You are on page 1of 3

Discussion Week 10

1. Describe an application of a logistic regression analysis to your area of research. Clearly


define your variables and state the null and alternative hypotheses.
My example focuses upon using binomial logistic regression analysis and I want to look at the
relationship between unemployment and ethnic. The model has a dichotomous dependent
variable and logistic regression is appropriate because there would be a categorical outcome
(George & Mallery, 2012; Week Ten Study Notes, 2013). The variables used in this study
include behavior problems with two levels No and YES and another variable sis ethnic group
with two levels White and Black. I am interested in modeling the variation in the probability of
being perceived to have behavior problems and modeling the variations in the proportions with
perceived behavior problems amongst black boys compared with white boys.

Hypothesis for this study is as:
Hypotheses:
Ho: Behavior problem likelihood can be predicted by ethnicity.
H1: Behavior problem likelihood can be predicted by ethnicity

2. Present the results from your application assignment using APA format in accordance
with the Publication Manual of the American Psychological Association. State a null and
alternative hypothesis and include a measure of effect size.

APA Results of Week 10 Assignment
From a sample of 537 participants, a binary logistic regression was applied which tested the level
of empathy tendency toward helpfulness. A new variable diempath was created from empathy
total, which is categorical dichotomous with two categories of high and low values of empathy
(coded as 0 = 5.11 and 1= > 5.11). The new variable diempath was used as a dependent
variable and the variables sympathy, anger, efficacy, and ethnicity were used as predictors for
the regression with significance value of less than .05 (p < .05). Ethnicity was coded as 1=white,
2=black, 3=Hispanic, 4=Asian. Each ethnic group was compared against an other category.

H0= ethnicity, anger, empathy, sympathy, and self-efficacy have no relationship to predicting
empathy leading to helping
H1= the predictor variables do have a relationship to predicting empathy leading to helping

The model was statistically significant,
2
(7, N = 537) = 23.154 p = .002.The results indicate
that ethnicity, anger, and efficacy were not significant predictors of helpfulness; however
sympathy was a significant predictor with odds ratio of 1.30. This suggests that for every unit of
sympathy expressed by the helper towards the friend in need, the helper was over one time more
likely to help. Using the Nagelkerke method, this model explains 5% of the variance in whether
or not the sympathy felt towards a friend in need influenced the desire to help. Table 1 outlines
the results of the logistic regression mode.

Table 1:
Results of the Logistic Regression Model (N = 537)

Variable B S.E. OR
95% C.I. (OR)
Lower Upper
sympathy .282* .077 1.325 1.140 1.540
anger .051 .058 1.052 .939 1.179
efficacy .169 .095 1.185 .984 1.426
ethnic

White -.084 .144 .919 .693 1.220
Black .122 .252 1.129 .689 1.850
Hispanic -.072 .209 .930 .617 1.402
Asian .321 .222 1.379 .892 2.132
*: P < 0.05

3. Answer your quarterly question related to the weeks assignment.
Compare the assumptions required for multiple and for logistic regressions

Assumptions for Multiple Regression:

The errors are normally distributed
The mean of the errors is zero
Errors have a constant variance
The model errors are independent

Assumptions for Logistic Regression.

Logistic regression does not assume a linear relationship between the dependent and
independent variables.
The dependent variable must be a dichotomy (2 categories).
The independent variables need not be interval, nor normally distributed, nor linearly
related, nor of equal variance within each group.
The categories (groups) must be mutually exclusive and exhaustive; a case can only be in
one group and every case must be a member of one of the groups.
Larger samples are needed than for linear regression because maximum likelihood
coefficients are large sample estimates. A minimum of 50 cases per predictor is
recommended

Reference
George, D., & Mallery, P. (2012). IBM SPSS Statistics 21 step by step: A Simple guide and
reference (12th ed.). Boston, MA: Pearson.
Bruin, J. 2006. newtest: command to compute new test. UCLA: Statistical Consulting Group.
Retrieved on November 7, 2013 from http://www.ats.ucla.edu/stat/stata/ado/analysis/

You might also like