AMOS – Analysis of Moment Structures Rick Zimmerman, Olga Dekhtyar HIV Prevention Center University of Kentucky Overview Overview of Structural Equation Models (SEM) Introduction to AMOS User Interface AMOS Graphics Examples of using AMOS Predictors of Condom Use using latent variables Structural Equation Models Structural Equation Modeling (SEM) An extension of Regression and general Linear Models Also can fit more complex models, like confirmatory factor analysis and longitudinal data. Structural Equation Modeling Ability to fit non-standard models, databases with autocorrelated error structures time series analysis Latent Curve Models, databases with non-normally distributed variables databases with incomplete data. T-test Bivariate Correlation ANOVA Family Tree of SEM Multi-way ANOVA Multiple Regression Factor Analysis Repeated Measure Designs Path Analysis Growth Curve Analysis Structural Equation Modeling Confirmatory Factor Next Workshop: Analysis November 9 Exploratory Factor Analysis See you there! Latent Growth Curve Analysis Structural Equation Modeling (SEM) Exogenous variables=independent Endogenous variables =dependent Observed variables =measured Latent variables=unobserved Structural Equation Graphs .10 Error : R2 Observed Variable .15 : Loading Latent Variable Example: Condom Use Model Observed variables for Impulsive decision making IDMA1R IDMC1R IDME1R Respondent Sex IDMJ1R SEX1 Impulsive Impulsive Decision Making Legend Observed Variables .15 FRBEHB1 Peer norms about condoms Latent SXPYRC1 ISSUEB1 Condom attitude Variables Loadings Condom Use Example: Condom Use Model Dependent Independent IDMA1R IDMC1R IDME1R IDMJ1R SEX1 Impulsive Independent Legend Observed Variables .15 FRBEHB1 Dependent Latent Variables Loadings SXPYRC1 ISSUEB1 Dependent Dependent Example: Condom Use Model eidm1 eidm2 eidm2 eidm4 IDMA1R IDMC1R IDME1R IDMJ1R SEX1 Impulsive efr1 FRBEHB1 ISSUEB1 Legend Observed Variables .15 Latent Variables Loadings SXPYRC1 eSXYRC1 eiss Example: Condom Use Model eidm1 eidm2 eidm2 eidm4 IDMA1R IDMC1R IDME1R IDMJ1R SEX1 Impulsive efr1 FRBEHB1 ISSUEB1 Legend Observed Variables .15 Latent Variables Loadings SXPYRC1 eSXYRC1 eiss Example: Condom Use Model eidm1 eidm2 eidm2 eidm4 .28 IDMA1R .24 IDMC1R .48 IDME1R .45 IDMJ1R .53 .49 .69 .67 Impulsive -.15 efr1 .15 -.19 Latent Variables Loadings SEX1 -.10 .13 .03 FRBEHB1 .38 Legend Observed Variables -.06 .11 .05 ISSUEB1 SXPYRC1 .15 eSXYRC1 eiss SEM Assumptions A Reasonable Sample Size a good rule of thumb is 15 cases per predictor in a standard ordinary least squares multiple regression analysis. [ “Applied Multivariate Statistics for the Social Sciences”, by James Stevens] researchers may go as low as five cases per parameter estimate in SEM analyses, but only if the data are perfectly well-behaved [Bentler and Chou (1987)] Usually 5 cases per parameter is equivalent to 15 measured variables. SEM Assumptions (cont’d) Continuously and Normally Distributed Endogenous Variables NOTE: At this time AMOS CANNOT handle not continuously distributed outcome variables SEM Assumptions (cont’d) Model Identification P is # of measured variables [P*(P+1)]/2 Df=[P*(P+1)]/2-(# of estimated parameters) If DF>0 model is over identified If DF=0 model is just identified If DF<0 model is under identified Missing data in SEM Types of missing data MCAR Missing Completely at Random MAR Missing at Random MNAR Missing Not at Random Handling Missing data in SEM Listwise Pairwise Mean substitution Regression methods Expectation Maximization (EM) approach Full Information Maximum Likelihood (FIML)** Multiple imputation(MI)** The two best methods: FIML and MI SEM Software Several different packages