KEY CONCEPTS AND IDEAS IN CHAPTER 4
[NOTE: Numbers in parentheses refer to pages in the fifth edition of Research Methods in Psychology by Shaughnessy, Zechmeister, and Zechmeister (2000), where these concepts and ideas are more fully defined.]
accidental sample A type of nonprobability sample that results when availability and willingness to respond are the overriding factors used in selecting respondents; generally low in representativeness. (153)
actuarial prediction A prediction of people's typical or average behavior based on reliable correlations between variables (e.g., predicting students' college GPA based on SAT scores). (145)
biased sample A sample in which the distribution of characteristics is systematically different from that of the parent population. (151)
correlation A correlation exists when two different measures of the same people, events, or things vary together; the presence of a correlation makes it possible to predict values on one variable by knowing the values on the second variable. (129)
correlation coefficient A statistic that indicates how well two measures vary together; absolute size ranges from 0.0 (no correlation) to 1.00 (perfect correlation); direction of covariation is indicated by the sign of the coefficient, a plus (+) indicating that both measures covary in the same direction and a minus (-) indicating that the variables vary in opposite directions. (129)
correlational research Research which has the goal of identifying predictive relationships among naturally occurring variables. (128)
cross-sectional design A survey research design in which one or more samples of the population is selected and information is collected from the samples at one time.
element Each member of the population of interest. (151)
interviewer bias Occurs when the interviewer tries to adjust the wording of a question to "fit" the respondent or records only selected portions of the respondent's answers.
longitudinal design A survey research design in which the same sample of respondents is interviewed more than once. (166)
margin of error In survey research, an estimate of the difference between a result obtained from a sample (e.g., the sample mean) and the corresponding true population value (e.g., population mean). (139)
nonprobability sampling A sampling procedure in which there is no way to estimate the probability of each element's being included in the sample; two common types are accidental sampling and purposive sampling. (153)
population The set of all the cases of interest. (150)
probability sampling A sampling procedure in which the probability that each element of the population will be included in the sample can be specified. (153)
purposive sample A type of nonprobability sample in which the elements to be included in the sample are selected by the investigator on the basis of special characteristics of the respondents. (154)
representativeness A sample is representative to the extent that it has the same distribution of characteristics as the population from which it was selected; our ability to generalize from sample to population is critically dependent on representativeness. (151)
response bias A threat to the representativeness of a sample which occurs when some participants selected to respond to a survey systematically fail to complete the survey (e.g., due to failure to complete a lengthy questionnaire or to return a phone call). (158)
sample Something less than all the cases of interest; in survey research, a subset of the population actually drawn from the sampling frame. (151)
sampling frame A specific listing of all the members of the population of interest; an operational definition of the population. (151)
selection bias A threat to the representativeness of a sample which occurs when the procedures used to select a sample result in the over- or underrepresentation of a significant segment of the population. (151)
simple random sample A type of probability sample in which each possible sample of a specified size in the population has an equal chance of being selected. (155)
social desirability The pressures on survey respondents to answer as they think they should respond in accordance with what is most socially acceptable, and not in accordance with what they actually believe. (170)
stratified random sample A type of probability sample in which the population is divided into subpopulations called strata, and random samples are drawn from each of these strata. (155)
successive independent samples design A survey research design in which a series of cross-sectional surveys is done and the same questions are asked of each succeeding sample of respondents. (163)
The Nature of Correlations (129)
Correlation and Causality (131)
- A correlation assesses the extent to which two variables covary; the correlation coefficient is a quantitative index of the direction and magnitude of this relationship.
- The direction of a correlation coefficient can be negative or positive.
- The magnitude of a correlation coefficient ranges from -1.0 (a perfect negative relationship) to +1.0 (a perfect positive relationship); a correlation coefficient of 0.0 indicates no relationship.
- Scatterplots are graphical displays of the relationship between two variables.
Path Analysis (132)
- When two variables are related (correlated), we can make predictions for the variables; however, we cannot make inferences about the cause of the relationship.
- When a relationship between two variables can be explained by a third variable, the relationship is said to be "spurious."
Questionnaires as Instruments (135)
- Path analysis is a statistical technique used to help researchers understand the potential causes of correlational relationships.
- Mediators are variables that help to explain the relationship between two variables.
- A moderator variable affects the strength and direction of the relationship between two variables.
- Although path analysis helps researchers to interpret correlational studies, they cannot make definite causal statements about the relationships between variables.
Margin of Error (138)
- Most survey research relies on questionnaires to measure variables.
- Demographic variables describe the characteristics of people who are surveyed.
- Self-report scales are used to assess people's judgments or to identify individual differences among respondents.
- Two methods to develop individual differences scales are the Likert method and factor analysis.
- The accuracy and precision of questionnaires requires expertise and care in their construction.
Reliability and Validity (141)
- Although survey researchers select a sample to represent a population, they recognize that it is unlikely the sample will exactly describe the population due to sampling error.
