AMOS Analysis of Moment Structures .


69 views
Uploaded on:
Description
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
Transcripts
Slide 1

AMOS – Analysis of Moment Structures Rick Zimmerman, Olga Dekhtyar HIV Prevention Center University of Kentucky

Slide 2

Overview of Structural Equation Models (SEM) Introduction to AMOS User Interface AMOS Graphics Examples of utilizing AMOS Predictors of Condom Use utilizing inert factors

Slide 3

Structural Equation Models

Slide 4

Structural Equation Modeling (SEM) An expansion of Regression and general Linear Models Also can fit more unpredictable models, as corroborative component examination and longitudinal information.

Slide 5

Structural Equation Modeling Ability to fit non-standard models , databases with autocorrelated blunder structures time arrangement investigation Latent Curve Models , databases with non-ordinarily circulated factors databases with deficient information .

Slide 6

Family Tree of SEM Latent Growth Curve Analysis T-test ANOVA Multi-way ANOVA Repeated Measure Designs Growth Curve Analysis Structural Equation Modeling Multiple Regression Path Analysis Bivariate Correlation Confirmatory Factor Analysis Next Workshop: November 9 See you there! Consider Analysis Exploratory Factor Analysis

Slide 7

Structural Equation Modeling (SEM) Exogenous variables= free Endogenous factors = subordinate Observed factors = measured Latent variables= surreptitiously

Slide 8

.10 : R 2 Error .15 Structural Equation Graphs Observed Variable : Loading Latent Variable

Slide 9

Observed factors for Impulsive basic leadership Peer standards about condoms Condom disposition Example: Condom Use Model Respondent Sex IDMA1R IDMC1R IDME1R IDMJ1R SEX1 Impulsive Decision Making FRBEHB1 ISSUEB1 Legend SXPYRC1 Latent Variables Observed Variables .15 Condom Use Loadings

Slide 10

Dependent Example: Condom Use Model Independent IDMA1R IDMC1R IDME1R IDMJ1R SEX1 Impulsive Independent FRBEHB1 ISSUEB1 Legend SXPYRC1 Latent Variables Observed Variables .15 Dependent Loadings

Slide 11

Example: Condom Use Model eidm1 eidm2 eidm4 eidm2 IDMA1R IDMC1R IDME1R IDMJ1R SEX1 Impulsive FRBEHB1 efr1 eiss ISSUEB1 Legend SXPYRC1 Latent Variables Observed Variables eSXYRC1 .15 Loadings

Slide 12

Example: Condom Use Model eidm1 eidm2 eidm4 eidm2 IDMA1R IDMC1R IDME1R IDMJ1R SEX1 Impulsive FRBEHB1 efr1 eiss ISSUEB1 Legend SXPYRC1 Latent Variables Observed Variables eSXYRC1 .15 Loadings

Slide 13

.28 .24 .48 .45 .03 .05 .15 Example: Condom Use Model eidm1 eidm2 eidm4 eidm2 IDMA1R IDMC1R IDME1R IDMJ1R .49 .69 .67 SEX1 - .06 .53 Impulsive - .19 - .10 - .15 .13 FRBEHB1 efr1 eiss ISSUEB1 .11 .38 Legend SXPYRC1 Latent Variables Observed Variables eSXYRC1 .15 Loadings

Slide 14

SEM Assumptions A Reasonable Sample Size a decent general guideline is 15 cases for every indicator in a standard common slightest squares different relapse investigation . [ " Applied Multivariate Statistics for the Social Sciences" , by James Stevens] specialists may go as low as five cases for each parameter gauge in SEM examinations, however just if the information are flawlessly all around carried on [ Bentler and Chou (1987)] Usually 5 cases for each parameter is identical to 15 measured factors .

Slide 15

SEM Assumptions (cont\'d) Continuously and Normally Distributed Endogenous Variables NOTE: At this time AMOS CANNOT deal with not consistently dispersed result factors

Slide 16

SEM Assumptions (cont\'d) Model Identification P is # of measured factors [P*(P+1)]/2 Df=[P*(P+1)]/2-(# of assessed parameters) If DF>0 model is over distinguished If DF=0 model is simply recognized If DF<0 model is under distinguished

Slide 17

Missing information in SEM Types of missing information MCAR Missing Completely at Random MAR Missing at Random MNAR Missing Not at Random

Slide 18

Handling Missing information in SEM Listwise Pairwise Mean substitution Regression strategies Expectation Maximization (EM) approach Full Information Maximum Likelihood (FIML)** Multiple imputation(MI)** The two best techniques: FIML and MI

Slide 19

SEM Software Several unique bundles exist EQS, LISREL, MPLUS, AMOS , SAS, ... Give at the same time general trial of model fit individual parameter gauge tests May look at all the while Regression coefficients Means Variances even over numerous between-subjects bunches

