LATENT DIFFERENCE SCORE MODELS Bob Vandenberg Terry College of Business Department of Management...

download LATENT DIFFERENCE SCORE MODELS Bob Vandenberg Terry College of Business Department of Management University of Georgia.

of 62

  • date post

    12-Jan-2016
  • Category

    Documents

  • view

    219
  • download

    0

Embed Size (px)

Transcript of LATENT DIFFERENCE SCORE MODELS Bob Vandenberg Terry College of Business Department of Management...

  • LATENT DIFFERENCE SCORE MODELSBob Vandenberg

    Terry College of BusinessDepartment of ManagementUniversity of Georgia

  • OverviewFramework for workshopIntroductionChangeWhy longitudinal?Other longitudinal statistical procedures and shortcomingsLatent difference: Only a T1 and T2 measureNonlatent methods and shortcomingLatent methodsType DR and Type modelsEmpirical examplesObserved variable onlySingle-Indicator ApproachMultiple-Indicator Approach

  • My Primary SourcesMcArdle, J.J., & Nesselroade, J.R. (2014). Longitudinal Data Analysis Using Structural Equation Models. American Psychological Association: Washington, D.C.Particularly Chapter 9 Newsom, J.T. (2015). Longitudinal Structural Equation Modeling: A Comprenhensive Introduction. Routledge: New York.Particularly Chapter 9

  • AssumptionsNeed an understanding of CFA and structural modelsExtension of basic modelsUsed Mplus 7.x (Muthn & Muthn, 1998-2013)Simply my personal choice can also be completed in LISREL, EQS, AMOS, or Mx.Some focus on equations . . .but I am very application oriented and thus, will be presenting lots of empirical examples with syntax

  • IntroductionChangeWhy is it so important?#1 = because the theoretical frameworks of interest to us explicitly or implicitly imply itUnderlying the use of terms such as related to or associated with is a true desire to say as X changes so does Y; or a change in X causes a change in YReality is we neither operationalize change nor use a design that permits true causal inferencesConcern here is with change, not cause

  • IntroductionComplexities in Changes Over TimeMultiple Levelse.g., changes in newcomer information seeking over time.Cross-Levelschanges in P-G fit, P-O fit over time.Multivariaterelationships linking changes in different focal variables.

  • IntroductionResearch DesignsCross-sectional research designCollects data on a variable at one time point.Longitudinal research designCollects data on a variable at multiple time points.Note:Longitudinal research designs are meant to examine changes over time. A prospective study that uses variable A at T1 to predict variable B at T2, with A measured only at T1 and B measured only at T2, is NOT a longitudinal research design.

  • IntroductionWhy longitudinal?McArdle & Nesselroade (2014) give 5 reasons1 = direct identification of intraindividual changeComes in many forms such as level, frequency and amplitude2 = direct identification of interindividual differences in intraindividual changeChange patterns differ meaningfully between individuals

  • IntroductionWhy longitudinal?5 reasons (continued)3 = analysis of interrelations in or among changesE.g., change in commitment impact change in performance?4 = Analysis of determinants of intraindividual changeWhat causes the changes in levels, amplitude and frequency?5 = Analysis of determinants of interindividual differences in intraindividual changeWhat causes differences in the pattern

  • IntroductionLongitudinal Research DesignsMy webcast in 2007 focused on latent growth models where you have 3 or more repeated measuresTodays webcast focuses on using latent models with only two points in timeLatent Difference Score Models

  • Latent DifferenceNonlatent method 1 Subtract T1 from T2 to get a simple difference scoreIf positive spoke of gain; if negative of lossWhile widely used prior to 70s, Chronbach & Furby (1970) heavily criticized it and use dropped off dramatically.Saw a parallel phenomenon in mid-90s with congruence score and Jeff Edwards severe critique

  • Latent DifferenceNonlatent method 1C&F (1970) criticism based on . . .Y1n = yn + e1n and Y2n = yn + e2n where y is same unobserved true scores and e is two random errors then . . . Dyn = Y2n Y1n= (yn + e2n) (yn + e1n)= (yn yn) + (e2n e1n)= e2n e1n= etn ; that is, change is completely composed of random error

  • Latent DifferenceNonlatent method 2Autoregression or Type A modelRegress T2 onto T1 in Step 1 and residual regressed onto control and substantive variables in subsequent steps

  • Latent DifferenceNonlatent method 2Criticism 1: Means assumed to be zeroHowever, in models depicting change, sample means carry useful statistical informationInformation needed to estimate both inter- and intraindividual differences in change or growthCriticism 2: By controlling T1, model eliminates all predictors except those that predict changes in rank order of observations over time.

