page 343 Table 13.1 Regression of estimated 1980 U.S. census undercount of area characteristics, for 66 central cities, state remainders, and states.
GET FILE='D:ericksen.sav'. regression /statistics=defaults tol /dep=undcount /method=enter perc_min crimrate poverty diffeng hsgrad housing city countprc.
Model  Variables Entered  Variables Removed  Method 

1  Percentage of households counted by conventional personal enumeration, Percentage of housing in small, multiunit buildings, Percentage having difficulty speaking or writing English, Percentage poor, Rate of serious crimes per 1000 population, City 1=yes, 0=no, Percentage age 25 or older who had not finished high school, Percentage black or Hispanic(a)  .  Enter 
a All requested variables entered.  
b Dependent Variable: Preliminary estimate of percentage undercount 
Model  R  R Square  Adjusted R Square  Std. Error of the Estimate 

1  .841(a)  .708  .667  1.42647 
a Predictors: (Constant), Percentage of households counted by conventional personal enumeration, Percentage of housing in small, multiunit buildings, Percentage having difficulty speaking or writing English, Percentage poor, Rate of serious crimes per 1000 population, City 1=yes, 0=no, Percentage age 25 or older who had not finished highschool, Percentage black or Hispanic 
Model  Sum of Squares  df  Mean Square  F  Sig.  

1  Regression  280.795  8  35.099  17.249  .000(a) 
Residual  115.985  57  2.035  


Total  396.780  65  



a Predictors: (Constant), Percentage of households counted by conventional personal enumeration, Percentage of housing in small, multiunit buildings, Percentage having difficulty speaking or writing English, Percentage poor, Rate of serious crimes per 1000 population, City 1=yes, 0=no, Percentage age 25 or older who had not finished highschool, Percentage black or Hispanic  
b Dependent Variable: Preliminary estimate of percentage undercount 

Unstandardized Coefficients  Standardized Coefficients  t  Sig.  Collinearity Statistics  

Model  B  Std. Error  Beta  Tolerance  VIF  
1  (Constant)  1.771  1.382  
1.282  .205  

Percentage black or Hispanic  7.983E02  .023  .566  3.531  .001  .200  5.009  
Rate of serious crimes per 1000 population  3.012E02  .013  .303  2.317  .024  .299  3.344  
Percentage poor  .178  .085  .324  2.101  .040  .216  4.625  
Percentage having difficulty speaking or writing English  .215  .092  .214  2.333  .023  .611  1.636  
Percentage age 25 or older who had not finished high school  6.129E02  .045  .211  1.369  .176  .216  4.619  
Percentage of housing in small, multiunit buildings  3.496E02  .025  .139  1.419  .161  .534  1.872  
City 1=yes, 0=no  1.160  .771  .203  1.505  .138  .283  3.538  
Percentage of households counted by conventional personal enumeration  3.699E02  .009  .372  3.997  .000  .591  1.691  
a Dependent Variable: Preliminary estimate of percentage undercount 
page 343 Table 13.2 Correlations among eight predictors of the 1980 U.S. census undercount.
CORRELATIONS /VARIABLES=perc_min crimrate poverty diffeng hsgrad housing city countprc.

Percentage black or Hispanic  Rate of serious crimes per 1000 population  Percentage poor  Percentage having difficulty speaking or writing English  Percentage age 25 or older who had not finished highschool  Percentage of housing in small, multiunit buildings  City 1=yes, 0=no  Percentage of households counted by conventional personal enumeration  

