Category: AMOS

Why use SPSS AMOS for Structural Equation Modelling (SEM)?

Analyzing structural relationships is made feasible with the help of Structural Equation Modelling. It is the amalgamation of factor analysis and regression analysis. These techniques are used to study the relationship between structural variables and constructs in latent variables. This method is preferred by researchers because of its ability to perform multiple operations in a single analysis. 

SEM is applied in building complex models using latent variables. A graphical or programming interface is necessary to observe and obtain variables, such as SPSS AMOS software. 

Characteristics of efficient SEM

  1. SEM facilitates controlling and understanding of the analyzes.
  2. Assumptions derived from the different analysis techniques are transparent in nature and could be tested.
  3. Creative graphical user interface is employed and it allows rapid model debugging.
  4. Model fit and individual parameter estimate tests are run alongside the process.
  5. Models involving estimation of the relationship between latent variables are kept free from errors that could arise in measurement.
  6. Models are customized to fit using different database of error structures that are autocorrelated.
  7. Multiple linear models can be fit under a single unifying framework.

About SPSS AMOS

Being an additional SPSS module, SPSS AMOS is specially used for performing structural equation modelling, path analysis, confirmatory factor analysis (CFA). AMOS is programmed to represent SPSS derived results in visual models, during structural equation modelling.

Attributes of  SPSS AMOS

  1. Simple drawing tools are used for drawing models.
  2. Computations regarding SEM are quickly resolved.
  3. Results obtained by both graphical and non-graphical applications.
  4. Various methods are used to calculate SEM coefficients. 
  5. Manual drawing of model is allowed.

How SEM is executed by using SPSS AMOS?

  • Attach data – By choosing suitable file name, data to be used for SEM analysis is attached in the SPSS AMOS software. 
  • Represent cause-effect relationship – Using a single headed arrow to draw the cause-effect relationship between the unobserved and observed variables. 
  • Name the variable – Using object properties in a graphical window, the variables in the AMOS is named. 

A model that is consistent in its theoretical understandings and parameter estimations using respective statistics is significant for research purposes. It falls on the researchers to rule out occurring model alternatives by efficiently testing and identifying fitting models. Hence, SEM using SPSS AMOS alleviates that and contributes to better research knowledge.

Purpose of using SPSS AMOS to obtain SEM results

SEM is conducted using AMOS specifically for the following advantages:

  1. The number of variables used and the variables used for the purpose of SEM analysis can be studied using the text output variable summary.
  2. The observed variables and unobserved variables are differentiated from the model.
  3. To ensure the normal distribution of the SEM model, different tests are used.
  4. These tests derived outputs with normality of data.
  5. Proper estimates of different effects that have on data is provided in output.
  6. The different tests performed on regression weight, direct, indirect, total effect affecting standardized loading of factor, residua, correlation and covariance to study effects and produce appropriate results.
  7. The statistics of the model are ensured to fit the desired results.
  8. Complications arising while drawing the model are prompted by error messages or failure to deliver the output.

Merits of SEM 

SEM involves using confirmatory path analysis which comes to the rescue by avoiding possible mistakes. Here, factors are specified exactly in numbers and loading patterns.

Since SEM uses different analysis techniques, it facilitates examination of more than one at a time. It replaces EFA and Regression by considering potential measurement errors too.

A sneak peek to performing structural equation modeling using AMOS

Research often involves testing the relationship between variables in a hypothesis. While various quantitative techniques can be used for this purpose, it is structural equation modeling(SEM) approach that provides a visual & easy to interpret display of the causal relationship between the variables. 

Structural equation modeling, a combination of factor and multiple regression analysis, is a multivariate statistical analysis technique that evaluates the structural relationships between latent constructs and measured variables. Structural equation modeling is of two types.

  1. Measurement model 

This type of model represents the theory specifying how the measured variables group together to demonstrate the theory.

 

  • Structure model 

 

Here, the theory defining how the constructs are related to other constructs are represented. 

In research, SEM technique depends on the popular statistical software known as Analysis of Moment Structures (AMOS). This is because AMOS produces tabular outputs and graphic models using user-friendly tool. However, prior to performing SEM, it is a must to consider few assumptions such as:

  • Linearity – There should be a linear relationship between the endogenous and exogenous variables. 
  • Outlier – Since outlier impacts the model significance, the data should be free of outliers.
  • Multivariate normal distribution – Maximum likelihood approach is used for multivariate distribution. Also, the small changes in the multivariate results in the larger difference in chi-square test. 
  • Sequence – There exists a cause and effect relationship between the exogenous and endogenous variables.  
  • Model identification – The models must be exactly or over-identified as the under-identified models aren’t considered. 
  • Uncorrelated error terms – The error terms are uncorrelated with other variable error terms. 
  • Non-spurious relationship – This implies that the observed variance should be true. 

