STATISTICAL SOFTWARE
LISREL SEM Analysis Service
Model Specification You Can Rerun and Defend
Our LISREL SEM analysis service estimates confirmatory factor, path and full structural models in LISREL, working from your data or covariance matrices and syntax, with the input checked against the intended model and the syntax and output provided where agreed so that each result can be rerun.
No Commitment Required
Strictly Confidential
In-House PhD-Level Statisticians
Fixed Project Fee
Same-Day Response
What Is LISREL Used For in Research?
LISREL is a commercial structural equation modeling program from Scientific Software International, long used for confirmatory factor analysis, path analysis and models that combine latent variables with structural paths. Researchers choose it because its SIMPLIS and LISREL syntax writes a model as explicit statements that can be read, archived and rerun, and because PRELIS prepares the data and the covariance or correlation matrix the model is fitted to. Many published scales and measurement studies were analyzed in it, which makes continuity with earlier work a common reason to stay. OnlineSPSS, a specialist statistical consultancy for academic, clinical and professional research, provides a LISREL SEM analysis service in which an in-house PhD-level statistician estimates the agreed model in LISREL and documents each specification decision.
WHEN TO ENGAGE
Four Situations in Which Researchers Bring Us a LISREL Model
Most LISREL requests start from one of four model states, and a LISREL statistician handles each differently: specified and ready, failing to converge, inherited from someone else or under reviewer challenge.
Model Ready to Estimate The measurement and structural model are specified from theory and the data are collected, and you need the model estimated, its fit judged with care and every parameter reported with its standard error.
Solution That Will Not Behave A matrix is not positive definite, a variance comes out negative or the output describes a model other than the one intended, and the cause is traced in the input before anything is interpreted.
Syntax to Check or Extend Input written by a supervisor or an earlier project is read line by line against the model diagram, then corrected or extended, so the analysis rests on what the file does and not on what it was meant to do.
Reviewer Questions on Fit A reviewer challenges the fit indices, the sample size or a respecification, and the model has to be examined again, compared with defensible alternatives and reported openly.
ANALYSES IN LISREL
Models Estimated in LISREL for Your Research Question
Each model is specified from theory and checked for identification before estimation, and its card links to the method page where the modeling choices are explained.
Structural Equation Modeling (SEM)
A LISREL structural equation modeling project estimates the measurement model and the structural paths together, with fit judged from several indices and indirect effects reported with their standard errors.
Confirmatory Factor Analysis (CFA)
LISREL CFA analysis tests a factor structure fixed in advance, with each item allowed to load only on its intended factor, then reads loadings, residual variances and factor correlations beside the fit.
Path Analysis
LISREL path analysis models directional relations among observed variables, with the model checked for identification and direct and indirect effects interpreted within what the design supports.
Measurement Invariance
Groups are compared in a multigroup LISREL model that holds loadings and then intercepts equal, so latent means are compared only where scalar invariance holds, fully or partially with the freed parameters stated.
Latent Growth Curve Modeling (LGM)
Repeated measures are placed on intercept and slope factors, with the time coding and the means structure set so that the slope has a stated meaning.
Mediation Analysis
An effect transmitted through one or more mediators is estimated inside the model, with total, direct and indirect effects reported and a causal reading offered only where the design allows it.
COMMON LISREL ISSUES
Where LISREL Models Go Wrong Before the Fit Statistics
A model can run, converge and look reasonable while describing something else, so each input and specification check below is made before any estimate is interpreted.
Matrix and Variable Order The covariance matrix, the variable list and the labels must agree position by position. A shifted order still runs, so each label is checked against the data it names.
Matrix Type and Sample Size A correlation matrix supplied where a covariance matrix is expected changes the standard errors and tests, and the sample size must travel with a matrix. Both are confirmed before estimation.
Scale and Defaults Set Deliberately Each latent variable needs a stated scale, and a parameter that the program frees or fixes by default can change the model. The estimated model is compared line by line with the diagram.
Suggested Modifications in the Output LISREL lists the paths and error covariances whose release would lower chi-square most. A suggestion is adopted only when item content or theory explains it, and a model revised this way is tested in a holdout or new sample where one exists.
Ordinal Items Treated as Continuous Few-category, skewed items are analyzed from polychoric correlations with an estimator built for them, with the PRELIS asymptotic matrix computed from the same cases and variable order, or the treatment as continuous is justified.
Standardized or Completely Standardized LISREL prints a standardized solution, which scales only the latent variables, and a completely standardized solution, which also scales the observed ones. Loadings and paths reported as standardized are taken from the completely standardized solution, and the table says which.
DELIVERABLES
What You Receive From a LISREL Analysis
The files follow a documented modeling workflow, agreed in the scope and provided where applicable, so that your team can reopen and rerun every model reported.
Syntax Files The LISREL or SIMPLIS syntax for every model estimated, commented and named so that each model in the tables matches one file.
Output Files The full output of each run, kept with its syntax as the record behind every estimate, standard error and fit statistic in the results.
Matrices and Data The covariance or correlation matrix with means and sample size, or the prepared data file, with a note on variable order and the treatment of missing data.
Tables and Diagrams Publication-ready tables and path diagrams in the format agreed, with standardized and unstandardized estimates, standard errors and the fit statistics.
Statistical Interpretation Each parameter and fit result is explained within what the design supports, and the limitations of the model are stated plainly.
Model Log A record of the models tested, the changes made and the reason for each, with the LISREL version noted so that a coauthor can reproduce the run.
