Chapter 10: Sampling and Data Collection in Quantitative Studies

Population

All of the cases a researcher is interested in

Sampling

Selection of a portion of the population (a sample) to represent the entire population

Eligibility Criteria

Characteristics that define the population. There will be inclusion and exclusion criteria

Strata

Subpopulations of a population (ex: female, employed, sexually active)

Target Population

The entire population of interest

Accessible Population

Portion of target population that is accessible to the researcher. Sample of population is drawn from accessible population.

Representative Sample

Sample whose key characteristics closely represent those of the population. This is a sampling goal in quantitative research. More easily achieved when you use

- Probability sampling

- Homogeneous populations

- Larger samples

Sampling Bias

Systematic over/underrepresentations of segments of the populations on key variables when the sample is not representative

Nonprobability Sampling

Not random selection. Less likely to produce representative sampling.

Probability Samping

Random selection. Each element has equal, independent chance of being selected.

Types of Nonprobability Sampling

- Convenience sampling

- Quota sampling

- Consecutive sampling

- Purposive sampling

Convenience Sampling

Using most readily available participants

Quota Sampling

Identifies strata and then identifies how many people you need from each strata to reach a particular quota that is representative of the population

Consecutive Sample

Recruits all people from an accessible population over a certain time period for a specific sample size.

Purposive Sample

Based on knowledge of population you can pick particular people. Really useful when you want sample of experts

*Which type of sampling is most vulnerable to bias?*

A. Convenience sampling

B. Snowball sampling

C. Quota sampling

D. Purposive sampling

*A. Convenience sampling*

Rationale: Although it is the most widely used approach for quantitative researchers, convenience sampling is the most vulnerable to sampling biases.

Types of Probability Sampling

- Simple random sampling

- Stratified random sampling

- Systemic sampling

Simple Random Sampling

- Researcher establishes sampling frame (list of population elements) and elements are numbered and then numbers are randomly drawn to the desired size

- No guarantee of representative sample.

Stratified Random Sampling

- Population is divided into strata and then then randomly selected within each strata.

- Aim is to enhance representativeness (like in quota sampling).

- Not possible if researcher lacks info on strata

Systemic Sampling

Every X number of case from a list.

- Size of population is divided by size of desired sample to obtain sampling interval.

T or F: Stratified random sampling is associated with a larger sampling error but it is more efficient.

*False*

Rationale: Stratified random sampling enhances representativeness; cluster sampling is associated with a larger sampling error but is considered more efficient.

Sample Size

Number of study participants in a study.

- Adequacy is a key determinant of sample quality in quantitative research

- General rule is the larger the sample the better

Power Analysis

Process through which sample size needed can be estimated

Statistical conclusion validity decreases when

sample is too small. You risk gathering data that won't support the hypothesis even if it is correct.

Basic Decision is the use of

new data or existing data.

Approaches to data collection vary along 4 dimensions:

- Structure

- Quantifiability

- Research obtrusiveness

- Researcher objectivity

Examples of available data

- Hospital records

- School records

- Corporate records

- Letters, diaries, meeting minutes, etc

- Photographs

Use of available data is very desirable as collecting new data is extremely labor intensive

Major Types of Data Collection Methods

- Self-reports

- Observation

- Biophysiologic measures

Data needs to...

- have structure

- be quantifiable

- have objectivity

Structured Self-Reports

Data are collected with a formal instrument, such as

- Interview Schedule

- Questionnaire

Interview Schedule

Questions are prespecified but asked orally, either in-person or by telephone

Questionnaire

Prespecified in written form, to be self- administered

Types of Questions in Structured Experiment

- Closed-ended

- Open ended

Advantages of Interviews vs Questionnaires

- Higher response rates

- Appropriate for more diverse audiences

- Opportunities to clarify questions/determine comprehension

- Opportunity to collect supplementary data through observation

*Which of the following would be an advantage of using a questionnaire?*

A. Higher response rates

B. Diversity of audience is not a problem.

C. Lower cost

D. Questions can be clarified if needed.

*C. Lower cost*

Rationale: Because each subject does not have to been seen face-to-face, questionnaires are associated with lower costs that interviews.

Composite Psychosocial Scales

Scales allow for fine discrimination amongst people w/different attitudes, perceptions, or traits

- Likert scales

- Summated rating scales (composite scales)

Likert Scales

- Have several items expressing viewpoints

- On an agree/disagree continuum

- Responses to items are summed up to compute a total scale score

Visual Analog Scale

Used to measure subjective experiences, like pain or nausea

- Measures on a straight line, w/end-points labeled as extreme limits of sensation

Response Set Bias

Biases reflecting the tendency of some people to respond to items in a characteristic way. Has nothing to do w/item content.

- *Social desirability response set bias* (answers in a way that is socially acceptable)

- *Extreme response set* (always answer w/extreme, and leads to distortions)

- *Acquiescence response set* (agree/disagree regardless of content

Self-Reports

- Allow access to info otherwise not available to researchers

- Most serious issues that come up w/these are validity and trustworthiness

- Have to assume that respondents have been candid even though we know that humans have a tendency to present self in most positive light.

Observation

Structured observation of prespecified behaviors

- Can be used to gather info about condition of individuals, verbal/nonverbal communication, activities, environmental conditions, etc.

- Researchers have flexibility in regard to focus of observation (specific or broad), concealment (don't have to let others know they are being observed), and method of recording observations (can be made through senses or can be recorded, etc.)

Category System

Formal system for systematically recording the incidence/frequency of prespecified behaviors/events. Most common approach to structured observation

- Exhaustive system

- Nonexhaustive system

Exhaustive System

*All behaviors* of a specific type are recorded, and each behavior is assigned to one mutually exclusive category

Nonexhaustive System

Only *specific behaviors* are recorded

Rating Scales

On a descriptive continuum. Ratings can occur

- At specific intervals

- Upon occurrence of certain events

- After an observational session

Observational Sampling

- *Time Sampling* (Sampling of time intervals for observation)

- *Event Sampling* (observation of integral events)

Evaluation of Observational Methods

- Excellent method for capturing many clinical phenomena and behaviors

- Potential issue w/reactivity when people are aware they are being observed

- Risk of observational biases from observer

Biophysiologic Measures

Strong on accuracy, objectivity, validity, and precision. Can be cost-effective, but caution is needed for their use and advanced skills are needed for interpretation

- *In Vivo Measurements* (performed directly within or on living organisms, like BP)

- *In Vitro measurements* (performed outside organisms body like urinalysis)

*Which of the following would be classified as an in vivo biophysiologic measure?*

A. Tissue biopsy

B. Blood glucose level

C. Bacterial culture

D. Body temperature

*D. Body temperature*

Rationale: In vivo measures are those performed directly within or on living organisms, such as blood pressure, body temperature, and vital capacity measurement.