Transcript Lecture 5b
CRIM 430
Sampling & Data Collection
Simple Random
List of elements in sampling frame
Number each element
Select a number from the random numbers
table arbitrarily
The number selected indicates which element
should be selected first
Move in constant direction on the random
number tables until all sample spots are filled
Simple Random Example
Sampling Frame
1. Frank Blue
2. Dana Boots
3. David Heinz
4. Jane Dear
5. John Doe
6. John Smith
7. Sara White
Random Numbers
01209
30254
52459
If you wanted a
sample of 3, who
would you pick given
the random numbers
above?
Systematic Sampling
List of elements from sampling frame
Put in order (alphabetical, chronological, etc.)
Calculate the nth element for selection by
dividing the number desired by the total
number in the sampling frame
Start a random point and go down the list,
selecting every nth element until you’ve
reached your sample size
Stratified Sampling
Modification to previous methods
Ensures a degree of representativeness
First, population is organized into
homogeneous subsets (all males, all females)
Sample elements are selected using the
simple random or systematic method
Once appropriate sizes for each group is me,
the sample elements are placed back
together to create the whole sample.
Sample distributions of stratified variable
equals the population distribution
Disproportionate Stratified
Used when members of a population vary
widely in size (e.g., 90% Male & 10%
White)—it ensures that you receive an
appropriate number of “rare” cases
Same methods as stratified sampling except
the distribution of the stratified variable is
greater in the sample than it is in the
population (I.e., the sample may=50/50
males and females)
The “rare” cases are over-sampled
To compensate for this, data can be weighted
during analysis to reflect population
distributions
Multi-Stage Cluster Sampling
Used when there are many “layers” to the
target population (e.g., police officers from all
metropolitan departments across the United
States)
First, apply sampling to select departments
Next, apply sampling to select police officers
Sampling is done in stages or clusters
More clustering results in potentially less
representativeness
Non-Probability Sampling
Probability sampling designs are not possible
in many situations
Non-probability sampling is an alternative;
however, the samples are not representative
of the population from which they are drawn
Non-probability sampling designs are prone to
selection bias
Non-Probability sampling designs are,
therefore, weaker than probability sampling
designs
Non-Probability Sampling Designs
Purposive or Judgmental Sampling:
Identifying a sample based on the presence
of a particular characteristic
Quota Sampling: Identifying a sample using a
matrix to represent the characteristics of the
population
Convenience Sampling: Sample is selected
because access is easy and convenient
Snowball Sampling: Using one respondent to
provide contact to 2-3 additional
respondents—continuous process to identify a
larger sample
Types of Data Collection
Self-Report Data
Data derived from the respondent him/herself
Key=Ask questions to subjects
Example: National Crime Victimization Survey
Official Data
Data derived from agency records or databases
Key=Examine written records
Example: Uniform Crime Reports
Observation Data
Data derived from watching the activities of people or
events; information is coded by observer
Key=Watching behavior
Example: Coding the behavior of detention officers and
offenders at a correctional institution
Data Collection: Asking Questions
Asking questions provides an indirect measure
or substitute for making observations—Used
to capture things such as experiences with
crime, attitudes and beliefs
Self-administered surveys
Mailed surveys
In-person structured interviews
Telephone interviews
Focus groups
Asking Questions, Cont’d.
Types of questions included in surveys
Open-ended
Close-ended
Statements with levels of agreement
Contingency questions (if yes, proceed; if no, skip
to)
Presentation of questions in a survey
Should be clear—avoid ambiguity & confusion
Keep items short and to the point
Keep items neutral and unbiased
Add disclaimers/introductions to provide
respondent with direction
Assessment
Strengths
Useful in describing large populations
Standardized surveys improve strength of measurement
Flexible during planning
Provides opportunity to capture a lot of information
Weaknesses
Limited in the information it can capture
Does not capture the context of the situation
Not flexible during implementation
Relies on the truthfulness & memory of respondent
Data Collection: Written Records
Published Statistics
Compiled statistics produced and distributed for public
consumption
Example: UCR
Nonpublic Agency Records
Records kept by agency for processing purposes
Not available for public consumption
Example: Probation case files
New Data Collected by Agency Staff
New information collected as part of the agency process in
order to investigate a research question
Example: Use of a new screening tool
Written Records, Continued
Other Related Sources:
Content Analysis
Reviewing narratives, usually written, to
identify patterns and themes
Example: Newspaper reports of crime over time
Secondary Data Analysis
Data are originally collected one set of
researchers and then made available to other
researchers for analysis
Example: Arrestee Drug Abuse Monitoring Data
Assessment
Strengths
In general, the availability of these data is much easier than
self-report
The cost can be significantly less than self-report
Conducive to large numbers
Weaknesses
Access is sometimes limited especially with regard to cj
information
Information is limited by agency priorities
Data are rarely flexible and are defined by agency not the
research question
Quality of data is sometimes questionable due to missing
and inconsistent reporting of information
Data Collection: Observation
Structured observation=quantitative
List of items that an observer will code while
observing behavior
Observers use a standard code sheet that contains
items and close-ended responses
Items are completed during or immediately after
the observation occurs
Unstructured observation=qualitative
General descriptions recorded in a narrative
Transcripts of taped descriptions or written notes
used to to identify themes and patterns
Assessment
Strengths
Flexible to the needs of the research question
Allows for more thorough investigation of certain
situations
Weaknesses
Time consuming and expensive
Collection of data is potentially impacted by data
collector’s bias
Analysis of data can be subjective and open to
bias, especially in the case of unstructured
observations
Limited sample size unless extremely well-funded
Multiple Measures
Many studies utilize different types of data to
answer research questions
For example, using both official and self-report
data to measure variables
Using multiple methods of data collection
builds on the strengths of each method
individually and minimizes their weaknesses
Multiple methods can increase the reliability
and validity of the data you collect
Multiple methods, however, are often
expensive and time-consuming
Deciding Which Method to Use
Decide on your method based on:
Research Question: What type of information does
your research question require?
Availability/access to the sample: How available is
a sample and what is your access?
Size of the sample: How large of a sample do you
need?
Time and resources: How much time and money
do you have at your disposal?
Ethical Issues
All research is bounded and defined by
professional code of ethics
Social science research is particularly subject
to ethical codes because it almost always
includes humans subjects
When conducting research, it is necessary to
balance the potential benefits from doing the
research against the possibility of
psychological, emotional, and physical harm
IRB
All human subject research conducted at a
University must be reviewed and approved by
the Institutional Review Board
The IRB ensures that federally defined
safeguards are applied in all types of research
with humans
Code of Federal Regulations Title 45, Chapter 46,
U.S. Department of Health & Human Services
Additional rules apply to two populations
considered particularly vulnerable:
Prisoners
Children
IRB Safeguards
Safeguards include:
Written consent form must be used to request
participation in the study
Written list of benefits and costs of participation
Subject must voluntarily participate
Subject must be guaranteed anonymity or
confidentiality
Analysis of data in the aggregate
Protection from deceit by researchers