2015 Lecture 09 Chapters 12 13 14 15 Evaluation and

Download Report

Transcript 2015 Lecture 09 Chapters 12 13 14 15 Evaluation and

Lecture 09
Applied Social Research Ch 14
Available Data Ch 12
Content Analysis Ch 13
Uses of SOCIAL RESEARCH
Applied Research
Explanatory Research
Critical Research
Exploratory Research
Descriptive Research
 Exploratory Research Ground-breaking research
on a relatively unstudied topic or in a new area
 Descriptive Research Research designed to
describe groups, activities, situations or events with
a focus on structure, attitudes or behavior “how the land lies”
 Explanatory Research Research that seeks to
explain the cause of a phenomenon, and typically
asks a “What causes what?” or “Why is it this way?”
research question
 Critical Research Research with the goal of
critically assessing some aspect of the social world
 Applied Research Research intended to be useful
in the immediate future and to suggest action or
increase effectiveness in some area (eg evaluation
or participatory action research)
100
200
300
400
500
 A research project can have more than one purpose.
http://iar.csusb.edu/
 Applied Research is
research with a
practical purpose
and helps to
create, modify, and
implement
programs and
activities that make
a difference in
people’s lives
 Most Social
Research
balances
applied and
basic research
http://www.buzzle.com/articles/difference-between-basic-and-applied-research.html
Is the following basic or applied?
How?
Database Links to study networksResearch May Give Potential
Homicide Victims A Heads Up
One tool researchers have used to identify high-risk group
members is social network analysis (SNA). Yale
sociologist Andrew Papachristos has developed a mathematical
method to identify the structures of street groups using
connections contained within police records of arrests, field
stops, and similar data. SNA recognizes that the overwhelming
majority of suspects and victims in homicides and shootings
know one another. In fact, Papachristos has demonstrated that
they are often part of a closely linked social network — that is,
shooters and victims tend to be closely linked to many other
shooters and victims.
Research on these social networks has found that individuals
with links to street groups and those closely linked to homicide
victims are at elevated risk of becoming shooters or victims
themselves. SNA has helped communities identify group
members to notify. Using co-arrest and field stop records, SNA
can create graphs of connections between street group
members. A GVI partnership can use these maps to identify
“brokers” — people with a profusion of group-related
connections — and people closely connected to homicide victims
and suspects. These high-risk group members can then be given
custom notifications to give the city’s antiviolence message the
greatest reach and to disrupt street group networks.
http://nnscommunities.org/ourwork/innovation/social-network-
http://www.npr.org/2014/10/09/354
754588/research-may-give-potentialhomicide-victims-a-heads-up
http://iar.
csusb.edu
/
The Policy
Cycle
http://trochim.human.cornell.edu/kb/pecycle.htm
http://trochim.human.cornell.edu/kb/pecycle.htm
To document a social problem we can
gather data from multiple sources:
• Primary data is original data
collected by the researcher
• Secondary data is collected by
someone other than the researcher
• Has been the dominant mode
of social science research in
the twentieth century.
http://www.westernseminary.edu/transformedblog/2011/12/05/problem-4-the-unsocial-gospel/
• A major advantage of using available
data is that data are usually cheap and
convenient.
• Usable secondary data can be found by
doing a literature review or by searching
the Internet.
http://trochim.human.cornell.edu/kb/pecycle.htm
To document a social problem we can
gather data from multiple sources:
• Primary data is original data
collected by the researcher
• Secondary data is collected by
someone other than the researcher
• Has been the dominant mode
of social science research in
the twentieth century.
Look
Familiar?
http://www.westernseminary.edu/transformedblog/2011/12/05/problem-4-the-unsocial-gospel/
• A major advantage of using available
data is that data are usually cheap and
convenient.
• Usable secondary data can be found by
doing a literature review or by searching
the Internet.
http://sda.berkeley.edu/
Data come in all sorts of
formats…
• Raw
• Half Baked
• Fully Cooked
• Statistics that have been
made into statistical
documents are called
existing statistics.
http://www.outsidethebeltway.com/public-acceptance-of-same-sex-marriage-at-all-time-high/
Look
Familiar?
http://sda.berkeley.edu/
• Be cautious to avoid
ecological fallacymisleadingly making
• Existing Statistics are
inferences about certain
statistics provided by large
types of individuals from
organizations
information about the
groups they belong to,
that might not be
• In general, the unit of
exclusively comprised of
analysis in studies based on
those individuals.
existing statistics is not the
individual.
http://www.fbi.gov/about-us/cjis/ucr/crime-in-the-u.s/2013/crime-in-the-u.s.-2013/tables/table-10/table-10pieces/table_10_offenses_known_to_law_enforcement_by_alabama_by_metropolitan_and_nonmetropolitan_counties_2013.xls
Social Problems (and other areas of
interest) can also be documented
using Content Analysis- studying
available recorded data other than
secondary data.
Could collect data from such sources
as:
• Newspapers
• Want ads
• Diary entries
• Email messages
It doesn’t have to be applied.
Might study such things as:
• Women’s portrayal through art
• Election coverage on various web
sites
• Facebook accounts for social
activity among teens
• Social networks through obituaries
http://trochim.human.cornell.edu/kb/pecycle.htm
Units of Analysis
are the units
about which
information is
collected and
units of
observation are
the units from
which information
is collected.
Generally, quantitative content
analyses tend to be deductive and
qualitative content analyses tend to
https://depts.washington.edu/uwiser/tp_content_analysis.htm
be inductive.
It doesn’t have to be applied.
If you were interested in the
Might study such things as:
difference between commercials
• Women’s portrayal through art
during children’s and adult’s TV
• Election coverage on various web
programs,
sites
Units of observation= child’s and
• Facebook accounts for social
adult’s TV programs
activity among teens
Units of analysis = commercials
• Social networks through obituaries
attached to those programs
Advantages
of Content
Analysis
http://media.tumblr.com/f8d13397889a2b902cafdde41d5315da/tumblr_inline_mudey3mkH11rkv4z0.jpg
 Can be conducted with a relatively
small expenditure of time, money,
and person power.
 Doesn’t require you get everything
right the first time
 Is unobtrusive
Outcome evaluations usually begin with a question
like- What is the reason the program is doing what
it is doing?
http://iar.csusb.edu/
http://trochim.human.cornell.edu/kb/pecycle.htm
http://iar.csusb.edu/
 Typically intended to assess the impact of a
specific policy or program
 If the Independent Variable affects the
dependent, this is called a causal hypothesis
IV

