Like him or not

[alebrewer]

It's ok to try to understand it. But when you are in a position of educator and positioning yourself as an expert, you should be clear about what you know and what you don't. Just because AI gives you the ability to do an analysis, doesn't mean the conclusions you draw are credible. Competence matters.

Here, the MAIs are chalking it up to homeowner intelligence. I felt the need to black his name out because it is just an unbelievably daft comment. Why are apples more expensive than oranges? Because people who buy apples are smarter!
View attachment 98952

Well, you failed in your attempt to black out his name...

That said, what would you have expected Jim to say in a professional forum? That George was/is daft? He's not that tactless.

PS - I don't disagree with you RE the 'position of authority' argument. But for this post, I don't see where he's providing misleading information? What is it you think is wrong with his hypotheses RE this particular post of his?

 

[CGinMN]

Could one impact the other, and how was it considered in the analysis? One of many problems.

 

[alebrewer]

Yet another question to be considered for sure. Which is why, I think, it takes a lot of hutzpah to put your analyses out there for others to criticize. I don't know the guy and I'm not associated with the AI, so I personally have no opinion of Jim, other than that he's trying to use AI and data analysis to further our profession. Just got through with GD's class on Stats and r - and Jim came in and devoted several hours of his time to helping the class. I thought it was a pretty nice gesture.

 

[TerryRohrer]

I think this analysis is on the same level, with equal veracity in the results, as the "Research Note" from Freddie Mac that created and promoted a biased appraiser narrative (albeit without the obvious political agenda).

 

[CGinMN]

In his post, he is stating that it relates to things like appraiser incentives to value refis higher to make loans work, appraiser bias and objectivity concerns, etc. This is done without even attempting to understand the nature of the data or control for obvious differences. I just think it is irresponsible.

 

[alebrewer]

For sure he proffers several ideas that might serve as hypotheses for the 'surprising trends' (to use his words), but I didn't notice where he stated them as causal? I'd suspect lax refinance underwriting might be a culprit as well. I'm not sure how you'd test that hypothesis, though.

 

[jay trotta]

IMO- value differences depend on "pressure" from those who need the deal. IE: 5% not significant / 20-30% Huge and should be an indicator something is not right, whether in data form or otherwise.

 

[CGinMN]

For sure he proffers several ideas that might serve as hypotheses for the 'surprising trends' (to use his words), but I didn't notice where he stated them as causal?

This is why, as George likes to say, problem identification comes first.

 

[alebrewer]

This is why, as George likes to say, problem identification comes first.

for appraisals - yes. For data analysis, sometimes the data exposes a problem to be identified.

 

[djd09]

have you ever seen jim post here....