Sampling

1
Sampling

• The group chosen to be studied in specific
  research projects
• In quantitative research, generalizations are
  made about the population from conclusions drawn
  about the sample
• Sampling procedures used are critical to the
  generalizability of study conclusions
• Certain procedures lend themselves to
  generalizability more than others

2
Population - a defined aggregate or set of
persons, objects or events that met a specified
set of criteriaSample - a subgroup of the
population which serves as a reference group from
which to draw conclusions about the population
3
Generalizing from study conclusions in
qualitative studies is not a requirement of the
work. Participants are chosen for appropriateness
to the research question.
4
Composition of Populations

• People
• Events
• Institutions
• Blood samples
• Nail clippings
• Saliva
• or just about any unit of interest.

5
TARGET POPULATION
ACCESSIBLE POPULATION
SAMPLE
6
Subject Criteria

• Inclusion Criteria
• qualifying characteristics of subjects for given
  study
• Example
• Subjects will have primary epilepsy
• Must have had a seizure in the last year
• Will be 60 years of age or older

• Exclusion Criteria
• criteria which exclude subjects from a study
• Example
• Epilepsy will result from trauma, stroke,
  infection or tumor
• Subjects will not have had a temporal lobectomy

7
Sampling Techniques

• Probability Sampling
• Subjects randomly chosen - every potential
  subject has equal chance of being chosen
• Nonprobability Sampling
• Subjects chosen by nonrandom methods most
  commonly used in clinical studies

8
Strengths of Random Sampling

• Allows for estimation of sampling error.
• Sampling error is the difference between sample
  statistics and population parameters
• In other words, the variation between the average
  of values found in the sample and those found in
  the population
• Random sampling gives the greatest confidence in
  the validity of findings in a study

9
Techniques of Random Sampling

• Simple random sampling
• Systematic sampling
• Stratified random sampling
• Proportional stratified sample
• Disproportional sampling
• Cluster sampling

10
Nonprobability Sampling

• Convenience sampling
• Consecutive sampling
• Quota sampling
• Purposive sampling
• Snowball sampling

11
Sample size needs to be large enough to generate
statistical power - that is, the sample needs to
be large enough to demonstrate significant
differences between groups when they exist
12
What is a survey?
13
A research method used to collect information or
data
14
examine, appraise, inspect, scrutinize, query
explore, chart, measure, describe,
compare,document
15
Survey research is a process used to collect data
which may use specially designed survey tools OR
existing standardized tools

Survey instruments are designed to collect
descriptive information
16
When would you use a survey?
What kind of research questions would you ask
with a survey?
17
DESCRIPTIVE EXPLORATORY
EXPERIMENTAL Describe Find Relationships
Cause and Populations Effect
Survey Research
(Portney and Watkins, p. 23)
18
Describe..

• What problems do therapists find with different
  brands of power wheelchairs?
• What are practice venues in which occupational
  therapists are employed?
• What kinds of work related injuries do
  occupational therapists incur?

19
Explore

• Is there a difference between occupational,
  physical and speech/language therapists use of
  natural environments for service delivery?
• What kinds of work-related injuries occur across
  various clinical practice settings?

20
Experimental

• Is there a difference in therapists attitudes
  towards persons with disabilities before and
  after a sensitivity training program?
• Is there a difference on fieldwork scores between
  OT students who experience problem-based learning
  and those who do not?

21
Steps in Survey Process

• Research question
• Choosing survey instruments
• Choice of sampling
• target population
• accessible population
• sample
• probability vs. nonprobability sampling

22
Sampling in Surveys

• Easier to randomize
• Select Ss from data bases or professional
  organizations
• Larger data collection area possible
• Regional comparisons possible
• Cluster sampling

23
Possible formats of Surveys

• Mail
• Telephone
• Interview
• Computerized

24
Decision-Making in Surveys

• Available funds
• Sample characteristics
• Age
• Literacy status
• Culture
• Technology available

25
Survey Instruments
26
Kinds of Questions

• Open ended
• Participant-generated responses
• Used when exhaustive choices are not known
• Hard to code
• Responses may be hard to understand

• Closed-ended
• Examiner-specified responses
• Exhaustive answer choices
• Easily coded
• Uniform response data

27
Scaling Responses

• Categorical scales (classification)
• Continuous scales (characteristics measured on a
  continuum)
• Summative scales (each item contributes equally
  to a total score)
• Cumulative scales (each item contributes
  increasing weight to final score)

28
Kinds of Scales

• Likert summative scales where response items
  are ranked
• Guttman cumulative scale where each item
  reflects increasing intensities of trait being
  measured

29
More scales..

• Semantic Differential summative scale where
  ends of the scale reflect polar opposite feelings
• Visual Analogue (VAS) scale reflecting
  uni-dimensional intensity of a characteristic

30
Fine-tuning survey instruments
31
Instrument Development

• Develop questions
• Content analysis make changes
• Pilot study make changes
• Conduct study

32

data analysis
33
Descriptive

• Descriptive stats mean, median, mode, range,
  standard deviation..etcetc
• Content analysis process of categorization of
  open ended question responses

34
Exploration

• Inferential procedures which ask if there is a
  relationship between variables
• cross tabs, correlations, chi square analysis

35
Experimental

• Inferential procedures
• t-tests (parametric data)
• Mann-Whitney U (non-parametric data)
• ANOVA

36
Problems with Survey Research

• Low response rates
• Poor quality instrumentation
• Sampling issues
• Generally descriptive, not experimental

37
Done properly, survey research can be a valuable
contributor to the research continuum!