I first admit that I am neither a mathematician nor a statistician. I struggled with math at school enough to get an undergraduate degree and wasn't used to everything until graduate school. Still, it was a standard business stats class that most people suffered while seeking an MBA. I searched for a real statistician, just as I did when I worked with a real IP lawyer in Google's Featured Snippet Legality article. Most importantly, we needed someone who wasn't working in the SEO space to avoid observer bias, that is, who unknowingly projected expectations into the study. My search led me to statistician Jen Hood.
Jen studied mathematics and economics at Bridgewater College in Virginia, and for most of her 15 years she has worked as a statistician. She was a Volvo data analyst. Since 2019, she has been working as an analytics consultant for her own company, Avant ghost mannequin effect Analytics, primarily helping small businesses without in-house analysts. advertisement Continue reading below In the first discussion, I talked about how most studies on SEO depend on the concept of statistical correlation. Statistical correlations indicate whether and how strongly a pair of variables are related, such as a particular aspect of a web page or the position of the page on Google's search engine results page. “
Most of the statistical work revolves around measuring correlation, even when predicting the future,” Jen said carefully. "But it's very difficult to prove a causal relationship." A causal relationship is the act of causing something to happen. That's the real reason things work that way. "I suspect that there is a significant amount of confirmation bias without knowing the details of how these companies are creating indicators," Jen continued. Confirmation bias occurs when the person performing the analysis wants to prove a pre-determined assumption. Instead of doing the actual work needed to confirm the hypothesis, fit the data until this assumption is proved. To give Jen a better idea of how these companies are creating data, I've shared some of the more popular