exist EQS, LISREL, MPLUS, AMOS, SAS, ... Provide simultaneously overall tests of model fit individual parameter estimate tests May compare simultaneously Regression coefficients Means Variances even across multiple between-subjects groups Introduction to AMOS AMOS Advantages Easy to use for visual SEM ( Structural Equation Modeling). Easy to modify, view the model Publication –quality graphics AMOS Components AMOS Graphics draw SEM graphs runs SEM models using graphs AMOS Basic runs SEM models using syntax Starting AMOS Graphics Start Programs Amos 5 Amos Graphics Reading Data into AMOS File Data Files The following dialog appears: Reading Data into AMOS Click on File Name to specify the name of the data file Currently AMOS reads the following data file formats: Access dBase 3 – 5 Microsft Excel 3, 4, 5, and 97 FoxPro 2.0, 2.5 and 2.6 Lotus wk1, wk3, and wk4 SPSS *.sav files, versions 7.0.2 through 13.0 (both raw data and matrix formats) Reading Data into AMOS Example USED for this workshop: Condom use and what predictors affect it DATASET: AMOS_data_valid_condom.sav Drawing in AMOS In Amos Graphics, a model can be specified by drawing a diagram on the 1. To draw an observed variable, click screen "Diagram" on the top menu, and click "Draw Observed." Move the cursor to the place where you want to place an observed variable and click your mouse. Drag the box in order to adjust the size of the box. You can also use in the tool box to draw observed variables. 2. Unobserved variables can be drawn similarly. Click "Diagram" and "Draw Unobserved." Unobserved variables are shown as circles. You may also use in the toolbox to draw unobserved variables. Drawing in AMOS To draw a path, Click “Diagram” on the top menu and click “Draw Path”. Instead of using the top menu, you may use the Tool Box buttons to draw arrows ( and ). Drawing in AMOS To draw Error Term to the observed and unobserved variables. Use “Unique Variable” button in the Tool Box. Click and then click a box or a circle to which you want to add errors or a unique variables.(When you use "Unique Variable" button, the path coefficient will be automatically constrained to 1.) Drawing in AMOS Let us draw: 1 1 1 Naming the variables in AMOS double click on the objects in the path diagram. The Object Properties dialog box appears. • OR Click on the Text tab and enter the name of the variable in the Variable name field: Naming the variables in AMOS Example: Name the variables IDM SEX1 FRBEHB1 ISSUEB1 1 1 eiss efr1 SXPYRC1 1 eSXPYRC1 Constraining a parameter in AMOS The scale of the latent variable or variance of the latent variable has to be fixed to 1. Double click on the arrow between EXPYA2 and SXPYRA2. The Object Properties dialog appears. Click on the Parameters tab and enter the value “1” in the Regression weight field: Improving the appearance of the path diagram You can change the appearance of your path diagram by moving objects around To move an object, click on the Move icon on the toolbar. You will notice that the picture of a little moving truck appears below your mouse pointer when you move into the drawing area. This lets you know the Move function is active. Then click and hold down your left mouse button on the object you wish to move. With the mouse button still depressed, move the object to where you want it, and let go of your mouse button. Amos Graphics will automatically redraw all connecting arrows. Improving the appearance of the path diagram To change the size and shape of an object, first press the Change the shape of objects icon on the toolbar. You will notice that the word “shape” appears under the mouse pointer to let you know the Shape function is