- The margin of error represents a quantitative estimate regarding the difference between sample results on a measure and the likely population value.
- Although the margin of error indicates how well survey results may describe a population, whether these results can be interpreted depends on the quality of the survey-research methods.
Predictions and Decisions (144)
- Reliability refers to the consistency of measurement, and is frequently assessed using the test-retest reliability method.
- Reliability is increased by including many similar items on a measure, by testing a diverse sample of individuals, and by using uniform testing procedures.
- Validity refers to the truthfulness of a measure: Does it measure what it intends to measure?
- Construct validity represents the extent to which a measure assesses the theoretical construct it is designed to assess; construct validity is determined by assessing convergent validity and discriminant validity.
Uses of Surveys (146)
- Actuarial prediction relies on correlational data to make predictions about people's average behavior.
- Predictions are improved when multiple sources of information are used.
- Statistical predictions, based on correlations, have been shown to be superior to predictions based on clinical judgment.
Characteristics of Surveys (148)
- Survey research is used to assess people's thoughts, opinions, and feelings.
- Surveys can be specific and limited in scope or more global in their goals.
- The best way to determine whether a survey is biased is to examine the survey procedures and analyses.
Sampling Techniques (149)
- Survey research involves selecting a sample (or samples) and using a predetermined set of questions.
Basic Terms of Sampling (149)
- Careful selection of a survey sample allows researchers to generalize findings from the sample to the population.
Approaches to Sampling (152)
- The identification and selection of elements that will make up the sample is at the heart of all sampling techniques; the sample is chosen from the sampling frame, or list of all members of the population of interest.
- Researchers are not interested simply in the responses of those surveyed; instead, they seek to describe the larger population from which the sample was drawn.
- The ability to generalize from a sample to the population depends critically on the representativeness of the sample.
- A biased sample is one in which the characteristics of the sample are systematically different from the characteristics of the population.
- Selection bias occurs when the procedures used to select a sample result in the overrepresentation or underrepresentation of some segment(s) of the population.
Survey Methods (157)
- Two approaches to selecting a survey sample are nonprobability sampling and probability sampling.
- With nonprobablity sampling techniques, accidental sampling and purposive sampling, there is no guarantee that every element in the population has an equal chance of being included in the sample.
- Accidental sampling relies on individuals' availability and willingness to respond to a survey; in purposive sampling, researchers select elements for the sample based on their special characteristics.
- With the probability sampling techniques, simple random sampling and stratified random sampling, researchers can estimate the likelihood their findings from the sample differ from the population.
- In simple random sampling each element of the population has an equal chance of being included in the sample; in stratified random sampling, the population is divided into subpopulations (strata), and random samples are drawn from the strata.
- Probability sampling is the method of choice for obtaining a representative sample.
Mail Surveys (158)
- Three methods for obtaining survey data are mail surveys, personal interviews, and telephone interviews.
Personal Interviews (160)
- Although mail surveys are quick and convenient, they may have the problem of response bias when individuals fail to complete and return the survey.
- Due to response bias, the final sample for the survey may not represent the population.
Telephone Interviews (161)
- Although costly, personal interviews allow researchers to gain more control over how the survey is administered.
- Interviewer bias occurs when survey responses are recorded inaccurately or when interviewers guide individuals' responses.
Survey-Research Designs (161)
- Despite some disadvantages, telephone interviews have become the method of choice for completing brief surveys.
Cross-Sectional Design (162)
- The three types of survey design are the cross-sectional design, successive independent samples design, and the longitudinal design.
Successive Independent Samples Design (163)
- In the cross-sectional design one or more samples are drawn from the population(s) at one point in time.
- Cross-sectional designs allow researchers to describe the characteristics of a population or the differences between two or more populations, and researchers can make predictions based on the correlational survey data.
Longitudinal Design (166)
- In the successive independent samples design, different samples of respondents from the population complete the survey over a time period.
- The successive independent samples design allows researchers to study changes in a population over time.
- The successive independent samples design does not allow researchers to infer how individual respondents have changed over time.
- A problem with the successive independent samples design occurs when the samples drawn from the population are not equally representative of the population
- that is, samples are not comparable.
Correspondence between Reported and Actual Behavior (169)
- In the longitudinal design, the same respondents are surveyed over time in order to examine changes in individual respondents.
- Because of the correlational nature of survey data, it is difficult to identify the causes of individuals' changes over time.
- As people drop out of the study over time (respondent mortality), the final sample may no longer be comparable to the original sample or represent the population.
- Individuals' survey responses may be influenced by the fact that they have completed the survey on more than one occasion.
- Survey research involves reactive measurement because individuals are aware their responses are being recorded.
- Social desirability refers to pressure respondents sometimes feel to respond as they "should" believe rather than how they actually believe.
- Researchers can assess the accuracy of survey response by comparing these results with archival data or behavioral observations.