on a big button that prompts you to add a photo.
11 times more profile views than what? What is the 11x multiplicative comparing to? Members who do not have a photo on their profile? Or the same members who now have the photo compared to when they had no photo but an otherwise identical profile?
The impression I get is that it’s the former: members with a photo get 11 times more profile views than members without a photo. Even if that’s not what they mean, that’s what a casual reader would think.
If that’s the case, there is an obvious problem of possible third variables not being controlled for. For example, it is very likely that members with a profile photo also put a lot more effort into maintaining their LinkedIn profile, which includes adding a summary about themselves (which shows up in the profile badge others see), and having more connections. That maintenance has to contribute to increased profile views.
In fact, LinkedIn also tells you that adding a summary relates to increased profile views:
</img>
Same questions apply. What is the 10x comparing to? Is that the unique benefit of adding a summary, or is the benefit of adding a profile photo – and other things – included? If you add a profile photo and a summary, do you get 110x more profile views, or is it closer to the figure you get for adding either, because hardly anyone who has a profile summary doesn’t have a photo and vice versa, and so the reason members with a photo get 11x the profile views is that almost all of them also have profile summaries and more connections?
I don’t have a strong guess as to whether LinkedIn controls for other variables before reporting the 11x and 10x figures1, but my weak guess is that they don’t. If I’m right, then those enticements on the buttons prompting you to add photos and summaries are misleading.
-
Although I’m sure their internal staff with access to the data know. ↩︎