active. Click and hold down your left mouse button on the object you wish to re-shape. Change the shape of the object to your liking and release the mouse button. Change the shape of objects also works on twoheaded arrows. Follow the same procedure to change the direction or arc of any double-headed arrow. Improving the appearance of the path diagram If you make a mistake, there are always three icons on the toolbar to quickly bail you out: the Erase and Undo functions. To erase an object, simply click on the Erase icon and then click on the object you wish to erase. To undo your last drawing activity, click on the Undo icon and your last activity disappears. Each time you click Undo, your previous activity will be removed. If you change your mind, click on Redo to restore a change. Performing the analysis in AMOS View/Set Analysis Properties and click on the Output tab. There is also an Analysis Properties icon you can click on the toolbar. Either way, the Output tab gives you the following options: Performing the analysis in AMOS For our example, check the Minimization history, Standardized estimates, and Squared multiple correlations boxes. (We are doing this because these are so commonly used in analysis). To run AMOS, click on the Calculate estimates icon on the toolbar. AMOS will want to save this problem to a file. if you have given it no filename, the Save As dialog box will appear. Give the problem a file name; let us say, tutorial1: Results When AMOS has completed the calculations, you have two options for viewing the output: text output, graphics output. For text output, click the View Text ( or F10) icon on the toolbar. Here is a portion of the text output for this problem: Results for Condom Use Model(see handout) The model is recursive. Sample size = 893 Chi-square=12.88 Degrees of Freedom =3 Maximum Likelihood Estimates FRBEHB1 ISSUEB1 FRBEHB1 ISSUEB1 SXPYRC1 SXPYRC1 <--<--<--<--<--<--- SEX1 SEX1 IDM IDM ISSUEB1 FRBEHB1 Estimate -.28 .30 -.38 -.57 .16 .49 S.E. .09 .08 .11 .10 .05 .04 C.R. -2.98 3.79 -3.29 -5.94 3.42 12.21 P .00 *** *** *** *** *** Standardized Regression Weights: (Group number 1 - Default model) FRBEHB1 ISSUEB1 FRBEHB1 ISSUEB1 SXPYRC1 SXPYRC1 <--<--<--<--<--<--- SEX1 SEX1 IDM IDM ISSUEB1 FRBEHB1 Estimate -.10 .12 -.11 -.19 .11 .38 Results for Condom Use Model Covariances: (Group number 1 - Default model) SEX1 <--> Estimate S.E. C.R. P -.02 .01 -2.48 .01 IDM Correlations: (Group number 1 - Default model) Estimate SEX1 <--> IDM -.08 Label Viewing the graphics output in AMOS • To view the graphics output, click the View output icon next to the drawing area. • Chose to view either unstandardized or (if you selected this option) standardized estimates by click one or the other in the Parameter Formats panel next to your drawing area: Viewing the graphics output in AMOS Unstandardized -.02 .17 Standardized -.08 .25 IDM IDM SEX1 -.57 SEX1 -.19 -.28 -.38 -.10 -.11 .30 .12 .02 FRBEHB1 efr1 FRBEHB1 ISSUEB1 1.94 1 1 .49 efr1 0.15 is the squared multiple SXPYRC1 1 2.80 eSXPYRC1 ISSUEB1 1.36 eiss .16 .06 correlation between Condom use and ALL OTHER variables .38 eiss .11 .15 SXPYRC1 eSXPYRC1 How to read the Output in AMOS See the handout_1 Modification of the Model Search for the better model Suggestions from: 1) theory 2) modification indices using AMOS Modifying the Model using AMOS View/Set Analysis Properties and click on the Output tab. Then check the Modification indices option Modifying the Model using AMOS Modification Indices (Group number 1 - Default model) Covariances: (Group number 1 - Default model) M.I. eiss <--> efr1 Chi-square decrease 9.909 Par Change .171 Parameter increase Modifying the Model