Slide 20

Introduction to AMOS

Slide 21

AMOS Advantages Easy to use for visual SEM ( Structural Equation Modeling). Simple to adjust, see the model Publication –quality design

Slide 22

AMOS Components AMOS Graphics draw SEM diagrams runs SEM models utilizing charts AMOS Basic runs SEM models utilizing grammar

Slide 23

Starting AMOS Graphics Start  Programs  Amos 5  Amos Graphics

Slide 24

Reading Data into AMOS File  Data Files The accompanying discourse shows up:

Slide 25

Reading Data into AMOS Click on File Name to indicate the name of the information document Currently AMOS peruses the accompanying information record 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 documents , renditions 7.0.2 through 13.0 (both crude information and framework groups)

Slide 26

Reading Data into AMOS Example USED for this workshop: Condom utilize and what indicators influence it DATASET: AMOS_data_valid_condom.sav

Slide 27

To draw a watched variable, click "Diagram" on the top menu, and snap " Draw Observed ." Move the cursor to the place where you need to put a watched variable and snap your mouse. Drag the container keeping in mind the end goal to alter the extent of the crate. You can likewise use     in the tool kit to draw watched factors. 2. In secret factors can be drawn also. Click "Diagram" and " Draw Unobserved ." Unobserved factors are appeared as circles. You may likewise use      in the tool stash to draw in secret factors. Attracting AMOS In Amos Graphics, a model can be determined by drawing an outline on the screen

Slide 28

Drawing in AMOS To draw a way, Click " Diagram " on the top menu and snap " Draw Path ". Rather than utilizing the top menu, you may utilize the Tool Box catches to draw bolts ( and ).

Slide 29

Drawing in AMOS To attract Error Term to the watched and in secret factors. Utilize " Unique Variable " catch in the Tool Box. Snap and afterward click a case or a hover to which you need to include mistakes or a novel factors. (When you utilize " Unique Variable " catch, the way coefficient will be naturally compelled to 1.)

Slide 30

Drawing in AMOS Let us draw:

Slide 31

Naming the factors in AMOS double tap on the articles in the way outline. The Object Properties discourse box shows up. Alternately Click on the Text tab and enter the name of the variable in the Variable name field:

Slide 32

Naming the factors in AMOS Example: Name the factors

Slide 33

Constraining a parameter in AMOS The size of the idle variable or change of the inactive variable must be settled to 1 . Double tap on the bolt amongst EXPYA2 and SXPYRA2 . The Object Properties exchange shows up. Tap on the Parameters tab and enter the esteem " 1 " in the Regression weight field:

Slide 34

Improving the presence of the way outline You can change the presence of your way graph by moving articles around To move a protest, tap on the Move symbol on the toolbar. You will see that the photo of a bit of moving truck shows up underneath your mouse pointer when you move into the drawing range. This tells you the Move capacity is dynamic. At that point snap and hold down your left mouse catch on the question you wish to move. With the mouse catch still discouraged, move the protest where you need it, and let go of your mouse catch. Amos Graphics will naturally redraw all associating bolts.

Slide 35

Improving the presence of the way outline To change the size and state of a protest, first press the Change the state of articles symbol on the toolbar. You will see that the word " shape" shows up under the mouse pointer to tell you the Shape capacity is dynamic. Snap and hold down your left mouse catch on the question you wish to re-shape. Change the state of the protest your preferring and discharge the mouse catch. Change the state of articles likewise deals with two-headed bolts. Take after similar method to alter the course or bend of any twofold headed bolt.

Slide 36

Improving the presence of the way chart If you commit an error , there are constantly three symbols on the toolbar to rapidly safeguard you out: the Erase and Undo capacities. To delete a question , basically tap on the Erase symbol and afterward tap on the protest you wish to eradicate. To fix your last drawing movement , tap on the Undo symbol and your last action vanishes. Every time you click Undo, your past movement will be evacuated. On the off chance that you alter your opinion , tap on Redo to reestablish a change.

Slide 37

Performing the investigation in AMOS View/Set ® Analysis Properties and tap on the Output tab. There is additionally an Analysis Properties symbol you can tap on the toolbar. In any case, the Output tab gives you the accompanying alternatives:

Slide 38

Performing the investigation in AMOS For our illustration, check the Minimization history, Standardized evaluations , and Squared various relationships boxes. ( We are doing this on the grounds that these are so ordinarily utilized as a part of examination ). To run AMOS , tap on the Calculate gauges symbol on the toolbar. AMOS will need to spare this issue to a document. on the off chance that you have given it no filename, the Save As exchange box will show up. Give the issue a record name; let us say, tutorial1 :

Slide 39

Results When AMOS has finished the estimations, you have two alternatives for survey the yield: content yield , representation yield . For content yield, tap the View Text ( or

Recommended
View more...