Percentage black or Hispanic  Pearson Correlation  1  .655  .738  .395  .535  .357  .758  .334 
Sig. (2tailed)  .  .000  .000  .001  .000  .003  .000  .006  
N  66  66  66  66  66  66  66  66  
Rate of serious crimes per 1000 population  Pearson Correlation  .655  1  .369  .512  .067  .532  .729  .233 
Sig. (2tailed)  .000  .  .002  .000  .595  .000  .000  .060  
N  66  66  66  66  66  66  66  66  
Percentage poor  Pearson Correlation  .738  .369  1  .152  .751  .335  .538  .157 
Sig. (2tailed)  .000  .002  .  .224  .000  .006  .000  .208  
N  66  66  66  66  66  66  66  66  
Percentage having difficulty speaking or writing English  Pearson Correlation  .395  .512  .152  1  .116  .340  .480  .108 
Sig. (2tailed)  .001  .000  .224  .  .352  .005  .000  .387  
N  66  66  66  66  66  66  66  66  
Percentage age 25 or older who had not finished high school  Pearson Correlation  .535  .067  .751  .116  1  .235  .315  .414 
Sig. (2tailed)  .000  .595  .000  .352  .  .058  .010  .001  
N  66  66  66  66  66  66  66  66  
Percentage of housing in small, multiunit buildings  Pearson Correlation  .357  .532  .335  .340  .235  1  .566  .086 
Sig. (2tailed)  .003  .000  .006  .005  .058  .  .000  .491  
N  66  66  66  66  66  66  66  66  
City 1=yes, 0=no  Pearson Correlation  .758  .729  .538  .480  .315  .566  1  .269 
Sig. (2tailed)  .000  .000  .000  .000  .010  .000  .  .029  
N  66  66  66  66  66  66  66  66  
Percentage of households counted by conventional personal enumeration  Pearson Correlation  .334  .233  .157  .108  .414  .086  .269  1 
Sig. (2tailed)  .006  .060  .208  .387  .001  .491  .029  .  
N  66  66  66  66  66  66  66  66 
page 355 Table 13.3 B. Fox's Canadian women's labor force participation data. T is year; L is women's labor force participation rate, in percent; F is the total fertility rate, per 1000; M is men's average weekly wages in 1935 dollars; W is women's average weekly wages; D is percapita consumer debt; and P is the percentage of parttime workers.
GET FILE='D:bfox.sav'. list year womwork fertil mwage fwage debt parttime.YEAR WOMWORK FERTIL MWAGE FWAGE DEBT PARTTIME
1946 25.30 3748 25.35 14.05 18.18 10.28 1947 24.40 3996 26.14 14.61 28.33 9.28 1948 24.20 3725 25.11 14.23 30.55 9.51 1949 24.20 3750 25.45 14.61 35.81 8.87 1950 23.70 3669 26.79 15.26 38.39 8.54 1951 24.20 3682 26.33 14.58 26.52 8.84 1952 24.10 3845 27.89 15.66 45.65 8.60 1953 23.80 3905 29.15 16.30 52.99 5.49 1954 23.60 4047 29.52 16.57 54.84 6.67 1955 24.30 4043 32.05 17.99 65.53 6.25 1956 25.10 4092 32.98 18.33 72.56 6.32 1957 26.20 4168 32.25 17.64 69.49 7.30 1958 26.60 4073 32.52 18.16 71.71 8.65 1959 26.90 4100 33.95 18.58 78.89 8.80 1960 27.90 4119 34.63 18.95 84.99 9.39 1961 29.10 4159 35.14 18.78 87.71 10.23 1962 29.90 4134 34.49 18.74 95.31 10.77 1963 29.80 4017 35.99 19.71 104.40 10.84 1964 30.90 3886 36.68 20.06 116.80 11.70 1965 32.10 3467 37.96 20.94 130.99 12.33 1966 33.20 3150 38.68 21.20 135.25 12.18 1967 34.50 2879 39.65 21.95 142.93 13.67 1968 35.10 2681 41.20 22.68 155.47 13.82 1969 36.10 2563 42.44 23.75 165.04 14.91 1970 36.90 2571 42.02 25.63 164.53 15.52 1971 37.00 2503 45.32 26.79 169.63 15.47 1972 37.90 2302 45.61 27.51 190.62 15.85 1973 40.10 2931 45.59 27.35 209.60 15.40 1974 40.60 1875 48.06 29.64 216.66 16.23 1975 42.20 1866 46.12 29.33 224.34 16.71
Number of cases read: 30 Number of cases listed: 30
page 358 Figure 13.6 Plot of C(p)p against p for the census undercount regression. Only subsets for which C(p)p < 10 are shown. The following capitalized letter are employed to label the predictors in each subset: Minority, Crime, Poverty, Language, High school, hOusing, cIty, and coNventaional. Ericksen, et al. (1989) selected the predictor subset MCN . (i.e., Minority, Crime and coNventaional).
NOTE: We do not know how to make this table in SPSS.
page 359 Table 13.4 "Best" subset regression models for Ericksen et. al.'s census undercount data. Coefficient standard errors are in parentheses.
GET FILE='D:ericksen.sav'.regression /statistics=defaults selection /dep=undcount /method=enter perc_min crimrate countprc /method=enter diffeng /method=enter poverty /method=enter hsgrad housing city.
Model  Variables Entered  Variables Removed  Method 