http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B6V7V-3WR4K4M-8&_user=735929&_handle=B-WA-A-W-AV-MsSAYZW-UUW-AAUDUAADDV-AAUVCEWCDV-YBACCWCBC-AV-U&_fmt=full&_coverDate=06%2F30%2F1999&_rdoc=8&_orig=browse&_srch=%23toc%235852%231999%23999779998%23102970!&_cdi=5852&view=c&_acct=C000040778&_version=1&_urlVersion=0&_userid=735929&md5=6e6b6f92758532f01a88230d7ec29380
DV
• Evaluation research: distinguished from basic research by its immediate
usefulness and its specific practical implications
• By the end of an outcome evaluation researchers should be able to answer
questions about the program’s success and ways the program or policy can
better serve its target population.
Outputs
Formative Analysis:
Usually Occurs in the early stages of
program development and design
Main purpose of process or
implementation studies: provide
information about the faithfulness of
the program to its original design
http://www.fsg-impact.org/images/upload/From%20Insight%20to%20Action(3).pdf
Outcomes
Summative Main Purpose:
Assessing whether Short,
Intermediate, and Log-Term
Outcomes are Met
A Natural Experiment For Policy AnalysisSome Parole Requirements Could Be
Increasing The Crime Rate
nd David Kirk - he's a sociologist at the
University of Texas - he immediately
realized that a natural experiment was
unfolding. And that's because he studies
recidivism. This is the likelihood that
people who are released from prison will
go back to a life of crime and go back to
prison. He's interested in whether
recidivism is shaped by whether the
people released from prison go back to
the neighborhoods they came from. Now,
in most parts of the country, people who
are released from prison often find their
way back to their old neighborhoods.
Many states actually require parolees to
go back to the counties they came from.
Kirk wanted to know whether was possible
that going to a different community - can
that help you to turn over new leaf.
http://www.npr.org/2014/07/08/329
731503/some-parole-requirementscould-be-increasing-the-crime-rate
Long Term Outcomes might look like:
Source: Adler and Clark 2015:368
Drugs Bite
http://www.dare-america.com/home/default.asp
 Video Overview of DARE
 http://sam.ntpi.spcollege.edu/spjc/vi
ew/eventListing.jhtml?eventid=5435
&c=13009

