This article describes a concern with relying on metrics that purport to show authority metrics. The industry increasingly relies on it but there may be compelling reasons to not rely on those metrics. Here are some thoughts for your consideration.
Correlation is not a ranking factor In a nutshell, third party Authority metrics are created with data that
correlates with sites that tend to rank well. However these correlations aren't necessarily factors that made them rank well.
For example, let's examine a hypothetical situation of sites that tend to rank well and also have large amounts of links and Facebook shares. That's a correlation. The Facebook likes didn't cause a site to rank well. The Facebook likes and large amount of links aren't necessarily related to each other. It may be simply that a
commercial site with a large amount of links and is
SEO optimized for ranking tends to rank well and often has a lot of FB likes (because they are SEO optimized).
Can you see how
correlations can be a rabbit hole of speculation? What's particularly troubling is that for every article published about correlated ranking symptoms there is virtually no
citations of scientific researchor filed patents to show that the claims are potentially true. One can make opinions and conspiracies about a wide range of potential ranking factors but if you can't cite any research to prove that this has even been researched, much less implemented at Google, then it's just air coming out of someone's mouth.
Let's return to the hypothetical situation. If we were to deduce that the high rankings were at least partially due to the high amount of links then we would have a strong chance of being correct. The reason is because there is substantial scientific documentation that search algorithms depend at least partially on inbound links for ranking websites. That's more than just a simple correlation because there is a scientific citation to back up the assertion.
Unfortunately, many of the ranking factor correlations one reads about do not feature such citations.
Questionable metrics used to calculate Authority Metrics Some of the factors that go into calculating Authority Metrics are questionable. For example, some use a version of Seed Set Based TrustRank, but that methodology has been
scientifically proven to be on shaky ground. [thesempost.com] This methodology was shown to be flawed many years ago, shortly after the Seed Set Based TrustRank method was proposed in a scientific research paper. Why do companies continue to use a method that was shown to be flawed? Are you wise to trust a metric that uses a compromised methodology?
Correlation does not imply causation An issue of concern is that many popular metrics are measuring correlations. This is a profound flaw. Correlations are a notoriously poor statistical signal for understanding an event. A better method would be to simply understand the science of Information Retrieval. It's taught at universities. There is no need to wander in the dark like the folk tale of the blind men touching an elephant, theorizing that the elephant is like a tree or a snake, depending on what part they are touching. Third party metrics that measure correlations are similar.
Matt Cutts, senior Google engineer [searchengineland.com]
Moz published a story today named Amazing Correlation Between Google +1s and Higher Search Rankings in which Matt Cutts responded to in Hacker News thread saying, “correlation != causation.”
Gary Illyes via Twitter [twitter.com]
Gary's being humorous here. However it can be seen as a reflection of his frustration with SEO's tendency to seize on false correlations.
þ
@methode I went to NYC, weather turned bad. Came to Las Vegas, started to rain and it's cold. Causation or correlation?
Wikipedia [en.wikipedia.org]
Correlation does not imply causation is a phrase used in statistics to emphasize that a correlation between two variables does not necessarily imply that one causes the other... For example, in a widely studied case, numerous epidemiological studies showed that women taking combined hormone replacement therapy (HRT) also had a lower-than-average incidence of coronary heart disease (CHD), leading doctors to propose that HRT was protective against CHD. But randomized controlled trials showed that HRT caused a small but statistically significant increase in risk of CHD. Re-analysis of the data from the epidemiological studies showed that women undertaking HRT were more likely to be from higher socio-economic groups (ABC1), with better-than-average diet and exercise regimens. The use of HRT and decreased incidence of coronary heart disease were coincident effects of a common cause (i.e. the benefits associated with a higher socioeconomic status), rather than a direct cause and effect, as had been supposed.