OTHER PLATFORMS
When Another Platform May Be More Appropriate
LISREL suits many measurement and path models, but the software follows the model, and some projects are served better elsewhere.
ILLUSTRATIVE EXAMPLE
How a LISREL Analysis May Look in Practice
An illustration of how a scale validation moves from an input matrix to a model that can be defended and rerun.
ILLUSTRATIVE EXAMPLE · PARENTING STRESS MEASUREMENT RESEARCH · UNITED STATES
A university family studies group wanted to confirm the factor structure of a parenting stress scale in two community samples and then ask whether it measured the same construct for mothers and fathers.
The factor structure was fixed from theory before estimation, and the items, answered on five-point scales, were analyzed from polychoric correlations with an estimator suited to ordinal data, with the asymptotic covariance matrix computed from the same cases. The first run converged with a solution that looked reasonable, but one loading sat far from the pattern the theory predicted. Comparing the label list with the data showed that two item columns had been exchanged when the matrix was built, so two labels named the wrong items and the model was describing variables it was not meant to describe. The matrix and labels were rebuilt together and the model was rerun, with fit judged from several indices instead of a single cutoff. The largest modification index pointed to a cross-loading with no theoretical basis, so it was reported as a question for a new sample and not added to the model.
Outcome: The group received the confirmatory factor model with its loadings, standard errors and fit statistics, the invariance results for mothers and fathers, the syntax and output for every model, the matrices used and a log of each decision.
This example does not describe a real client or project. We never share details of actual client work, in line with the confidentiality we promise every researcher. This illustrative example is included to show the type, scope, and standard of work involved in this kind of analysis.
RESEARCH DISCIPLINES
Research Fields Where LISREL Is Common
LISREL is often met in psychology, education and the social sciences, where scales and path models are central, and the same specification and fit questions arise in health and business research.
HOW WE ASSESS YOUR PROJECT
We Review Your Actual Research
Your assessment is based on a review of your actual research materials and analytical requirements, not a generic project category.
You Submit Your Project Details
Share whatever is available at the current stage, such as the research aim, protocol or proposal, dataset, variable list, manuscript or statistical requirements.
We Review Your Research and Define the Scope
Your statistically trained Research Coordinator reviews the materials to understand the study design, research questions, data structure, analytical requirements and expected deliverables. If clarification is needed, additional documents or information can be requested before the assessment is finalized.
Your Analysis Is Completed and Delivered
Once the scope and fixed project fee are agreed, the in-house statistician completes the work and the Research Coordinator remains the communication point through delivery, with the analysis, deliverables and timeline as agreed.
RESEARCHER VOICE
What Researchers Say
Feedback from a psychometrics researcher on keeping the covariance matrix, variable order and syntax aligned, and on outputs the project team could rerun.
“Keeping the analysis in LISREL mattered because the rest of the project team needed to rerun it later. They dealt with keeping the covariance matrix, variable order, and model syntax perfectly aligned, documented the analytical steps clearly, and gave us outputs that were much easier to audit and report.”
Statistical Support Within Your Research OnlineSPSS provides the agreed methodological, analytical and statistical reporting support, including the LISREL files described in the scope. The researcher or research team retains the scientific question, substantive interpretation, conclusions and overall responsibility for the research.
FREQUENTLY ASKED QUESTIONS
Questions About LISREL Analysis for Research
Short answers to what researchers ask before sharing a LISREL project, from the models we estimate and existing matrices to model fit, platform choice and confidentiality.
Which models do you estimate in LISREL?
Our LISREL analysis covers confirmatory factor models, path analysis, full structural models, multigroup models for measurement invariance and latent growth models. The model is fixed from theory before estimation, and the choice is confirmed during the assessment.
Can you work from my covariance matrix and existing syntax?
We work from the matrix or the raw data and from your syntax files, so the model is checked where it is specified. Existing input is read against the model diagram before it is extended, and a finished model can be checked through Independent Statistical Review, which compares the matrix, the variable order and the specification with what the paper reports. A published covariance matrix can be used for secondary analysis once its sample size, and the standard deviations behind a correlation matrix, are confirmed.
How do you judge and report model fit?
No single index settles it. We report the chi-square with its degrees of freedom, RMSEA with its confidence interval, CFI and SRMR, and read them as a pattern, not against fixed cutoffs. Good fit does not show that a model is true, so competing models are compared and any respecification is labeled exploratory.
Should my model stay in LISREL or move to Mplus or R?
Stay in LISREL when the model is a standard CFA, path or structural model and your team or earlier papers already use it, because continuity with those analyses has value. Move to Mplus or R when many categorical indicators, mixture classes or multilevel latent structures are central to the question, or when coauthors need scripts in an open language.
How do you protect my research data and confidentiality?
Your data files, covariance matrices and syntax are handled confidentially, and project information is not published. An NDA is available where required and can be arranged before sensitive materials are shared. Read more about confidentiality and data security.
What happens when I request an assessment, and how long does it take?
You share your research question, design and relevant materials, such as the dataset or matrix and any syntax, and a statistically trained Research Coordinator reviews the request and responds the same day. The analysis, deliverables, timeline and one fixed project fee are then confirmed before any work begins, so the timeline follows from the agreed scope.
Still have a question? Find more answers in our full FAQ.
NEXT STEP
Discuss Your
LISREL Analysis With Us
Share your research question, your data or covariance matrix, the model you have in mind and any existing syntax. We check the input and confirm the LISREL setup before any work begins.
No Commitment Required
Strictly Confidential
Fixed Project Fee
In-House PhD-Level Statisticians
Same-Day Response
Serving Researchers Since 2014