Research Methods Chapter 10 & 11

Median Split

A process in which participants are divided at the "middlemost" score of a variable, with 50% of participants falling above, and the 50% below the split.

Bivariate Correlation

Applied to examine the association between two variables and include the Pearson correlation, point-biserial correlation, and phi correlation.

Pearson Correlation

A bivariate correlation to examine the linear association between two quantitative variables.

Scatter Plot

A graph used to illustrate the correlation between two variables, with each point representing each participant's score on variable X and variable Y.

Point-Biserial Correlation

A bivariate correlation to examine the association between a binary variable and a quantitative variable.

Phi Correlation

A bivariate correlation to examine the association between two binary variables.

Restriction of Range

An artifact of limitations of measurement; when a range of scores on a measure are limited or truncated (i.e., cut-off).

Coefficient of Determination

The squared value of the Pearson correlation (r2). It represents the proportion of variance shared between two variables.

Regression Equation

Mathematical representation of the relationship between variables; has an unstandardized and standardized format.

Multiple Regression

An extension of the Pearson correlation used to estimate the relations involving multiple predictors on an outcome.

z Score

A standardization procedure ti ensure that all variables are on the same metric (M = 0, SD = 1) and is calculated by subtracting each person's score from the mean and dividing that by the pooled standard deviation.

Multiple R2 (Squared)

The proportion of percent of variation accounted for by the set of predictors on the outcome (% of how much IV influences the DV).

Experimental Control

Pursued in experimental methods, notable in randomized experiments, to ensure that the conditions or groups are initially equivalent so that the impact of the IV on the DV can be causally determined.

Statistical Control

Pursued in non-experimental methods to compensate for the fact that experimental control was not implemented as an attempt to rule out rival predictors/causes.

Partial Mediation

Path c' (vs. c) coefficient is reduced and still significant.

Full Mediation

Path c' (vs. c) coefficient is reduced to near zero and no longer significant.

Multilevel Model

An extension of multiple regression to estimate the relations of predictors on an outcome if the design involves a nested hierarchy of units.

Multilevel modeling is used interchangeably with what terms?

Multilevel regression modeling and hierarchical linear modeling

Intraclass Correlation

An index of interrater reliability for two or more coders of a quantitative variable used to assess the extent to which participants have more homogenous scores within higher-order grouping units relative to variability of participants scores across groupings in a multilevel model.

Nested

Occurs when data points are grouped inside larger, higher-level units

Structural Equation Model

A statistical technique that overcomes the limitations of a multiple regression analysis by allowing the researcher to estimate and test the strength of relationships among multiple predictors and multiple criterion variables.

Path Model

A type of structural equation model in which predictive relationships involving only measured variables are estimated.

Exogenous Variable

A variable not explained by a determinant or predictor, as postulated in a structural equation model.

Endogenous Variable

A variable explained by a determinant or predictor variable or variables, as postulated in a structural equation model.

Mediation

A third variable that serves as intermediary, to help explicate the chain of processes in a predictive relation.

Multiple Mediation

A model involving more than one mediator variable.

Latent Structural Equation Model

The result of the integration of confirmatory factor analysis and path analysis.

Latent Factor

A factor that has removed measurement error using techniques such as structural equation modeling.

E (Error)

Measurement error in a latent model

D (Disturbance)

Predictive error stemming from an endogenous latent factor not fully being explained by other factors in the model

Goodness-of-Fit Indicies

Degree of similarity between model and data (higher values are desirable).

Badness-of-Fit Indicies

Degree of discrepancy between the model and data (lower values are desirable).

Cross-Lagged Panel Model

Statistical method to determine the temporal direction of relations involving variables using a nonexperimental longitudinal data, and consisting of synchronous correlations, autoregression paths, and cross-lagged paths.

Synchronous Correlation

The nondirectional association of two different constructs assessed at the same time point.

Growth Curve Model

An analysis conducted to determine the longitudinal trajectory or shape of observations for participants measured at multiple time points.

Quadratic Trend

U-shaped trend

Cubic Trend

Two changes in direction of trend (characterized by 3 lines)

Non-Time Varying Predictor

Predictor that does no vary over time (i.e., race)

Time-Varying Predictor

Predictor that varies over time (i.e., attitudes)