Can't access the study, but the article doesn't elaborate on how they're able to make the jump from correlation to causation ("companies that respond to 75% of

Can't access the study, but the article doesn't elaborate on how they're able to make the jump from correlation to causation ("companies that respond to 75% of complaints on twitter openly also tend to fall in brand quality metrics" -> "responding to complaints on twitter openly cause companies to fall in brand quality metrics"), and they also don't talk about the effect size.