Flashcards
Ex Post Facto and Correlational Research
Flashcards do conjunto (43)
- 1*Research conducted after variation in the variable of interest has already been determined in the natural course of events. *Used when randomization and manipulation of variable characteristics is not possible.Ex Post Facto Research
(Latin for after the fact)
Also sometimes called casual comparative - 2What are the two basic designs of ex post facto research?*Begin with subjects differing on an independent variable -try to determine consequences
*Begin with subjects differing on a dependent variable-try to determine the antecendent (cause) - 3A characteristic that a subject possesses before the study beginsAttribute independent variable
- 4Approaches are very similar between these two types of researchEx Post Facto and Experimental
- 5Starts with equivalent groups and exposes them to different treatmentsExperimental Research
- 6Starts with different groups and tries to determine the consequences or antecendents of the differenceEx Post Facto Research
- 7Mistakenly attributing causation based on a relationship between two variablesPost hoc fallacy
- 8To establish a conclusion that X is the cause of Y, you should verify these three things:*Establish that a statistical relationships exists b/t x and y
*Establish that x preceded y in time
*Establish that other factors did not determine y - 9A relationship where two variables have no effect on each other, but are related because of a third variableSpurious Relationship
- 10A relationship where a change in one variable can be predicted from a change in the otherStatistical Relationship
- 11What are the 5 stages of Ex Post Facto Research?*State the problem
*Select your groups to compare
*Choose b/t proactive/retroactive design
*Collect data on independent and dependent variables
*Analyze and interpret data - 12Type of Ex Post Facto design that begins with subjects grouped on the basis of an independent variable (Ex. father present/father not present)Proactive ex post facto research
- 13Type of Ex Post Facto design that seeks possible antecedent causes for a pre-existing dependent variableRetroactive ex post facto research
- 14What are some alternative explanations in ex post facto research?*Common Cause
*Reverse Causality
*Other possible independent variables - 15Strategies used to improve the credibility of ex post facto research provide...Partial Control
- 16What are some of the specific partial control strategies?*Matching
*Homogenous groups
*Build extraneous variables into the design
*Analysis of covariance
*Partial Correlations - 17This partial control strategy is likely to greatly reduce the number of subjects that can actually be used in the final analysisMatching
- 18True or False: Using homogeneous groups reduces external validity of the studyTrue
- 19This partial control strategy adjusts scores on the dependent variable for any initial differences on the extraneous variableAnalysis of Covariance (ANCOVA)
- 202x4x2 higher-order interaction F test is an example of this type of partial control strategy:Extraneous Variables
- 21Investigates the extent to which variables are related and the direction of the relationshipCorrelational Research
- 22How does correlational research differ from ex post facto?*Correlational relates two or more variable measures from the same group of subjects
*Ex post facto compares two or more groups on the same variable - 23In correlational research, the sign + or - indicates what?The direction of the relationship
*Height and weight are positively correlated
*Temperature and heating bills are negatively correlated - 24What do + 1.00 and - 1.00 represent and why?*A perfect positive and perfect negative relationship
*Size of the correlation and coefficient indicates the strength of the relationship b/t the variables. - 25What are the uses of Correlation?*Assessing relationships
*Assessing consistency
*Prediction - 26True or False: Often a correlational study is mainly exploratoryTrue: Identify patterns of relationships based on theory
- 27True or False: Correlations can be used to measure consistencyTrue: They can be used to measure consistency (or lack of) in a wide variety of cases
- 28Reliability (Consistency) of a test can be assessed through:*Correlating test-retest
*equivalent forms
*split-half scores - 29True or false: When you find two variables are correlated, you can use one variable to predict the other.True
*Ex. freshman GPA from high school grades and aptitude tests
(prediction may not hold for every case) - 30What are the stages to designing a correlational study?*Specify the relationship by asking a question about the relationship b/t two variables of interest
*Determine how the constructs will be quantified
*Select the sample to represent the population
*Collect the data and calculate the coefficient of correlations - 31Name the 6 correlation coefficients*Pearson's product moment correlation-r (interval or ratio data)
*Spearman's Rho or Rank- (P)
(data are ranks rather than raw scores)
*Kendall's tau b
(ordinal data-square contingency tables)
*Kendall's tau c
(ordinal data-rectangular contingency tables)
*Phi Coefficient -
(Nominal variables-2x2 contingency table)
*Cramer's V-
(Nominal variables-other than a 2x2 table) - 32Look at Measures of Relationships Table- What are the three types?*Nominal
*Ordinal
*Interval - 33What are the considerations for interpreting a correlation coefficient?*Absolute size and predictive value
*Correlation as effect size (look at Cohen's stuff)
*Assess in relation to other correlations of the same variables
*Statistical significance
*Practical Utility - 34Does correlation indicate a cause and effect relationship?No
- 35This is used to determine the relationship b/t two variables when the third is removedPartial Correlation
- 36A correlational procedure that examines the relationships among several variables, enables researchers to find the best possible weighting of two or more independent variables to yield a maximum correlation with a single dependent variable.Multiple Regression
- 37A statistical procedure related to multiple regression, but differs in that the criterion is a categorical variable rather than a continuous oneDiscriminant analysis
- 38Analyzes the intercorrelations among a large set of measures to identify a smaller number of common factorsFactor analysis
- 39Adds more than one dependent variable to the prediction equationCanonical correlations
- 40An independent variable that an investigator can directly manipulateActive Independent Variable
- 41This is an alternative explanation in Ex Post Facto research, in which the independent and dependent variable are two separate results of a third variableCommon Cause
- 42This is an alternative explanation in Ex Post Facto research, in which the reverse of the suggested hypothesis could also account for the findingReverse Causality
- 43The assumption that a statistically significant correlation also has practice signficanceSignificance Fallacy