Page 11 - Regression Guideline
P. 11
An"Example"of"SAVVI’s"Regression" Modeling"Steps"
Step&3:&Bivariate&Linear&Regression&
&
Once"a"set"of"properJes"is"idenJfied"for"the"analysis,"the"outliers"removed,"and"a"Jme" adjustment"has"been"calculated,"the"third"step"in"the"SAVVI"regression"approach"is"to"run"a"set" of"bivariate"linear"regressions"(see"the"sidebar"explaining"linear"regression)."This"analysis" examines"the"effect"of"each"property"characterisJc"individually"without"controlling"for"the" effects"of"any"other"property"characterisJc."Shown"in"the"table"below"are"sample"results"that" might"appear"in"an"actual"report."
""
The"results"show"the"coefficient"that"is"the"best"predictor"of"sale"price"for"each"variable,"the" percentage"of"the"sale"price"accounted"for"by"that"variable"individually,"and"whether"the" esJmate"is"staJsJcally"significant"with"at"a"95%"(*)"or"99%"(**)"level"(shown"in"bold)."Variables" that"are"not"staJsJcally"significant"at"least"at"the"95%"threshold"do"not"reliably"account"for" differences"in"the"sales"prices"of"the"properJes"examined."Results"are"ordered"from"largest"to" smallest"by"the"percentage"of"price"variaJon"accounted"for"by"each"property"characterisJc."In" the"model"for"main"square"feet,"for"example,"this"property"characterisJc"accounts"for"about" 37%"of"the"price"variaJon"among"properJes;"garage"spaces"account"for"about"26%,"and"so"on." The"coefficient"for"main"square"feet"means"the"esJmated"price"of"the"property"increases" $172.14"for"every"addiJonal"square"foot"while"the"price"increases"$65,832.95"for"a"garage" space."A"negaJve"coefficient"such"as"for"Style(SL)"means"the"price"of"a"property"with"that" characterisJc"decreases."
10"