1  Percentage of households counted by conventional personal enumeration, Rate of serious crimes per 1000 population, Percentage black or Hispanic(a)  .  Enter 
2  Percentage having difficulty speaking or writing English(a)  .  Enter 
3  Percentage poor(a)  .  Enter 
4  Percentage of housing in small, multiunit buildings, City 1=yes, 0=no, Percentage age 25 or older who had not finished high school(a)  .  Enter 
a All requested variables entered.  
b Dependent Variable: Preliminary estimate of percentage undercount 

R  R Square  Adjusted R Square  Std. Error of the Estimate  Selection Criteria  

Model  Akaike Information Criterion  Amemiya Prediction Criterion  Mallows' Prediction Criterion  Schwarz Bayesian Criterion  
1  .798(a)  .638  .620  1.52302  59.405  .409  12.676  68.164 
2  .818(b)  .669  .647  1.46699  55.385  .385  8.515  66.333 
3  .828(c)  .686  .659  1.44207  54.032  .377  7.320  67.170 
4  .841(d)  .708  .667  1.42647  55.211  .385  9.000  74.918 
a Predictors: (Constant), Percentage of households counted by conventional personal enumeration, Rate of serious crimes per 1000 population, Percentage black or Hispanic  
b Predictors: (Constant), Percentage of households counted by conventional personal enumeration, Rate of serious crimes per 1000 population, Percentage black or Hispanic, Percentage having difficulty speaking or writing English  
c Predictors: (Constant), Percentage of households counted by conventional personal enumeration, Rate of serious crimes per 1000 population, Percentage black or Hispanic, Percentage having difficulty speaking or writing English, Percentage poor  
d Predictors: (Constant), Percentage of households counted by conventional personal enumeration, Rate of serious crimes per 1000 population, Percentage black or Hispanic, Percentage having difficulty speaking or writing English, Percentage poor, Percentage of housing in small, multiunit buildings, City 1=yes, 0=no, Percentage age 25 or older who had not finished high school 
Model  Sum of Squares  df  Mean Square  F  Sig.  

1  Regression  252.966  3  84.322  36.352  .000(a) 
Residual  143.814  62  2.320  


Total  396.780  65  



2  Regression  265.504  4  66.376  30.843  .000(b) 
Residual  131.276  61  2.152  


Total  396.780  65  



3  Regression  272.006  5  54.401  26.160  .000(c) 
Residual  124.774  60  2.080  


Total  396.780  65  



4  Regression  280.795  8  35.099  17.249  .000(d) 
Residual  115.985  57  2.035  


Total  396.780  65  



a Predictors: (Constant), Percentage of households counted by conventional personal enumeration, Rate of serious crimes per 1000 population, Percentage black or Hispanic  
b Predictors: (Constant), Percentage of households counted by conventional personal enumeration, Rate of serious crimes per 1000 population, Percentage black or Hispanic, Percentage having difficulty speaking or writing English  
c Predictors: (Constant), Percentage of households counted by conventional personal enumeration, Rate of serious crimes per 1000 population, Percentage black or Hispanic, Percentage having difficulty speaking or writing English, Percentage poor  
d Predictors: (Constant), Percentage of households counted by conventional personal enumeration, Rate of serious crimes per 1000 population, Percentage black or Hispanic, Percentage having difficulty speaking or writing English, Percentage poor, Percentage of housing in small, multiunit buildings, City 1=yes, 0=no, Percentage age 25 or older who had not finished high school  
e Dependent Variable: Preliminary estimate of percentage undercount 

Unstandardized Coefficients  Standardized Coefficients  t  Sig.  

