Appraisers' Free Forum

Deleted

Deleted

Pina Colada wrote:

Now what is srong with that R^2? How big should it be before it is OK for appraiser use?

there are no arbitrary benchmarks.

Are there any un-arbitrary (sic) benchmarks? You guys seem to see significance or not in these R^2s based on something. That is what I am looking for. I want to know that.

I separated out those sales that were above the regression line and located on upper floors - the ones you were complaining about. The R2 for the remaining sales in 80% and for the separated-out sales its 66%. (See attachment). So, whatever it is that is making those sales stand out, both subsets show a relationship between floor and price.

So now what? Do we adjust for floor or not based on this stuff, and if so, how much? The amount in the 66% relationship or in the 80% or something in between?

Edd,
Sometimes you'll look at a scatterplot and see two or more separate groups of data, as is shown in PC's earlier document. That kind of grouping indicates that something else is going on aside from the single element you're testing.

In the data we're looking at in these examples, clearly the floor causes an increase in price--regardless of what else is happening. There's a separate submarket involving some other quality, but we don't know yet what that might be.

From only the linear regression in the second graph of the first document, we cannot predict a price because the scatter is too great (R^2 of 5%). But we can say that prices increase with floor height. In the latter document, floor height varies about the same whether you're looking at the lower group or the upper group. Other data sets may be different; the upper group's regression line may slope more than the lower group's, or vice-versa.

In this case, we could say that since both groups' regression lines have about the same slope, then the adjustment for floor level would be the same regardless of which group the subject exemplifies. As a bonus, you've been able to tell that there's some other quality that needs to be addressed in your report. Maybe it's a better view (oceanside vs. street side); perhaps proximity to the elevator, fire exits, or access to the parking garage. Or maybe that high group is just a better-quality condo. Whatever it is, you have to figure out what's causing the price disparity.

Thanks

I hope I never get reviewed by any of you guys.

mr rex wrote:
We have a 4 story furniture store downtown. The Fire Department has a snorkel truck just in case. They often drive the snorkel by the furniture store, salivating and trying to conceal their woodies.

Deleted

Pina Colada wrote:

From only the linear regression in the second graph of the first document, we cannot predict a price because the scatter is too great (R^2 of 5%).

I did not create the graph to "predict a price." Floor number is not the only variable. Fixating on the spread of the points, that are spread out for several reasons, is a good way ot overlook at how they are all moving in the same direction.

If you don't think that's how to do, what's the Plante recipe? Make up a theory based on how someone's grandmother doesn't want to be on a high floor?

Well, I guess its interesting to see the points are all moving in the same direction, which direction from the context of this thread is otherwise termed price. So the chart tells us that the higher we go the higher price. Why is that not predictable even though you may not have had prediction in mind? Incidentially, if you weren't working with a price prediction in mind, what were you doing?

Incidentially, if you weren't working with a price prediction in mind, what were you doing?

Isn't there a difference between actual price prediction and working with something that could be used for price prediction. Just because I am trying, very badly obviously, to explain how a spark plug wrench works, doesn't mean I am doing a tune-up.

So the chart tells us that the higher we go the higher price.

I don't think any of the charts indicated that. And without searching the posts, I am quite sure I didn't write that.

I think maybe I should just recant everything I posted, agree that there is no human way possible to discern if floor differences affect prices in high-rise buildings (even when it is staring you right in the face), observe that this is just an "art," that we are all "subjective," that markets are crazy, that the R[sup]2[/sup] is too low, and return to concocting PFA theories of market psychology based on Aunt Gertrude's acrophobia.

Pina Colada wrote:I think maybe I should just recant everything I posted, agree that there is no human way possible to discern if floor differences affect prices in high-rise buildings (even when it is staring you right in the face), observe that this is just an "art," that we are all "subjective," that markets are crazy, that the R[sup]2[/sup] is too low, and return to concocting PFA theories of market psychology based on Aunt Gertrude's acrophobia.

Yeah right and then hell will freeze over right away.

Isn't there a difference between actual price prediction and working with something that could be used for price prediction. Just because I am trying, very badly obviously, to explain how a spark plug wrench works, doesn't mean I am doing a tune-up.

I didn't see the tool, I just saw what it did to your chart and the chart looked all tuned up to me. Did you make all of this stuff up? We've been hypotheticalized , folks.

Yeah right and then hell will freeze over right away.

as part of global warming.

Did you make all of this stuff up?

Every word of it.

Has anyone ever consistently been able to prove a floor level adjustment in a condo?

No one.