Design and Implement Basic Study Designs
CH 3: CHOOSING THE STUDY SUBJECTS
1. Good choice of study subjects:
- Purpose of ensuring that the findings in the study accurately
represent the population of interest
- Specify sample:
o Can be studied at acceptable cost in time and money (=
modest size and convenient to assess)
o Yet large enough to control for random error
o Representative enough to allow generalizing findings
Basic Terms and Concepts:
1. Populations and samples:
- Population = complete set of people with specified
characteristics
- Sample = subset of the population
- Defining features of a population:
o Target population = large set of people throughout the
world to which results may be generalized; defined by
clinical and demographic characteristics
o Accessible population = subset of the target population
available for study; defined by geographical and temporal
factors
o Intended study sample = subset of accessible population
that investigator seeks to include in the study
o Actual study sample = group of subjects that participate in
the study
2. Steps in designing the protocol:
- Specify the clinical and demographic factors of target population
that will serve the RQ
- Use geographic and temporal criteria to specify study sample
that is representative and practical
,Selection Criteria:
1. Establishing selection criteria
- Inclusion criteria = define the main factors of target population
that pertain to the RQ
o Decisions should be sensible, can be used consistently
through the study, and can be clearly described to others
- Effect modification = an effect in one race that is different from
that in other races
o An interaction
o The number needed to statistically test for it is quite large;
most studies are not powered enough to detect it
- Exclusion criteria = individuals who meet the inclusion criteria
and are suitable for study
o But, have features that may interfere with:
The success of follow-ups
The quality of data
The acceptability of randomized treatment
o General rule = have as few exclusion criteria as possible
2. Clinical vs community populations
- Clinical samples are easier to find and best to sample from
primary care practices
- Community samples represent healthy populations; not fully
representative of general population
Sampling:
1. Non-probability samples:
- Convenience sample = meet entry criteria and are easily
accessible
- Consecutive sample = reduces selection biases by consecutively
selecting subjects who meet the entry criteria
- Validity from drawing inferences from any sample:
o The premise that it sufficiently represents the accessible
population
, 2. Probability samples:
- The gold standard for ensuring generalizability
o Uses random process to guarantee that each unit of the
population has a specified chance of being included in the
sample
- Simple random sample:
o List all people in the population from which sample will be
drawn
o Select subset at random
o Most commonly used in clinical research
- Systematic sample:
o List the population and select sample by preordained
periodic process
o Susceptible to errors caused by natural periodicities in the
population
o Allows investigator to predict and manipulate the subjects
included
o Rarely better than simple random sample
- Stratified random sample:
o Divide population according to features, such as sex or
race, into subgroups (= strata)
o Take random sample from each stratum
o Less common populations can be overrepresented in order
to yield similar incidence estimates across all groups
- Cluster sample:
o Random sample of natural groupings (= clusters)
o Useful when population is widely dispersed
o The effective sample size will be smaller than the number
of subjects
Recruitment:
1. Goals:
, - Recruit a sample that accurately represents the target population
(= reducing chance of wrong answer to RQ due to systematic
error (bias))
- Recruit sufficient sample size (= reduce chance of wrong answer
to RQ due to random error (chance))
2. Achieving representative sample:
- Design phase = choose target and accessible populations, and
approaches to sampling
- Implementation = guard against errors in applying entry criteria,
and enhance successful strategies
- Response rates = population of selected subjects who consent to
be enrolled; influences validity of inferring that the enrolled
sample represents the population
- Non-response = concern especially for descriptive studies
o Compromises generalizability and influences the
conclusions
o Reduce by repeated contact attempts and improve
efficiency and attractiveness of the study
3. Recruiting sufficient number of subjects:
- Approaches to the problem of falling short in the rate of
recruitment
o Estimate the magnitude of recruitment problem empirically
with pretest
o Plan the study with an accessible population larger than
believed necessary
o Make contingency plans for additional subjects
CH 4: PLANNING THE MEASUREMENTS
1. Measurements = describe phenomena in terms that can be
analyzed statistically
- Validity of study = depends on how well the variables designed
for the study represent the phenomena of interest
