HW 4
Part A.
Problem 1.To determine what factors influence public opinion about favoring death penalty for murder:
1. Replicate the SPSS output provided below. Run Logistic regression using the following variables from the GSS 2004: cappun, age, polviews, reborn, sex, religion. Start with identifying the dependent and the independent variables. Draw your model using boxes and a
ows diagram.
2. Give the full logistic regression analysis of the SPSS output. Specifically, interpret the
EXP (B) coefficient (odds) for the significant results. The coding for the dependent variable is presented in the SPSS output. The coding for the other variables is available in the GSS 2004 data set. Go to Utilities – Variables – Click on any variable you need information about. Please read carefully the questions and the possible answers. The variables’ coding will help you to interpret the odds co
ectly. For each variable indicate the level of measurement and the codes for each category of the nominal and ordinal variables.
3. Calculate the probability of favoring death penalty for murder for a catholic man, moderate in political views and without “born again” experience.
4. Calculate the probability of favoring death penalty for murder for a non-believer woman with a “born again” experience and extremely liberal in her political views.
Logistic Regression
Case Processing Summary
Unweighted Casesa
N
Percent
Selected Cases
Included in Analysis
553
39.1
Missing Cases
862
60.9
Total
1415
100.0
Unselected Cases
0
.0
Total
1415
100.0
a. If weight is in effect, see classification table for the total number of cases.
Dependent Variable Encoding
Original Value
Internal Value
oppose
0
favo
1
Pay attention to the dependent variable coding. The interpretation of the dependent variable depends on how the variable is coded!
Categorical Variables Codings
Frequency
Parameter coding
(1)
(2)
(3)
(4)
RS RELIGIOUS PREFERENCE
PROTESTANT
312
.000
.000
.000
.000
CATHOLIC
139
1.000
.000
.000
.000
JEWISH
6
.000
1.000
.000
.000
NONE
91
.000
.000
1.000
.000
OTHER (SPECIFY)
5
.000
.000
.000
1.000
Block 0: Beginning Block
Classification Tablea,
Observed
Predicted
FAVOR OR OPPOSE DEATH PENALTY FOR MURDER
oppose
favo
Percentage Co
ect
Step 0
FAVOR OR OPPOSE DEATH PENALTY FOR MURDER
oppose
0
175
.0
favo
0
378
100.0
Overall Percentage
68.4
a. Constant is included in the model.
b. The cut value is .500
Variables in the Equation
B
S.E.
Wald
df
Sig.
Exp(B)
Step 0
Constant
.770
.091
70.943
1
.000
2.160
Block 1: Method = Ente
Omnibus Tests of Model Coefficients
Chi-square
df
Sig.
Step 1
Step
58.050
8
.000
Block
58.050
8
.000
Model
58.050
8
.000
Model Summary
Step
-2 Log likelihood
Cox & Snell R Square
Nagelkerke R Square
1
632.281a
.100
.140
a. Estimation terminated at iteration number 20 because maximum iterations has been reached. Final solution cannot be found.
Classification Tablea
Observed
Predicted
FAVOR OR OPPOSE DEATH PENALTY FOR MURDER
oppose
favo
Percentage Co
ect
Step 1
FAVOR OR OPPOSE DEATH PENALTY FOR MURDER
oppose
47
128
26.9
favo
35
343
90.7
Overall Percentage
70.5
a. The cut value is .500
Variables in the Equation
B
S.E.
Wald
df
Sig.
Exp(B)
Step 1
age
-.002
.006
.182
1
.670
.998
polviews
.325
.075
18.934
1
.000
1.384
reborn
.704
.227
9.601
1
.002
2.022
sex
-.630
.200
9.919
1
.002
.533
relig
14.505
4
.006
relig(1)
-.485
.242
4.013
1
.045
.616
relig(2)
-.583
.929
.394
1
.530
.558
relig(3)
-1.064
.285
13.978
1
.000
.345
relig(4)
19.549
XXXXXXXXXX
.000
1
.999
3.089E8
Constant
-.346
.658
.276
1
.599
.708
Problem 2. The Exam Practice Problem
To determine what factors influence public opinion about abortion replicate the logistic regression analysis example from above using the following variables from the GSS 2006: abany, age, education, sex, religion. Start with identifying the dependent and the independent variables. Draw your model using boxes and a
ows diagram. Provide coding for the nominal and ordinal variables. Provide the full logistic regression analysis. Specifically, interpret the EXP (B) coefficients (the odds) for the significant results.
PART B
Hugh Crean, A.D. Hightower, and Marjorie Allan (2001) “School-Based Child Care
for Children of Teen parents: Evaluation of an U
an Program Designed to Keep Young
Mothers in School. Educational Evaluation and Program Planning. 24: XXXXXXXXXX.
1. Was there any difference in the ethnic distribution of participating and non-participating mothers? Why do you think so?
2. Was there any difference in their ages when they give birth to their child? How can you prove this?
3. Why logistic regression method was used in this analysis?
4. What were the independent and dependent variables? [Indicate their level of measurement , unit of measurement and/or coding in the tables below]
Variable Name
Level of Measurement
Unit of measurement/coding
Independent/Dependent Variable
5. What were the results? Interpret the impact of each of the five independent variables on graduation using only B coefficient and p (sig). columns in Table 4 on page 272.
