SmartPLS and AMOS do different jobs: SmartPLS estimates PLS-SEM, judged by loadings, AVE, HTMT and R² (Hair et al., 2019), while AMOS estimates covariance-based SEM, judged by model fit such as CFI ≥ .95, RMSEA ≤ .06 and SRMR ≤ .08 (Hu & Bentler, 1999). So the real choice is between methods, not programs. Both are commercial; free tools such as lavaan for CB-SEM (Rosseel, 2012) and seminr for PLS-SEM, both in R, produce the same estimates, and browser tools let you run either without installing anything.
Side-by-side comparison
| SmartPLS | AMOS | |
|---|---|---|
| Method | PLS-SEM (also PLSc, and some CB-SEM in recent versions) | CB-SEM (maximum likelihood and others) |
| Interface | Drag-and-drop path diagram | Drag-and-drop path diagram |
| Main output | Loadings, reliability, AVE, HTMT, bootstrapped paths, R², f², PLSpredict | Estimates, standard errors, χ², fit indices, modification indices |
| Formative constructs | Straightforward | Need identification constraints |
| Platform | Windows and macOS | Windows only, sold with IBM SPSS |
| Licence | Commercial; check the vendor for current student terms | Commercial; often provided through university site licences |
Which one do you need?
Decide on the method first. If you are testing an established theory and your examiner expects model fit, you need CB-SEM, so AMOS or a free equivalent. If you aim to explain and predict, have formative constructs or a complex model, PLS-SEM is the better fit, so SmartPLS or a free equivalent (Hair et al., 2019). Then choose software you can access for the full duration of your project, including after your university licence expires; journals often ask for re-analysis months later.
Free alternatives
- lavaan (R): the standard free package for CFA and CB-SEM, with robust estimators, multigroup models and all common fit indices (Rosseel, 2012). Models are written in a short text syntax.
- seminr (R): PLS-SEM with bootstrapping, HTMT and PLSpredict, following the reporting logic of Hair et al.
- JASP and jamovi: free point-and-click statistics programs whose SEM modules run lavaan in the background.
- Browser tools: no installation; useful on a laptop where you cannot install software, or on a Chromebook or tablet.
Worked example: same model, different tools
A student has a three-construct model (Service Quality → Satisfaction → Loyalty) with 280 responses. She runs CFA in AMOS at the university lab and gets χ²(51) = 118.6, CFI = .962, RMSEA = .069, SRMR = .041. At home she runs the same model in lavaan with maximum likelihood. The estimates and fit indices match to the second or third decimal, because both programs fit the same likelihood function. Against Hu and Bentler (1999), CFI and SRMR meet the guideline; RMSEA is slightly above .06, which she reports honestly. In APA style she writes CFI = .96 (two decimals, no leading zero).
Her classmate runs a PLS-SEM version in SmartPLS and in seminr. Path coefficients agree to three decimals; bootstrap t values differ slightly between runs because bootstrapping draws random subsamples. Reporting the number of subsamples (10,000; Hair et al., 2022) and, where possible, the random seed makes this transparent.
Switching tools half-way
If you start in one program and finish in another, re-run the whole model in the final tool rather than mixing output. Check that the estimator, missing-data handling and bootstrap settings match, and report only the final tool's results. Small differences in the third decimal are normal; differences in the first decimal point to a different model or data set.
What examiners care about
No examiner should mark you down for using free software. They will mark you down for an unjustified method, missing statistics or thresholds without sources. Name the software and version in your method chapter and cite it (R packages have formal citations; lavaan's is Rosseel, 2012).
Run it free
Run PLS-SEM in the browser with the free PLS-SEM tool, or CFA and CB-SEM with fit indices in the CB-SEM tool (Vietnamese interface). Both are checked against R (seminr and lavaan).
FAQ
Is SmartPLS better than AMOS?
Neither is better; they implement different methods. Choose the method that fits your research objective.
Can AMOS do PLS-SEM?
No. AMOS is a covariance-based SEM program.
Will lavaan give the same results as AMOS?
With the same estimator and model, yes, within rounding. Differences usually come from default settings, such as how missing data are handled.
Do I need to learn R to use lavaan?
A little. Alternatively, JASP and jamovi run lavaan through menus.
References
- Hair, J. F., Hult, G. T. M., Ringle, C. M., & Sarstedt, M. (2022). A primer on partial least squares structural equation modeling (PLS-SEM) (3rd ed.). Sage.
- Hair, J. F., Risher, J. J., Sarstedt, M., & Ringle, C. M. (2019). When to use and how to report the results of PLS-SEM. European Business Review, 31(1), 2–24. https://doi.org/10.1108/EBR-11-2018-0203
- Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6(1), 1–55. https://doi.org/10.1080/10705519909540118
- Rosseel, Y. (2012). lavaan: An R package for structural equation modeling. Journal of Statistical Software, 48(2), 1–36. https://doi.org/10.18637/jss.v048.i02