CH 3: CHOOSING THE STUDY SUBJECTS
1. Good choice of study subjects:
- Purpose of ensuring that the findings in the study accurately
represent the population of interest
- Specify sample:
o Can be studied at acceptable cost in time and money (=
modest size and convenient to assess)
o Yet large enough to control for random error
o Representative enough to allow generalizing findings
Basic Terms and Concepts:
1. Populations and samples:
- Population = complete set of people with specified
characteristics
- Sample = subset of the population
- Defining features of a population:
o Target population = large set of people throughout the
world to which results may be generalized; defined by
clinical and demographic characteristics
o Accessible population = subset of the target population
available for study; defined by geographical and temporal
factors
o Intended study sample = subset of accessible population
that investigator seeks to include in the study
o Actual study sample = group of subjects that participate in
the study
2. Steps in designing the protocol:
- Specify the clinical and demographic factors of target population
that will serve the RQ
- Use geographic and temporal criteria to specify study sample
that is representative and practical
,Selection Criteria:
1. Establishing selection criteria
- Inclusion criteria = define the main factors of target population
that pertain to the RQ
o Decisions should be sensible, can be used consistently
through the study, and can be clearly described to others
- Effect modification = an effect in one race that is different from
that in other races
o An interaction
o The number needed to statistically test for it is quite large;
most studies are not powered enough to detect it
- Exclusion criteria = individuals who meet the inclusion criteria
and are suitable for study
o But, have features that may interfere with:
The success of follow-ups
The quality of data
The acceptability of randomized treatment
o General rule = have as few exclusion criteria as possible
2. Clinical vs community populations
- Clinical samples are easier to find and best to sample from
primary care practices
- Community samples represent healthy populations; not fully
representative of general population
Sampling:
1. Non-probability samples:
- Convenience sample = meet entry criteria and are easily
accessible
- Consecutive sample = reduces selection biases by consecutively
selecting subjects who meet the entry criteria
- Validity from drawing inferences from any sample:
o The premise that it sufficiently represents the accessible
population
, 2. Probability samples:
- The gold standard for ensuring generalizability
o Uses random process to guarantee that each unit of the
population has a specified chance of being included in the
sample
- Simple random sample:
o List all people in the population from which sample will be
drawn
o Select subset at random
o Most commonly used in clinical research
- Systematic sample:
o List the population and select sample by preordained
periodic process
o Susceptible to errors caused by natural periodicities in the
population
o Allows investigator to predict and manipulate the subjects
included
o Rarely better than simple random sample
- Stratified random sample:
o Divide population according to features, such as sex or
race, into subgroups (= strata)
o Take random sample from each stratum
o Less common populations can be overrepresented in order
to yield similar incidence estimates across all groups
- Cluster sample:
o Random sample of natural groupings (= clusters)
o Useful when population is widely dispersed
o The effective sample size will be smaller than the number
of subjects
Recruitment:
1. Goals:
, - Recruit a sample that accurately represents the target population
(= reducing chance of wrong answer to RQ due to systematic
error (bias))
- Recruit sufficient sample size (= reduce chance of wrong answer
to RQ due to random error (chance))
2. Achieving representative sample:
- Design phase = choose target and accessible populations, and
approaches to sampling
- Implementation = guard against errors in applying entry criteria,
and enhance successful strategies
- Response rates = population of selected subjects who consent to
be enrolled; influences validity of inferring that the enrolled
sample represents the population
- Non-response = concern especially for descriptive studies
o Compromises generalizability and influences the
conclusions
o Reduce by repeated contact attempts and improve
efficiency and attractiveness of the study
3. Recruiting sufficient number of subjects:
- Approaches to the problem of falling short in the rate of
recruitment
o Estimate the magnitude of recruitment problem empirically
with pretest
o Plan the study with an accessible population larger than
believed necessary
o Make contingency plans for additional subjects
CH 4: PLANNING THE MEASUREMENTS
1. Measurements = describe phenomena in terms that can be
analyzed statistically
- Validity of study = depends on how well the variables designed
for the study represent the phenomena of interest