using AMOS 2.38, .17 -.02 1.45, .25 IDM SEX1 -.57 -.38 -.28 .30 3.74 5.58 0, 1.94 1 efr1 FRBEHB1 .49 ISSUEB1 1 .16 .17 3.08 SXPYRC1 1 0, 2.80 eSXPYRC1 SEE Handout # 2 for the whole output 0, 1.36 eiss Examples using AMOS Condom Use Model with missing values Confirmatory Factor Analysis for Impulsive Decision Making construct Multiple group analysis How to deal with non-normal data Missing data in AMOS Full Information Maximum Likelihood estimation • View/Set -> Analysis Properties and click on the Estimation tab. • Click on the button Estimate Means and Intercepts. This uses FIML estimation Recalculate the previous example with data “AMOS_data.sav” with some missing values Missing data in AMOS The standardized graphical output. -.10 IDM SEX1 -.18 -.09 -.10 .12 .02 .05 FRBEHB1 efr1 ISSUEB1 .37 eiss .08 .14 SXPYRC1 eSXPYRC1 Missing data in AMOS Example: see the handout #3 Confirmatory Factor Analysis with Impulsive Decision Making scale Need to fix either the variance of the IDM1 factor or one of the loadings to 1. 0, 0, e2 e1 1 1 0, 0, e4 e3 1 IDMA1R IDMC1R IDME1R 1 IDMJ1R 1 idm1 0, Confirmatory Factor Analysis with Impulsive Decision Making scale e1 e2 e3 .30 IDMA1R .26 IDMC1R .55 .51 idm1 e4 .47 IDME1R .69 Multiple Correlation .47 IDMJ1R .69 Factor Loadings Chi-square = 11.621 Degrees of freedom = 2, p=0.003 CFI=0.994, RMSEA=0.042 Confirmatory Factor Analysis with Impulsive Decision Making scale What if want to compare two NESTED models for Impulsive Decision Making Model? 1) error variances equal for all 4 measured variables 2) error variances are different Confirmatory Factor Analysis with Impulsive Decision Making scale: the error variances are the same Need to give names to the error variances, by double clicking on the error variance. The Object properties will appear, click on the Parameter and type the name for the error variance( e1, e2...) in the Variance box. Confirmatory Factor Analysis with Impulsive Decision Making scale 0, e1 0, e2 e1 1 0, e3 0, e4 e2 e3 e4 1 1 1 IDMA1R IDMC1R 1 IDME1R 0, idm1 IDMJ1R Confirmatory Factor Analysis with Impulsive Decision Making scale: error variances are the same Click MODEL FIT , then Manage Models In the Manage Models window, click on New. In the Parameter Constraints segment of the window type “e1=e2=e3=e4” Now there are two nested models Confirmatory Factor Analysis with Impulsive Decision Making scale error variances are the same error variances are different 0, .45 0, .48 0, .48 e1 0, .48 e2 1 e3 1 2.18 IDMA1R 2.44 1.00 IDME1R 1.15 1.50 2.28 IDMJ1R 1.36 0, .19 idm1 Chi-square = 56.826, df=5, p=0.000 0, .47 e2 1 1 2.24 IDMC1R e1 e4 1 0, .58 0, .43 0, .48 e3 1 2.18 IDMA1R 1 2.44 1 2.24 IDMC1R 1.00 e4 IDME1R 1.03 1.48 2.28 IDMJ1R 1.40 0, .19 idm1 Chi-square = 11.621, df=3, p=0.003 Confirmatory Factor Analysis with Impulsive Decision Making scale: error variances are the same Compare Nested Models using Chi-square difference test: Model2( errors the same) Chi-square = 56.826, df=5, p=0.000 Model1 ( errors are different) Chi-square = 11.621, df=3, p=0.003 Chi-squaredifference=56.826-11.621=45.205 df=5-3=2 Chi-squarecritical value=5.99 Significant Model 2 with Equal error variances fits WORSE than Model 1 Confirmatory Factor Analysis with Impulsive Decision Making scale: error variances are the same Nested Model Comparisons Assuming model Error are free to be correct: Model Errors are the same P NFI Delta-1 3 45.205 .000 .026 DF CMIN IFI RFI TLI Delta-2 rho-1 rho2 .026 .032 .032 Multiple group analysis WHY: test the equality/invariance of the factor loadings for two separate groups HOW : 1) test the model to both groups separately to check the entire model 2) the same model by multiple