$ per Square Foot Significance

[slacker]

You will be amazed at how few adjustments are really necessary if you use the correct sequence of adjustments, assuming you did a good job of comp selection.

Austin,

If you need to use a graph to explain your adjuments, maybe you should be using different comps to begin with.

I like the idea of defining and supporting your adjustments graphically but sure seems like a lot of work / time. Could you e-mail me a sample spreadsheet, I'd like to see it.

Thanks

 

[Austin]

Slacker: You don't use graphs in the way you just stated. You use them to guide yourself through the correct sequence of adjustments like a road map. For example, when you begin an appraisal after selecting the best comps, you have no idea what is influencing price. You might think you do, but you don't. The object of the game is to find what is influencing price and what is not and my system does just that. The secret is to not go into an appraisal with preconcieved ideas nor can you import data. I have had McKissocks class but nothing I do is remotely related to his methods. I have read a lot of books on the subject and my opinion is that there are a lot of people that know the math but they know nothing about appraisal theory, and the appraisal theory they need to know is not in circulation as of this time. An appraiser that writes for the NAIFA or whatever their correct name is, has written two articles on regression and is writing a third. He sent me a copy of his first two articles for review and his presentation of the regression method is the best I have read. I can't recall his name but have it at the home computer. If you go over there to that forum and check the index you will find it.

There are a lot of comments above that I would like to respond to especially about regression. Based on my experience with regression, the reason no body has been successful at it is for the reason just stated above and the same reason AVM won't work in all situations. That is, you can't go into a problem with 16 variables that you just know affect price. You have to use regression as a tool to extract value influencing factors in a stepwise progression, otherwide you just confuse the situation by screwing up the test statistics. If you load up the regression with value factors that do not affect price, the program will take normal random variance and allocate it to these factors. Test statistsics are not a measure of the accuracy of the regression for this reason. You can have a perfect regression equation that predicts prices perfectly and have terrible test statistics.

I will be glad to give you guys any information I have, but my dad is 80 years old and is having heart surgery next Wednesday so I won't be around the forum much the next week or so. I have to finish my work fast and will be behind when I get back. Wish us luck.
Austin

 

[Farm Gal]

Ed

Programs like MPSS (?) and SPPS (the one I worked on) used to require planning and top-notch University computer systems... yoru desktop can now do this sort of anyalysis. The programs are somewhat expensive - $1,500 +/- but the time and energy to learn how to run them is the missing link.

I am quite convinced that the resultant anyalysis would be VERY informative, however een what Austin is doing took a lot of tie and effort to learn how to do it as he said RIGHT... THAT is the part I am willing to pay little OR big money for: the thousands (if I had to guess) of hours he has spent 'fiddling' to achieve an understanding of what 'really' matters even if it is specific to his area: as in SOME of it will carry over generally!.

Then taking that and the power of one of these more complex programs - WOW.

someday, someday soon.

 

[Fred]

"An appraiser that writes for the NAIFA or whatever their correct name is, has written two articles on regression and is writing a third. He sent me a copy of his first two articles for review and his presentation of the regression method is the best I have read. I can't recall his name "

Douglas G. Smith, IFAS; from SW Montana

 

[slacker]

For example, when you begin an appraisal after selecting the best comps, you have no idea what is influencing price.

And neither does your graph or regression. There's no doubt that you can make conclusions to specific levels confidence but guess what? You never know for sure. I have never seen a regression based on a 100% confidence level.

I'm sure you could build a regression showing that weather conditions at the time of the contract signing have an influence on price as well. Maybe that should be a new line on the comp grid??

And if you don't know for sure, you need to rely on your experience and knowledge of current market conditions.

I think regression is a great tool for supporting conclusions from a specific perspective but there's no way can prove anything 100% with them. Unfortunately, regression lives side by side with experience in their own little "Artsy-Fartsy" world.


By manipulating the right numbers you can prove anything. Does anyone remember seeing this shortly after 9/11?

The date of the attack: 9/11 - 9 + 1 + 1 = 11
September 11th is the 254th day of the year: 2 + 5 + 4 = 11
After September 11th there are 111 days left to the end of the year.
119 is the area code to Iraq/Iran. 1 + 1 + 9 = 11
Twin Towers - standing side by side, looks like the number 11
The first plane to hit the towers was Flight 11
But ....There's More.......
State of New York - The 11th State added to the Union
New York City - 11 Letters
Afghanistan - 11 Letters
The Pentagon - 11 Letters
Ramzi Yousef - 11 Letters (convicted of orchestrating the attack on the WTC in 1993)
Flight 11 - 92 on board - 9 + 2 = 11
Flight 77 - 65 on board - 6 + 5 = 11

The numbers all add up so there must be something to this right? WRONG!!! It's just a crafty way of manipulating some numbers.

No mathematical model could ever place a number on "Oh Honey!! It's just what I've always wanted!! I've just got to have it!!"

And I still think your graph's are an interesting angle and would like to see one in action.

"Get your facts first, and then you can distort them as much as you please. (Facts are stubborn, but statistics are more pliable.)
Mark Twain (1835 - 1910)"

 

[Austin]

slacker: Stick to your voodoo methods or astrology or whatever you use.
The following articles are not for you slacker. No pictures or crossword puzzles.

by Douglas G. Smith, IFAS
Is There a Blue Box AVM In Your Future?

http://www.naifa.com/gram/2002apr/smith-apr02.html

Multiple Regression Analysis - An Appraiser's Perspective
Part II

http://naifa.com/gram/2002jul/smith-jul02.html

These are the articles on regression I mentioned in an above post. If you are interested, check them out, they are good. Slacker and his cronies can go over to the Watercooler and work cross word puzzles or get into a spitting contest with Gregg Goodpasture.

 

[TEL2002]

Pam,

Try asking the REAgent if he/she had first reduced the $ per square foot by the cost of the larger concrete driveway or the cost of the added landscaping, or the larger deck etc. Maybe this will enlighten him that all $ per square foot are not created equally.

Austin,

I am going to print out your comments and pass on to my partner. He just completed a Multiple Regression Analysis course out in Denver, and I'll bet we would like to chat with you. We are now pricing the purchase of SPSS so we can do our own 'number crunching'. (Hoping my daughter can buy it at her college at a drastically reduced rate.)

You were right on.

 

[Farm Gal]

At the risk of segregating the forum even further :roll: ,

I am beginning to think there ought to be a seperate forum for those of us who want to know EVERYTHING about Multi-LinearRegression, Linear Regression as run through Excell or other spreadsheets and 'better analyis through numbers' in general.

Those who want to know and seriously focus on this matter (like me) would appreciate it! Those who think it another form of 'statistical manipulation for the purpose of proving what you want to prove' aka 'lies, damn lies, and statistics' can skip the threads...

Any thoughts?

 

[EDWARD BERRY]

Lee Ann,

That is a good sugestion.

I took a stat course but there was only about 1 chapter on MR. it was not enough for me to work any practical(real estate) problems.

There are several public domain programs that would proably work for us.

Perhaps we could all use one set of residential data, use the same MR program and work thru it . Maybe some guru either from our industry or an instructor from a Univ. could monotor our progress until we were able to work it on our own.

We coul use screen dumps, e-mail, and follow each participant's work until we got it correct.

Again, thanks-ed in arkansas

 

[Farm Gal]

Anyone else have any thoughts?

perhaps we can start out in the Technical Forum... rather than a whole new one...

(I slept on it :lol: naps do wonders for my thinking functions.... :wink: )