Chapter 7 Flashcards
Non-experimental research
describes participants, traits, scores, and other characteristics without direct or active intervention.
Does not assess causal relationships, though it may suggest them in some designs (e.g., causal-comparative)
Characteristics of Descriptive Studies
Describes a phenomenon
Useful when an area is first being investigated
Many studies, descriptive included, have more than one design type incorporated (so only parts of the study may be truly descriptive)
Criteria for Evaluating Descriptive Studies
Conclusions about relationships should be made with caution
Subjects and instrumentation should be well described
Relationship (Association)
The finding that one variable varies systematically with another variable
Relationship are important because:
They identify potential causes of relationships
They identify variables that can be investigated in true experiments
They allow the prediction of one variable from another
Characteristics of Comparative Studies
Investigates the relationship of one variable to another by comparing values of the DV among differing groups (which serve as the IV)
Does not assess causation, because individuals have not been randomly assigned to an intervention
Does allow prediction of the DV as a function of group membership
Criteria for Evaluating Comparative Studies
Subjects, instrumentation, and procedures should be well defined
Identify the criteria for establishing different groups
Do not infer causation from comparative research designs
Graphic presentations should not distort the results
Characteristics of Correlational Studies
Involves two (or more) variables
A scatterplot is constructed for each set of two variables
Pairs of variables are related using correlation coefficients
The size of correlations is limited by the reliability of the variables
Both the x and y variables need to be sampled over a wide range
If large numbers of variables are correlated, some correlations will show up as significant due to chance alone
Large samples may yield significant correlations that have very small effect sizes
Prediction Studies
A correlation between two variables means that one variable can predict the other variable through a procedure known as regression
Prediction in a new sample of subjects is always ________________ than in the sample that was used to derive the original prediction equation
smaller
Factors That Can Limit Prediction
Reliability of the scores
Length of time between criterion and predictor
Complexity of the criterion variable
Multiple Regression Analysis
Uses multiple predictors to predict a single criterion
Yields a multiple correlation coefficient, R
Best when the multiple IVs have only low to moderate correlations with each other
Logistic Regression
Used when the criterion variable (DV) is dichotomous
Criteria for Evaluating Correlational Studies
Causation should rarely be inferred from correlation
The reported correlation should not be higher or lower than the actual relationship
Practical significance should not be confused with statistical significance
The size of the correlation should be sufficient for the use of the results
Prediction studies should report accuracy of prediction for new subjects
Procedures for collecting data should be indicated clearly
Correlational studies that claim explanations should be examined for other influential factors
Causal-Comparative Designs
These designs compare groups of pre-existing groups
The experimenter cannot assign participants to conditions in these studies
Because random assignment is not possible, these designs are essentially correlational, and not experimental
Ex Post Facto Studies
The experimenter takes advantage of a situation that happened in the past
Subjects who experienced different conditions are then compared on one or more DVs
Because random assignment of subjects to conditions does not occur, causation cannot be determined (it is a correlational design)
Subject matching helps reduce subject characteristic threats to validity
Criteria for Evaluating Causal-Comparative and Ex Post Facto Studies
The primary purpose of the research should be to investigate causal relationships when an experiment is not possible
The presumed causal condition should have already occurred in an ex post facto study
Potential extraneous variables should be recognized and considered
Differences between groups being compared should be controlled
Causal conclusions should be made with caution
survey methods
Paper-based, mailed surveys
Web-based, Internet surveys
Telephone-based interviews
Steps in Designing a Survey
Define the purpose and objectives
Identify resources needed and target population
Choose an appropriate survey method
Word questions carefully
Design the survey (how the survey will appear to the respondent)
Develop directions (should accompany the survey proper)
Develop a letter of transmittal (cover letter)
Pilot test!!!
Cross-Sectional Survey Research
Information is collected from one or more samples or populations at one point in time
Cohort Effects
Subjects at different ages may be qualitatively different from one another or in terms of their experiences, and therefore their responses may not be directly comparable
While differences in responses from different age groups are consistent with developmental differences, only ______________________ can truly establish developmental trends over time
longitudinal survey data
Trend Study
A general population (that may change over time) is studied over time, and different samples of subjects from the population are used at each data collection point
Cohort Study
A specific population (that does not change over time) is studied over time, and different samples of subjects from the population are used at each data collection point
Panel Study
A specific population (that does not change over time) is studied over time, and the same sample of subjects is used at each data collection point
Longitudinal Survey Research
Information is collected from the same group of subjects over a specified length of time
Longitudinal Survey Research Types
Trend Study
Cohort Study
Panel Study
Odds Ratio
An indication of the power of the IV as an increase or decrease in the odds of the DV occurring
E-mail Survey:
Subject is sent an e-mail with the survey as an attachment
Web-based Survey
An e-mail directs the subject to a web site that contains the survey
Advantages and Disadvantages of Internet-Based Surveys
Advantages
Reduced cost and time of administration
Responses can be quickly gathered and tabulated
Disadvantages
Responses not equally likely among different racial and SES groups (e.g., those who don’t have Internet access)
Respondents may give biased answers, because they think their responses will not be anonymous
Internet-Based Survey Design Principles
Show only a few questions per screen
Show both question and response categories on the same screen
Avoid excess scrolling
Limit the use of matrix format questions
Direct respondents to simply click on their response to closed-end questions
Use error messages that refer to the specific location needing attention
Use graphics, hypertext, and colors (but also use good web design principles
Password-protect Web-based surveys
Indicate progress toward completing the survey