IV

http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B6V7V-3WR4K4M-8&_user=735929&_handle=B-WA-A-W-AV-MsSAYZW-UUW-AAUDUAADDV-AAUVCEWCDV-YBACCWCBC-AV-U&_fmt=full&_coverDate=06%2F30%2F1999&_rdoc=8&_orig=browse&_srch=%23toc%235852%231999%23999779998%23102970!&_cdi=5852&view=c&_acct=C000040778&_version=1&_urlVersion=0&_userid=735929&md5=6e6b6f92758532f01a88230d7ec29380
DV
DARE Evaluation as an Example
http://www.dare-america.com/home/default.asp
 Research Question: Is DARE effective?
 Possible hypotheses
 DARE possible dependent variables
 Attitudes towards
drugs
 knowledge of drugs
 drug use (behavior)
Independent Variable
A 2002 DARE Evaluation
http://www.dare-america.com/home/default.asp
 SAMPLE:
 More than 15,000 students from 83 high
schools and 122 middle schools in 6 U.S.
cities (Detroit, Houston, Los Angeles,
Newark, New Orleans and St. Louis) are
the sample for this study.
http://www.dare-america.com/officers/Officers_Training/Defaultfe17.asp?N=Officers_Training&M=24&S=0
 STUDY DESIGN: Experiment
 Half of the schools were randomly
selected to receive the new DARE
program which now begins in 7th grade
and is designed to be more interactive.
 The students in the other schools will
receive the original DARE program or no
programs and will be the control groups.
 Data will be collected before the
program begins in 7th grade and then
periodically through the 11th grade.
Pretest-Posttest Control Group Experiment
2 Groups
Random
Assignment
 DATA
 The variables include students’
knowledge, attitudes, and behaviors.
The measures will be answers to
questions.
Adler and Clark p. 408
 Results….
Adler and Clark
 Turn to Text Page 408
 Which Dependent Variables would use Chi Square
and which Difference of Means t test?
 Turn to p.410 and 411
 Notice the charts that show 2 categories
percentage using alcohol and other drugs (in last
30 days) Does this look familiar?
 Chi- Square for % in a category
 Now notice Table 14.1 and figure 14.3 these
are comparing ????
 Means or average. Also since it is a
continuous variable, can have standard
deviations and means.
Drug Use: Lifetime Prevalence
Just like a chart made
with a Cross-Tab
IV: Dare or not
DV: Drug use or not
Drug Use Within Last 30 Days
 We can see differences but are they
significant?
 Which type of signif test would we use for
an IV and DV with 2 categories each?
 Alcohol (Y and N)
 Dare (Y and N)
Chi-Square
 If we re-code things into 2 (or even 3
or 4) categories or if they are
naturally that way, we know we use
Chi-square to test significance.
 But what about age or grade level or
# of TV hours watched, these are
continuous. How do we test those?
 The rule of thumb is, if it makes
sense to calculate a mean (average),
i.e. if it is a continuous variable,
 Then use a comparison of means (Z
or t test)
 The t-test assesses whether the
means of two groups are statistically
different from each other. This
analysis is appropriate whenever you
want to compare the means of two
groups, and especially appropriate as
the analysis for the posttest-only
two-group randomized experimental
design.
http://www.socialresearchmethods.net/kb/stat_t.htm
http://www.socialresearchmethods.net/kb/stat_t.htm
http://www.socialresearchmethods.net/kb/stat_t.htm
 This leads us to a very important
conclusion: when we are looking at
the differences between scores for
two groups, we have to judge the
difference between their means
relative to the spread or variability of
their scores. The t-test does just this.
http://www.socialresearchmethods.net/kb/stat_t.htm
Hypothesis testing
chi-square
 The
test enabled
you to test hypotheses where
concepts were measured at the
nominal or ordinal level
 A difference-of-means test
allows you to test hypotheses when
your variables are measured at the
continuous level (interval or ratio)
14.3
 Adler and Clark
 Also look at p. 410 last sentence,
statistical significance. This is what
we are doing today, statistical
significance for diff of means b/t 2
groups.
Beyond here FYI
Lecture 10
Statistical Analysis Ch 15