Model  B  Std. Error  Beta  
1  (Constant)  2.224  .561  
3.965  .000 
Percentage black or Hispanic  7.861E02  .015  .557  5.337  .000  
Rate of serious crimes per 1000 population  3.630E02  .010  .366  3.614  .001  
Percentage of households counted by conventional personal enumeration  2.800E02  .008  .282  3.473  .001  
2  (Constant)  1.976  .550  
3.592  .001 
Percentage black or Hispanic  7.519E02  .014  .533  5.273  .000  
Rate of serious crimes per 1000 population  2.715E02  .010  .274  2.613  .011  
Percentage of households counted by conventional personal enumeration  2.730E02  .008  .275  3.512  .001  
Percentage having difficulty speaking or writing English  .209  .087  .208  2.414  .019  
3  (Constant)  .793  .860  
.921  .361 
Percentage black or Hispanic  .101  .020  .716  4.992  .000  
Rate of serious crimes per 1000 population  2.435E02  .010  .245  2.355  .022  
Percentage of households counted by conventional personal enumeration  2.933E02  .008  .295  3.796  .000  
Percentage having difficulty speaking or writing English  .184  .086  .183  2.126  .038  
Percentage poor  .110  .062  .200  1.768  .082  
4  (Constant)  1.771  1.382  
1.282  .205 
Percentage black or Hispanic  7.983E02  .023  .566  3.531  .001  
Rate of serious crimes per 1000 population  3.012E02  .013  .303  2.317  .024  
Percentage of households counted by conventional personal enumeration  3.699E02  .009  .372  3.997  .000  
Percentage having difficulty speaking or writing English  .215  .092  .214  2.333  .023  
Percentage poor  .178  .085  .324  2.101  .040  
Percentage age 25 or older who had not finished high school  6.129E02  .045  .211  1.369  .176  
Percentage of housing in small, multiunit buildings  3.496E02  .025  .139  1.419  .161  
City 1=yes, 0=no  1.160  .771  .203  1.505  .138  
a Dependent Variable: Preliminary estimate of percentage undercount 

Beta In  t  Sig.  Partial Correlation  Collinearity Statistics  

Model  Tolerance  
1  Percentage having difficulty speaking or writing English  .208(a)  2.414  .019  .295  .731 
Percentage poor  .240(a)  2.093  .040  .259  .423  
Percentage age 25 or older who had not finished high school  .124(a)  1.158  .251  .147  .506  
Percentage of housing in small, multiunit buildings  .068(a)  .747  .458  .095  .715  
City 1=yes, 0=no  .169(a)  1.280  .206  .162  .331  
2  Percentage poor  .200(b)  1.768  .082  .223  .411 
Percentage age 25 or older who had not finished high school  .049(b)  .449  .655  .058  .455  
Percentage of housing in small, multiunit buildings  .088(b)  1.002  .320  .128  .709  
City 1=yes, 0=no  .123(b)  .949  .346  .122  .323  
3  Percentage age 25 or older who had not finished high school  .154(c)  1.044  .301  .135  .239 
Percentage of housing in small, multiunit buildings  .053(c)  .591  .557  .077  .664  
City 1=yes, 0=no  .146(c)  1.147  .256  .148  .320  
a Predictors in the Model: (Constant), Percentage of households counted by conventional personal enumeration, Rate of serious crimes per 1000 population, Percentage black or Hispanic  
b Predictors in the Model: (Constant), Percentage of households counted by conventional personal enumeration, Rate of serious crimes per 1000 population, Percentage black or Hispanic, Percentage having difficulty speaking or writing English  
c Predictors in the Model: (Constant), Percentage of households counted by conventional personal enumeration, Rate of serious crimes per 1000 population, Percentage black or Hispanic, Percentage having difficulty speaking or writing English, Percentage poor  
d Dependent Variable: Preliminary estimate of percentage undercount 