5.1 Basic Estimation Techniques The relationships we theoretically develop in the text can be estimated statistically using regression analysis, Regression analysis is a method used to determine the coefficients of a a functional relationship. For example, if demand is P = a+bQ We need to estimate a and b.
Basic Estimation Techniques. The relationships we theoretically develop in the text can be estimated statistically using regression analysis, Regression analysis is a method used to determine the coefficients of a a functional relationship. For example, if demand is P = a+bQ - PowerPoint PPT Presentation
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
5.1
Basic Estimation Techniques The relationships we theoretically develop
in the text can be estimated statistically using regression analysis,
Regression analysis is a method used to determine the coefficients of a a functional relationship.
For example, if demand is P = a+bQ
We need to estimate a and b.
5.2
Ordinary Least Squares(OLS) Means to determine regression equation that
“best” fits data Goal is to select the line(proper intercept &
slope) that minimizes the sum of the squared vertical deviations
Minimize ei2 which is equivalent to
minimizing (Yi -(Y-hat)i)2
5.3
Standard Error of the Estimate Measures variability
about the regression equation
Labeled SEE If SEE = 0 all points
are on line and fit is perfect
)1(2
knei
5.4
Standard Error of the Slope Measures theoretical
variability in estimated slope - different datasets(samples) would yield different slopes
n
i
XX
SEESE
1
21
)()(
5.5
Variability in the Dependent Variable
The sum of squares of Y about its mean value is representative of the total variation in Y
2
1
)( YYTSSn
ii
5.6
Variability in the Dependent Variable
The sum of squares of Y about the regression line(Y-hat) is representative of the “unexplained” or residual variation in Y
n
ii YYRSS
1
2)ˆ(
5.7
Variability in the Dependent Variable
The sum of squares of Y-hat about Y-bar is representative of the “explained” variation in Y
n
ii YYESS
1
2)ˆ(
5.8
Variability in the Dependent Variable
Note, TSS = ESS + RSS If all data points are on the regression line,
RSS=0 and TSS=ESS If the regression line is horizontal, slope =
0, ESS=0 and TSS=RSS The better the fit of the regression line to
the data, the smaller is RSS
5.9
Describing Overall Fit - R2
The coefficient of determination is the ratio of the “explained” sum of squares to the total sum of squares
n
ii
n
ii
YY
e
TSSRSS
TSSESSR
1
2
1
2
2
)(11
5.10
Coefficient of Determination R2 yields the percentage of variability in Y
that is explained by the regression equation It ranges between 0 and 1 What is true if R2 = 1? What is true if R2 = 0?
5.11
Statistical Inference Drawing conclusions about the population
based on sample information. Hypothesis Testing
– which independent variables are significant?– Is the model significant?
Estimation - point versus interval– what is the rate of change in Y per X?– what is the expected value of Y based on X
5.12
Errors in Hypotheses Testing Type I error - rejecting the null hypothesis
when it is true Type II error - accepting the null hypothesis
when it is false Will never eliminate the possibility of error
- but can control their likelihood
5.13
Structuring the Null and Alternative Hypotheses
The null hypothesis is often the reverse of what theory or logic suggest the researcher believes; it is structured to allow the data to contradict it. In the model on the effect of price on quantity demanded, the researcher would expect price to inversely impact amount purchased. Thus, the null might be that price does not effect quantity demanded or it effects it in a positive direction.
5.14
Structuring the Null and Alternative Hypotheses
Model: QA=B0+B1PA+B2Inc+B3PB+
– QA = quantity demanded of good A
– PA = price of good A
– Inc = Income
– PB = price of good B
H0: B1 0
HA: B1 < 0 Law of Demand expectation
5.15
H0 : 1 = 0
Do Not Reject RejectReject
/2/2
5.16
H0 : 1 0
RejectDo Not Reject
5.17
H0 : 1 0
Do Not RejectReject
5.18
The t-Test for the Slope We can test the significance of an
independent variable by testing the following
H0 : k = 0 k = 1,2,….K
HA : k 0
Note if k = 0 a change in the kth independent variable has no impact on Y
5.19
The t-Test for the Slope The test statistic is
)ˆ(
ˆ0
k
Hkk SEt
5.20
T-Test Decision Rule The critical t-value, tc, is the value that
defines the boundary line separating the rejection from the do not reject region.
For a 2-tailed test if |tk| > tc, reject the null; otherwise do not reject
For a 1-tailed test if |tk| > tc and if tc has the sign implied by HA, reject the null; otherwise do not reject
5.21
F-Test and ANOVA F-Test is used to test the overall
significance of the regression or model Analysis of Variance = ANOVA ANOVA is based on the components of the
variation in Y previously discussed - TSS, ESS, and RSS
5.22
ANOVA Table
Source Sum of Sq df Mean Sq
Explain ESS K ESS/K
Residual RSS n-K-1 RSS/(n-K-1)
Total TSS n-1
5.23
F-Statistic
)1/(/
KnRSSKESSF
)1/()(
/)ˆ(2
2
KnYYKYY
Fi
i
5.24
Hypotheses for F-Test
H0: 1= 2=…..= K=0
HA: H0 is not true
Note the null suggests that all slopes are simultaneously zero and that the model would NOT be significant, ie. no independent variables are significant
5.25
Decision Rule for F-Test If F > Fc, reject the null that the model is
insignificant. Note this likely to be good news - your model appears “good”
Otherwise do not reject
5.26
Regression StatisticsMultiple R 0.954779929R Square 0.911604712Adjusted R Square 0.901205266Standard Error 13.29712264Observations 20