November 10, 2005 Lecture Notes

Research for Social Workers
Salem State University
School of Social Work
Jeff Driskell, MSW, PhD
Today’s Class
• Announcements/Check-in
• Lecture
▫ Statistical interpretation
• Work time
Think…Pair…Share
4 basic types of statistical tests:
Description
• Mean, standard deviation
• Median, Mode
• Percentage, frequency
Correlation
• Pearson’s correlation
Comparison
Prediction
•
•
•
•
Student’s t tests
Chi-square tests
ANOVA
Odds ratios
• OLS regression
• Logit regression
Comparison Tests
Chi-Square Tests
• Also known as:
▫ Chi-square goodness-of-fit test, commonly
referred to as the chi-square test
▫ Pearson’s chi-square test
▫ Yates’ chi-square test, also known as Yates'
correction for continuity
▫ Mantel-Haenszel chi-square test
▫ Linear-by-linear association chi-square test
Chi-Square Tests
• Assessing if there is a
relationship/comparison between two
categorical (nominal and/or ordinal)
variables
▫ Compare 2 or more groups (2 groups easiest).
 I.E. Is there a statistical difference between
groups?
• Reported as x2 in text and tables
Current Issues
• Abortion
• Gun rights
• Legalization of marijuana
Chi Square Example-
(Emlet et al. 2002)
What are t-tests
Students’ t-test: Why use it?
t-test: Requirements
• Requires Continuous variables for the DV
and categorical for the IV.
• DV variable being tested is normally distributed
in both groups.
▫ When these tests can’t be conducted, similar tests
can be used: (non-parametric)
 (Independent) Mann-Whitney U test
 (Paired) Binomial test or Wilcoxon signed-rank test
t-test choices
1. Independent t-test- The data come from two
independent populations/groups (e.g. women and men,
or students from section A and from section B)
<- between-subjects design
2. Dependent/paired t-test: The data tend to come from
related populations /groups. .comparing observations
from two occasions (e.g. “before treatment” and “after
treatment”) pre/post test
<- within-subjects design
Reporting Independent t-test
On average, participants experienced greater
anxiety to real spiders (M=47.00 SD= 11.02 )
than to pictures of spiders (M=40.00, SD=9.3)
This difference was not significant t(22)=-1.68,
p>.05.
df=Degrees
of freedom
t =statistic
p= Probability
level
Research Question ExampleIs there a significant difference in mean salary between
social worker and psychologists?
Olver et al., (2009)
Interpret
statement
Article
• In pairs
▫ Based on the article you selected for today’s class,
interpret one of the t-test results. What does it
mean? Be prepared to share with the class.
What is ANOVA?
.
ANOVA Requirements
• Requires:
▫ DV to be measured at the interval/ratio
level
▫ IV at the categorical level.
ANOVA tests
• Also known as:
▫
▫
▫
▫
Fisher’s test of variance
Fisher’s ANOVA
Fisher’s analysis of variance
One-way ANOVA
ANOVA
• Compare 3 or more groups on a continuous
variable (mean)
• Reported as F statistic but focus on significance
Example Research Question- 1
Is there a statistically significant difference in the
average salary among job types (social worker,
psychologist, Psychiatrists)?
• Post hoc- to determine which groups are sig
different.
Example Research Question- 2
Is there a statistically significant difference in
depression scores based on treatment approach?
(CBT, psychodynamic, eclectic)?
Narrative Example 1
One way analysis of variance was conducted to
determine if self esteem scores differed in
relation to the number of religious services
attended. There was an overall difference
between the means of the three groups F(3,
1938)= 1.58, p<.05. However the difference
was not significant.
Narrative Example 2
Teacher satisfaction differed significantly among
the three groups, F(2,12) = 8.18, p < .01.
Teachers were most satisfied in charter schools
(M = 9.00) and less satisfied in public and
private schools (both Ms = 6.00).
The Scheffe post hoc tests indicated that teacher
satisfaction in the charter schools differed
significantly from teacher satisfaction in public
and private schools (p < .05).
Odds ratio
Odds ratios
What does it tell you?
• What are the odds that one group is more likely
than another to experience one condition
▫ People with and without disabilities: Who is more
likely to access substance abuse treatment (Y/N)?
▫ Male vs. Female ex-offenders on postincarceration employment retention for a year or
more (Y/N)?
Odds ratios: Choices
Adjusted
 Seeing if one group is
more likely than the
other to evidence some
condition – controlling
for other factors
Unadjusted
 Just seeing if one
group is more likely
than the other to
evidence some
condition
An adjusted odds ratio:
• Needs to be logic to the adjusting
• Exploratory analyses of children placed in
foster care on a voluntary basis:
▫ “Children with ID were more likely than
children without ID to live in an institutional
setting (age, gender, co-morbid disabilityadjusted OR=1.27***)”
▫ ID=intellectual disability
Odds ratios:
How to interpret them
OR=1.5
1.5 times more likely…
OR=12.5
Almost 13 times more
likely…
OR=0.50
Fifty percent less likely…
OR=O.89
11 percent less likely…
Odds ratios:
What you are looking for
• OR = 1
▫ Condition equally likely in both groups
• OR > 1
▫ Looks like this: OR=2.34*
▫ Condition is more likely in the first group
• OR < 1
▫ Looks like this: OR=0.34*
▫ Condition is less likely in the first group
Table 3: Removal-related characteristics of foster children with and without ID†
Variable
Child's disability
Relinquishment
Neglect
Sexual abuse
Parental death
Physical abuse
Inadequate housing
Abandonment
Parental incarceration
Child's alcohol abuse
Child's behavior
Parental drug abuse
Child's drug abuse
Parental alcohol abuse
Parental inability to cope
Foster children
with ID
(N=17,714)
3,054 (17.3%)
542 (3.1%)
10,362 (58.5%)
1,768 (10.0%)
249 (1.4%)
3,132 (17.7%)
1,395 (7.9%)
928 (5.2%)
627 (3.5%)
244 (1.4%)
3,103 (17.5%)
1,432 (8.1%)
254 (1.4%)
1,264 (7.1%)
3,461 (19.6%)
Foster children
without ID
(N=655,536)
18,292 (2.8%)
11,849 (1.8%)
321,596 (49.4%)
45,436 (7.0%)
7,117 (1.1%)
103,880 (15.9%)
52,104 (8.0%)
38,306 (5.9%)
29,551 (4.5%)
10,301 (1.6%)
108,118 (16.6%)
85,635 (13.2%)
18,969 (2.9%)
47,722 (7.3%)
130,181 (20.0%)
Test
OR=6.97***
OR=1.81***
OR=1.71***
OR=1.42***
OR=1.21**
OR=1.17***
OR=1.11***
OR=0.88***
OR=0.87**
OR=0.81**
OR=0.71**
OR=0.68***
OR=0.46***
NS
NS
Note: All odds ratios are age and/or gender-adjusted with foster children without ID as referrents. ***
p<.001 **p<.01 *p<.05 †Reasons for removal are not mutually exclusive
Level of Measurement & Significance Tests
Chi-Square
IV & DV are nominal and/or ordinal
t-test
IV is nominal (group like men & women)
DV is Interval/Ratio (or a scale)
ANOVA
IV is nominal (group with 3 or more categories)
DV is I/R (or a scale)
Group Work Time