Ron Smith's Teaching Notes on ...
Updated Spring 2016 as a supplement to Professor Smith's textbooks
Strategic Planning for Public Relations and Becoming a Public Relations Writer, (Routledge/Taylor and Francis).
Public relations practitioners use several different research methodologies in their strategic planning, both the formative research that shapes a project or campaign, and the evaluative research that assess its outcomes and impact.
The systematic, objective and unbiased gathering of information
Primary Research: New research undertaken to address a specific research question or hypothesis
Secondary Research: Existing research that may be re-examined to address a specific research question or hypothesis
Formal Research: Research that enjoys a level of sophistication that allows it to be replicated and subject to an analysis of its reliability and validity
Informal Research: Research that does not enjoy a high degree of reliability and validity, often because of its use of nonprobability sampling
Reliability: Measure of data that is consistent, stable & dependable. Extent to which a particular technique is likely to yield the same or similar results. Repeatability of a research finding
Validity: Extent to which a measurement adequately reflects the real meaning of the concept under study. Technique actually measures what it seems to be measuring
Population: The public of interest to the researcher
Census: Gathering information from every member of the population
Sampling: Gathering information from a subgroup of the target population that represents and reflects the larger body
Optimal Sample Size: Nonexistent; depends on purpose; no optimum percentage of population. 384 often optimal number, with 4% plus/minus error. More not necessarily better
Extrapolation: Extending to the entire population research findings based on a sample of the population
Two Type of Research Methods
A research method that uses flexible & open-ended questioning, often with a small number of participants/respondents, and that cannot be extrapltated to large populations
Examples of Qualitative Research
- Field Observation
- Focus Group
- Intensive (aka In-Depth) Interview
- Case Study
A research method that uses standardized & closed-ended questionins, often with a large number of participants/respondents, and that generally can be extrapolated to large populations
Examples of Quantitative Research
- Content Analysis
Steps in Focus Group Research
- Define problem
- Select sample
- Determine number of groups necessary
- Prepare study mechanics
- Prepare focus group materials
- Conduct session
- Analyze data
- Prepare summary report
Steps in Survey Research
- Define problem
- Design the study (# participants, margin of error, response rate, logistics)
- Select sample
- Select type of administration (face-to-face, telephone, self-administer, group administer, mail, web-based)
- Develop questionnaire
- Pilot test
- Administer questionnaire
- Data processing
- Data analysis
- repare summary report
Nonprobability. Every member of the population does not have an equal chance of being selected for the sample; some techniques are more haphazard than others
Probability. Every member of the population has an equal chance of being selected for the sample
Nonprobability Sampling Techniques
Convenience Sample draws subjects because they are readily available, such as people walking through a shopping mall, students enrolled in a particular course, or participants in so-called (but not really) "random" street-corner interviews by reporters. This type of sampling generally is unreliable as an indicator of the entire population.
Volunteer Sample uses subjects who ask to be included in the research study, such as blog readers who respond to a questionnaire online, or radio talk show listeners who call into the program. Self-selected respondents generally represent extreme views pro or con.
Purposive/Judgment Sample includes research subjects chosen simply because they have a certain characteristic. For example, they may be people who were patients at a particular hospital, or they may drive a particular brand of automobile. Beyond that, however, they are not representative of an entire population of hospital patients or motorists.
Snowball Sample begins with a small group of individuals who are asked to identify others to participate in the research. This technique often is used for populations where member lists do not exist and whose members are hard to identify, at least to outsiders.
Quota Sample is an attempt to be more representative of the population. It involves the selection of subjects to fit a predetermined percentage. For example, if 40 percent of a company's employees have college degrees, then 40 percent of the sample in a study of employee attitudes should have college degrees. The limitation of quota sampling is that it gives no reasonable certainty that findings will be representative of the larger population.
Using Nonprobability Sampling
Need quick data
Need only general data
Working with low budget
Can tolerate high margin of error
Probability Sampling Techniques
Simple Random Sample is the basic type of probability sampling, exemplified by lottery-style drawings such as pulling names from a hat or numbers from a box or using a computer-generated list of random numbers corresponding with a list of members of a population. The major advantage of random sampling is that it produces a sampling in which everyone in the population has an equal chance of being selected. The two major disadvantages of simple random sampling are that the technique requires a sampling frame (an actual list of members of the population, who are assigned numbers), and that it does not necessarily produce a representative sampling if significant subsets exist within the population (thus it works best with a homogenous population).
Systematic Sample involves selection of subjects spaced at equal intervals, such as every 10th name on a membership roster. This method sometimes is combined with the simple random sampling technique by using a random selection for the initial selection. For example, a random start between one and 10, then every 10th item (so if the random start is 6, the subsequent items would be 16, 26, 36, and so on).
Stratified Sample is a modification of random and systematic techniques that involves a ranking of elements on a list, such as four subset lists (for example, freshmen, sophomores, juniors, and seniors) for a study of college students. By stratifying the list according to important demographic characteristics, the researcher is able to ensure a greater likelihood the sample will be representative of the entire population. Stratified sampling may be proportionate, with research sizes based on their proportion in the population; or it may be disproportionate (also called weighted) if particular attention is given to underrepresented members of a population. An example of weighting a sample would be a study of university students that might use every 20th in-state student but every 5th out-of-state student, in order to generate a large-enough number of out-of-state respondents to give an acceptably low margin of error for that subgroup of the student population.
Cluster Sample is used when the researcher can't readily generate a list of the entire population but can obtain listings of particular groups within the population. For example, research on students at every college in the state would have a difficult time putting together a complete sampling frame listing every student at every college. By using the cluster technique instead, the researcher would identify several schools (either randomly or through a stratified technique based on criteria such as size or location). For example, the list of college might be separated into two categories: private and public. Each of these categories might be subdivided into three groups: urban, suburban, rural. Then a number of schools would be randomly selected from each of the six groups. Finally the researchers would select individual students at those particular colleges (randomly, systematically, or through stratified techniques).
Using Probability Sampling
Need to generalize to entire population
Need reliable data
Driving important decisions
Working with adequate time & budget