If all the above-mentioned assumptions hold good for SEM, AMOS then continues with the structuring equation modeling process by assuming that the data has been modeled. Some of the methods used by AMOS to perform SEM include:

  1. Generalized least squares – This method estimates the coefficients in the linear regression model if there exists correlation amongst the residuals. 
  2. Unweighted least squares – This approach estimates residual errors to access the conditional mean. 
  3. Browne’s asymptotic distribution free – This type of method is largely recommended when samples containing the non-normal data and SEM involves analysis of covariance structure. 

Model construction in AMOS

On determining the type of model, the next step is to run AMOS by clicking ‘start’ menu and choosing the ‘AMOS graphic’ option. When the AMOS starts running, a window known as ‘AMOS graphic’ appears, in which the user can manually draw the SEM model.

  • Data input – Choose the file name from the data file option and attach the data in AMOS for further SEM analysis. The user can also select this option by clicking the ‘select data’ icon. 
  • Observed variable – Use rectangle icon and draw the observed variables.
  • Unobserved variable – Deploy circle icon and draw the unobserved variables. 
  • Covariance – To denote the covariance between variables, select a double-headed arrow.
  • Cause-effect relationship – To establish the relationship between the observed and unobserved variable, use single-headed arrow in AMOS.  
  • Naming the variable – It is important to determine the variables to work with them precisely. Click on the variable in the graphical window, select ‘object properties’ option and name the variables in the AMOS. 
  • Error term – The error term appears next to the unobserved variable and is often used to draw the latent variable. 

After running the analysis, the outputs are displayed on the graphic window. However, the graphic window will display only the error term weights, standardized & unstandardized regressions. Some of the results produced by AMOS include:

  • Variable summary 

AMOS and the text output variable provides the option of viewing how many variables and which variables have been used for SEM analysis process. The user can also the number of observed and unobserved variables present in the model. 

  • Accessing the normality 

In the SEM model, data must be normally distributed. AMOS provides skewness, text outputs, Mahalanobis d-squared test, Kurtosis and also gives information about the normality of the data.

 Modification index 

The modification index describes the reliability of the path in the SEM model. If the modification index value is huge, then more paths can be added to the SEM model.

  • Estimates 

The estimate option in the AMOS text output will provide the output for regression weight, residual, standardised loading factor, covariance, indirect effect, direct effect, correlation, total effect, and many more. 

  • Error message 

If there is any error in the SEM model drawing process, then AMOS will either give an error message or will not calculate the result. 

  • Model fit 

The model fit will provide the result for goodness fit model statistics and present goodness fit indexes such as RMR, GFI, BCI, TLI, RMSER, and many more.

 Additionally, AMOS enables the functioning of SEM analysis and makes it easy to arrive at the statistics (where direct measurements are not possible). 

How beneficial is using AMOS for data analysis?

AMOS can be defined as a module of SPSS that offers the user with a user interface in order to perform data analysis and structural equation modelling. In other words, AMOS can be described as statistical software based on windows and is used to perform various tasks related to data analysis. Using AMOS one can perform various analysis that include – data analysis, path analysis, deriving longitudinal data models, casual models etc. The researcher can use AMOS to interpret the structures of multiple data sets, analyse the collected data from a population set, analyse different models at the same time, calculate the test statistics automatically and identify the nested models. Hence, we can say it is beneficial to use AMOS for data analysis. Various Statistics consultation firms are there, that help the researchers in the process of data analysis using AMOS.

AMOS provides the researchers with a user-friendly and powerful software called Structural Equation Modeling (SEM). Using this software, the researcher can perform multiple realistic models, instead of developing multiple statistics and regression models. Using AMOS, one can estimate, specify and access their model in the form of a diagram and define the relationship between the variables. This helps the researcher in analysing and testing the data for validity and reliability. AMOS also allows the researcher to build models that reflect the relationships with the capability to use variables. The path analysis method helps the users to gain an insight to the casual models, so that the variable relationships can be strengthened. With AMOS, one can perform statistical estimation, which allows the users to – create multiple models based on non-numeric data, instead of assigning numeric values to the data. AMOS also allows the users to impute numeric values, so that they can develop a complete numeric database whenever required. Not only this, the users can insert missing data values into the database using AMOS. Hence, we can say that AMOS is very beneficial for data analysis, and one can get answers to all their questions during the process of data analysis. AMOS can run with multiple linear models that exhibit the quality of regression. It is easy to use, as it is integrated with SPSS.

The online Dissertation statistics help forums can provide the users with solution for all kinds of problems related to data analysis and structural equation modelling. With a variety of questions related to statistics, one can interpret the desired results related to data analysis, using these online forums.