group analysis Example: Do Males and Females can be fitted to the same Condom USE model? Need to have 2 separate data files for each group. data_boys and data_girls. Multiple group analysis • Select Manage Groups... from the Model Fit menu. • Name the first group “Girls”. • Next, click on the New button to add a second group to the analysis. • Name this group “Boys”. • AMOS 4.0 will allow you to consider up to 16 groups per analysis. • Each newly created group is represented by its own path diagram Multiple group analysis • Select File->Data Files... to launch the Data Files dialog box. • For each group, specify the relevant data file name. • For this example, choose the data_girls SPSS database for the girls' group; • choose the data_boys SPSS database for the boys' group. Multiple group analysis The following models fit to both groups (see handout) : Click Model Fit and Multiple Groups. Unconstrained all parameters different inineach This gives a –name to everyare parameter the group model in each group. Measurement weights – regression loadings are the same in both groups Measurement intercepts – the same intercepts for both groups Structural weights – the same regression loadings between the latent var. Structural intercepts – the same intercepts for the latent variables Structural covariates – the same variances/covariance for the latent var. Structural residuals – the same disturbances Measurement residuals – the same errors-THE MOST RESTRICTIVE MODEL Example: Multiple group analysis for Condom use Model 0, .48 0, .65 0, .47 0, .47 0, .47 eidm1 eidm2 eidm3 eidm4 1 2.33 1 2.60 1 2.39 0, .44 0, .39 eidm1 eidm2 eidm3 eidm4 1 2.43 1 2.21 IDMA1R IDMC1R IDME1R IDMJ1R 1 2.41 1 2.36 1 2.40 IDMA1R IDMC1R IDME1R IDMJ1R 1.58 1.41 1.00 1.04 0, .18 1.56 1.45 1.00 1.14 0, .16 Impulsive Impulsive -.64 -.62 -.38 -.28 4.35 2.72 FRBEHB1 4.12 ISSUEB1 0, 2.12 1 efr1 0, .62 FRBEHB1 0, 1.50 1 .40 eiss .11 ISSUEB1 0, 1.81 1 efr1 .62 eiss .26 2.16 SXPYRC1 SXPYRC1 1 1 0, 2.95 eSXPYRC1 UNCONSTRAINED MODEL 0, 1.13 1 3.63 Boys 3.06 0, 2.56 eSXPYRC1 Girls Example: Multiple group analysis for Condom use Model 0, .47 0, .48 eidm1 0, .64 eidm2 1 2.33 IDMA1R 0, .48 eidm3 1 2.60 IDMC1R 0, .63 eidm2 1 2.21 eidm4 1 2.39 IDME1R eidm1 0, .46 1 2.43 0, .43 eidm3 1 2.41 0, .40 eidm4 1 2.36 1 2.40 IDMA1R IDMC1R IDME1R IDMJ1R IDMJ1R 1.57 1.57 1.08 1.08 1.42 1.42 1.00 1.00 Impulsive Impulsive 0, .16 0, .18 -.50 -.45 -.50 -.45 4.35 2.72 FRBEHB1 4.12 3.06 ISSUEB1 FRBEHB1 1 efr1 0, 2.14 1 .40 .11 1 eiss efr1 3.62 ISSUEB1 0, 1.51 0, 1.81 .62 .26 0, 2.95 SXPYRC1 1 eSXPYRC1 Boys eiss 2.16 SXPYRC1 1 Measurement weights 0, 1.12 1 Girls 0, 2.56 eSXPYRC1 Example: Multiple group analysis for Condom use Model see handout Since Measurement Weights model is nested within Unconstrained . Chi-square difference test computed to test the null hypothesis that the regression weights for boys and girls are the same. However, the variances and covariance are different across groups. Example: Multiple group analysis for Condom use Model Chi-squarediff =68.901-65.119=2.282 df=29-26=3 NOT SIGNIFICANT FIT of the Measurement Weights model is not significantly worse than Unconstrained Handling non-normal data: Verify that your variables are not distributed joint multivariate normal Assess overall model fit using the BollenStine corrected p-value Use the bootstrap to generate parameter estimates, standard errors of parameter estimates, and significance tests for individual parameters Handling non-normal data: checking for normality To verify that the data is not normal. Check the Univariate SKEWNESS and KURTOSIS for each variable . • View/Set -> Analysis Properties and click on the Output