My guess is that Google, like Facebook, have a good impression of your CANOE/Big 5 personality type: Conscientiousness, Agreeableness, Neuroticism, Openness, Extroversion. And in that they have an extremely good assessment of your likelihood to repay loans.
I understand that this has been at the heart of the Cambridge Analytica political manipulation scandal. The basic methodology is take a large sample of people whose big 5 traits you know, and look at their browsing history. Through this you create a model: people visiting x websites are statistically more likely to be y personality. Then when they look at your patterns of web use, they are able to assign you a personality type. Or more specifically group you with others likely to be of that personality type. Of course this assignment may be incorrect, but of course over a large number of people using this methodology, they are more likely to be right than wrong across most people. And although I’m not a statistician, this is the power of big data.
Ok, I disagree here. The data analysis done so far describes what you look at, but not why. AFAIK CA identified which party you were likely to vote for, and what concerns (gun control, Clinton Foundation, Islam, etc) concerned you. Mostly just classify the topic of links people post.
There's a big jump to determining your levels of Conscientiousness, Agreeableness, Neuroticism, Openness, Extroversion. Do people post articles that imply a negative agreeableness? I don't know -- political articles are typically intended to persuade, so you can safely argue that the poster believes in the article they post. But I don't know what kind of assumptions you can make about someone's facebook posts -- much less their browsing / purchasing behavior -- to determine aspects of their personality.
Your disagreement here, as far as I understand it, is the central point of disagreement in the debate around the impact of the FB/Google on behavior. On one side, you have people saying that internet use can be indicative of personality type. Others say no: because they can’t discern the reason behind why you do what you do. I have done some statistics, multiple regression and the like a long time ago, but I’m not entirely expert enough to be fully confident in my conclusions, but look at it this way: (1) you may visit a gun site, not because you are interested in guns, but because you are doing market research for your employer or because you accidentally clicked on a link, (2) however 90% of people clicked on the link because they like guns, and that says something about their personality. Now of course 90% is not perfect, but the predictive power at large scale of that insight is huge.
What does looking at guns say about someone's personality? That's what I don't get. Are they fearful after a home invasion? Do they like marksmanship? Are they looking for a gift for someone? Are their friends into shooting, and they're looking for ways to hang out more with their friends? How do you know why? And if you did, does that tell you about their personality?
There's a large jump there between an interest and "Conscientiousness, Agreeableness, Neuroticism, Openness, Extroversion." I don't think any amount of "throw big data and other buzzword at it" will reduce that jump, because you're more likely to train against other attributes of your population than their personality.
