4# bobguy
Becasue the sample size is so small, it's hard to get accurate result.
So I classify the income into two classes: high, y=1 (income >=40000) and low y=0 (income <40000).
and then run logistic regression. I get the result as below: I think it's the only way.
The LOGISTIC Procedure
Model Information
Data Set _PROJ_.PALMORG
Response Variable Ownership Ownership
Number of Response Levels 2
Number of Observations 20
Model binary logit
Optimization Technique Fisher's scoring
Response Profile
Ordered Total
Value Ownership Frequency
1 1 7
2 0 13
Probability modeled is Ownership=1.
Class Level Information
Design
Class Value Variables
income_class 0 1
1 -1
Model Convergence Status
Convergence criterion (GCONV=1E-8) satisfied.
Model Fit Statistics
Intercept
Intercept and
Criterion Only Covariates
AIC 27.898 23.962
SC 28.894 25.953
-2 Log L 25.898 19.962
Testing Global Null Hypothesis: BETA=0
Test Chi-Square DF Pr > ChiSq
Likelihood Ratio 5.9360 1 0.0148
Score 5.4945 1 0.0191
Wald 4.4339 1 0.0352
1 10:09 Sunday, February 23, 2003 2
The LOGISTIC Procedure
Type 3 Analysis of Effects
Wald
Effect DF Chi-Square Pr > ChiSq
income_class 1 4.4339 0.0352
Analysis of Maximum Likelihood Estimates
Standard Wald
Parameter DF Estimate Error Chi-Square Pr > ChiSq
Intercept 1 -0.8959 0.6180 2.1014 0.1472
income_class 0 1 -1.3013 0.6180 4.4339 0.0352
Odds Ratio Estimates
Point 95% Wald
Effect Estimate Confidence Limits
income_class 0 vs 1 0.074 0.007 0.835
Association of Predicted Probabilities and Observed Responses
Percent Concordant 59.3 Somers' D 0.549
Percent Discordant 4.4 Gamma 0.862
Percent Tied 36.3 Tau-a 0.263
Pairs 91 c 0.775