tab. •Click on the button Tests for normality and outliers Handling non-normal data: checking for normality Assessment of normality Variable min max skew c.r. kurtosis c.r. IDM 1.182 3.727 .381 4.649 .496 3.025 SEX1 1.000 2.000 .182 2.222 -1.967 -11.997 FRBEHB1 1.000 6.000 -.430 -5.245 -.778 -4.748 ISSUEB1 1.000 4.000 -.431 -5.259 -1.387 -8.462 SXPYRC1 2.000 7.000 -.937 -11.436 -.715 -4.360 -3.443 -6.149 Multivariate Critical ratio of +/- 2 for skewness and kurtosis statistical significance of NON-NORMALLITY Multivariate kurtosis >10 Severe Non-normality Handling non-continuous data: Bootstrapping Use Bootstrapping Bootstrapping generates an estimate of the sampling distribution from the available data and computes the p-values and construct confidence intervals. Bootstrapping in AMOS generates random covariance matrices from the sample covariance matrix assuming multivariate normality Handling non-continuous data: Bootstrapping Bootstrapping is useful for estimating standard errors for statistics with complex distributions, for which there is no practical approximate However, Some limitations include: The “population” in nonparametric bootstrapping is merely the researcher’s sample If the researcher’s sample is small, unrepresentative, or the observations are not independent, resembling from it can magnify the effects of these features (see Rodgers, 1999) Bootstrap analyses are probably biased in small samples (just as they are in other methods)—that is, bootstrapping is not a “cure” Handling non-normal data:Bollen-Stine bootstrapping p-value•View/Set -> Analysis Properties and click on the Bootstrap tab. Check Perform bootstrap and Bollen- Stine bootstrap BOLLEN_STINE BOOTSTRAP performed only for dataset without any missing values (see handout #6: amos_data_valid_condom.sav) Handling non-normal data:Bollen-Stine bootstrapping p-value The model fits better than expected in 496 samples out of 500 samples (500-496)/500=0.010 So, p-value=0.01 < 0.05 - Model does not fit to the data very well Handling non-normal data:Bollen-Stine bootstrapping p-value Overall Model Fit: Chi-square=12.88; Degrees of freedom = 3 The expected(mean) value of Chi-square is 2.929. The mean value of Chisquare (2.929) serves as the critical chi-square value against which the obtained chi-square of 12.88 is compared In our example, results from the Bollen-Stine are the same as results for the overall model. Handling non-normal data: Bootstrapping Standard Errors Bootstrapping can be used to evaluate the estimates, by computing the Standard Errors of the estimates UnSELECT Bollen-Stine Bootstrap and Select Percentile Confidence Intervals and Bias-corrected confidence intervals Handling non-normal data: Bootstrapping Standard Errors Estimates using ML Relationship between Condom use and Peer Norms about Condom Bootstrap estimates is 0.487, with S.E.=0.04, Almost the same estimate produced by Bootstrap, 0.488 with S.e=0.042 Handling non-normal data: Bootstrapping Standard Errors 90% Percentile Method 90% Bias Corrected Percentile method Hope to see similar results for the estimates NOTE: BOOTSTRAP option works ONLY with COMPLETE data Handling non-normal data: Bootstrapping Standard Errors NOTE: BOOTSTRAP option works ONLY with COMPLETE data if missing is less than 5% , it is defensible to use LISTWISE deletion Sample size should be reasonably large with 200 for SEMs that contain latent variables ( by Nevitt and Hancock, 1998) Thank You! See you in a week! Upper critical values of chi-square distribution Degree of freedom Chi-square critical value 1 3.841 2 5.991 3 7.815 4 9.488 5 11.070 6 12.592 7 14.067 8 15.507 9 16.919 10 18.307 11 19.675
© Copyright 2024 Paperzz