Note: The original model is in the log odds, or logit. Therefore, the B coefficient is the effect of a one-unit change in an independent variable on the log odds of graduation. For example (just example, it is not in the table), the b coefficient for age is 0.3. This means that every additional year of age is to increase the log odds of graduation by 0.3
6. Explain the results of Table 5 on page 273.
PII: S XXXXXXXXXX
School-based child care for children of teen parents: evaluation of an
u
an program designed to keep young mothers in school
Hugh F. Creana,*, A.D. Hightowera, Marjorie J. Allan
aDepartment of Clinical and Social Sciences in Psychology, University of Rochester, Rochester, NY, USA
Department of Pediatrics, University of Rochester School of Medicine and Dentistry, Rochester, NY, USA
Abstract
This study examined the effects of the school-based Early Childhood Centers for Children of Teen Parents Program. Designed to keep
young mothers in school, the program provides needed support to u
an young mothers including free on-site child care for their infants and
toddlers, parenting classes, and refe
al to other service agencies. Archived school record information was collected on teen mothers who
participated in the program (n 81) and on a group of teen mothers who had applied for the program but did not receive services (n 89).
Controlling for pre-service differences, participant mothers were found to have better school attendance and deemed to be at lower overall
isk than were the non-participant young mothers. Signi®cant differences were also evident in the graduation rates of these young mothersÐ
70% of the participant mothers graduated, 28% of the non-participant young mothers graduated. Logistic regression co
ectly classi®ed
graduation/drop-out status in 76% of the cases. School attendance, mother's age at birth of the child, and participation/non-participation in
the program were signi®cant predictors. Percent core courses passed and average risk scores did not signi®cantly add to prediction.
Implications and future areas of study are discussed. q 2001 Elsevier Science Ltd. All rights reserved.
Keywords: Teen parents; School-based child care; Educational outcomes of teen parents; Program evaluation
Although recent research more critically questions the
extent and causal directions associated with many of the
negative outcomes associated with adolescent motherhood
(Geronimus & Korenman, 1992; Hoffman, Foster &
Furstenberg, 1993; Hotz, McElroy & Sanders, 1997;
Luker, 1996), the consequences of teen parenting remain
signi®cant. Poor, unwed teen mothers often have much to
overcome to succeed. Unwed teen mothers, when compared
with women of similar academic and socioeconomic back-
grounds who postpone childbearing, are more likely to drop
out of school (Allen & Pittman, 1986; Coley & Chase-Lans-
dale, 1998; Moore, Myers, Mo
ison, Nord, Brown &
Edmonston, 1993; Mott & Marsiglio, 1985); are less likely
to ®nd stable, meaningful employment; and are more likely
to rely on public assistance (Brindes & Jeremy, 1988;
Duncan, XXXXXXXXXXNearly seven in ten teen mothers go on
welfare before their child's fourth birthday and more than
40% of young mothers who receive AFDC do so over at
least a 10-year period (Allen & Pittman, XXXXXXXXXXSome
studies indicate that teen childbearers never achieve
economic parity with women who postpone childbearing
until their adult years (Furstenberg, Brooks-Gunn &
Chase-Lansdale, 1989; Coley & Chase-Lansdale, 1998).
Later childbearers are also more likely to enter stable
ma
iages than are those women who have children in
their early teens (Furstenberg, Brooks-Gunn & Morgan,
1987; McCarthy & Menken, 1979).
Nevertheless, teen mothers who can successfully manage
their educational career and social relationships do drasti-
cally improve the odds for themselves and their children.
Initiatives focused on providing needed educational and
socioemotional assistance can be effective (Furstenburg et
al., 1987; Hofferth, XXXXXXXXXXFurstenberg et al. (1987), fo
instance, found that teen mothers who had received educa-
tional assistance (in the form of a continuing educational
program and postpartum family planning services) had
etter long-term outcomesÐbeing more self-suf®cient
economically and having more stable and smaller families
than did non-program teen mothers. Educational programs
that focus on parenting skills tend to be successful in a
different yet complimentary way, leading to an improved
parent±child relationship and healthier overall development
in the child (Clewell, Brooks-Gunn & Benasich, 1989).
Related to both strategies, child care is the service most
frequently requested by adolescent mothers and the service
most likely to be unavailable (Flood, Greenspan &
Mundorf, 1985; Furstenberg et al., XXXXXXXXXXAlthough cost
Evaluation and Program Planning XXXXXXXXXX±275
XXXXXXXXXX/01/$ - see front matter q 2001 Elsevier Science Ltd. All rights reserved.
PII: S XXXXXXXXXX
www.elsevier.com/locate/evalprogplan
* Co
esponding author. Tel.: XXXXXXXXXX; fax: XXXXXXXXXX.
E-mail address: XXXXXXXXXX (H.F. Crean).
may be a prohibiting factor, high-quality child care has the
potential for enhancing teen mothers' lives as well as bene-
®ting the development of their children (Clewell et al.,
1989)