  • Latent DifferenceGroundwork for latent approachesLain by Nesselroade (1972; 1974)Dont assume exact same true scoreY1n = y1n + e1n and Y2n = y2n + e2n where y1 is unobserved true at T1 and y2 is unobserved true score at T2, then . . .Dyn = Y2n Y1n= (y2n + e2n) (y1n + e1n)= (y2n y1n) + (e2n e1n)= y + (e2n e1n)= y + etn

  • Latent DifferenceFramework for Remainder of SectionUse affective commitmentIllustrate two latent methods conceptuallyType DR: Regression of ChangesType : Regression Change ModelVery simple difference between themWill only use Type DR for empirical examplesEmpirical examplesObserved variable onlySingle-indicator approachMultiple-indicator approach

  • Latent DifferenceType DR: Regression of Changes 1 21 AC1 1 AC4 1 AC 2 24 1 1McArdle & Nesselroade (2014) notation

  • Latent DifferenceType : Regression Change Model 1 21 AC1 1 AC4 1 AC 2 1 0 1McArdle & Nesselroade (2014) notation

  • Latent DifferenceType DR vs. Type Substantively drive decisionIf you feel that T1 variable is a major factor in the change, then Type E.g. Weight loss greater for those heaviest to begin with; resource rich firms can change faster, then firms with lessIf you feel that T1 variable is not a major factor, then Type DRE.g., Most attitude research, doesnt consider T1 to be a factor in the change

  • Latent DifferenceObserved Variables ExampleType DR followed by Type Level 1 is there change?Averages of affective commitment at T1 and T4After this, build on the Type DRLevel 2 theory testinglatent difference in turnover intention as an outcome affective commitment diferenceorg fit as an antecedent to affective commitment change

  • Latent DifferenceType DR Syntax for Level 1Model:CH by AffAvgT4@1;AffAvgT4 on AffAvgT1@1;AffAvgT4@0;CH with AffAvgT1;[CH AffAvgT1 AffAvgT4@0];CH AffAvgT1;

  • Latent DifferenceCH by AffAvgT4@1; 1 21 AC1 1 AC4 1 AC 2 24 1 1

  • Latent DifferenceAffAvgT4 on AffAvgT1@1; 1 21 AC1 1 AC4 1 AC 2 24 1 1

  • Latent DifferenceAffAvgT4@0; 1 21 AC1 1 AC4 1 AC 2 24 1 1

  • Latent DifferenceCH with AffAvgT1; 1 21 AC1 1 AC4 1 AC 2 24 1 1

  • Latent Difference[CH AffAvgT1 AffAvgT4@0]; 1 21 AC1 1 AC4 1 AC 2 24 1 1Not shown but mean of AC4 fixed to 0

  • Latent DifferenceCH AffAvgT1; 1 21 AC1 1 AC4 1 AC 2 24 1 1

  • Latent DifferenceModel OutcomesJust Identified so perfect fitCH BY AFFAVGT4 1.000 0.000 999.000 999.000 AFFAVGT4 ON AFFAVGT1 1.000 0.000 999.000 999.000 CH WITH AFFAVGT1 1 -0.438 0.073 -5.975 0.000 Means AFFAVGT1 1 1.071 0.061 17.431 0.000 CH 2.017 0.065 31.210 0.000

  • Latent DifferenceModel Outcomes (cont.)

    Intercepts AFFAVGT4 0.000 0.000 999.000 999.000 Variances AFFAVGT1 21 1.133 0.093 12.247 0.000 CH 2 1.254 0.102 12.247 0.000 Residual Variances AFFAVGT4 0.000 0.000 999.000 999.000

  • Latent DifferenceType Syntax for Level 1 Model: CH by AffAvgT4@1; AffAvgT4 on AffAvgT1@1; AffAvgT4@0; CH on AffAvgT1; [CH AffAvgT1 AffAvgT4@0]; CH AffAvgT1;

  • Latent DifferenceType : Regression Change Model 1 21 AC1 1 AC4 1 AC 2 1 0 1

  • Latent DifferencesType Outcomes CH BY AFFAVGT4 1.000 0.000 999.000 999.000 CH ON AFFAVGT1 1 -0.387 0.056 -6.845 0.000 AFFAVGT4 ON AFFAVGT1 1.000 0.000 999.000 999.000 Means AFFAVGT1