Get AI summaries of any video or article — Sign up free
First Basic Model using GSCA in #SmartPLS thumbnail

First Basic Model using GSCA in #SmartPLS

Research With Fawad·
4 min read

Based on Research With Fawad's video on YouTube. If you like this content, support the original creators by watching, liking and subscribing to their content.

TL;DR

GSCA in SmartPLS estimates path relationships between observed variables and components while producing dedicated model fit statistics.

Briefing

Generalized Structured Component Analysis (GSCA) in SmartPLS is presented as a new way to run structural equation models that estimates path relationships between observed variables and components. The core practical takeaway is that GSCA lets researchers keep the same modeling workflow—dragging constructs, connecting paths, and calculating results—while adding a clear set of model fit outputs to judge how well the specification accounts for the observed data.

In the session, a GSCA model is created using an existing project and data set. The model type is switched to GSCA, a first model is saved, and the researcher specifies the structural relationship of interest: the impact of “assurance” on “organizational performance.” Once the path is connected and the model is calculated, the output includes familiar measurement and structural elements. Outer loadings are reported for the measurement side, and the results are described as matching what would be obtained under earlier approaches—highlighting that the technique changes, not the basic interpretation of loadings and relationships. Reliability and validity checks also appear, including composite reliability (not shown in the immediate view but indicated as available), along with discriminant validity presented in graphical form.

The main difference emphasized comes with model fit assessment. GSCA provides multiple fit statistics, including SRMR and fit indices intended to quantify how closely the estimated model reproduces the observed data. The guidance given is straightforward: SRMR values below 0.08 (closer to zero) indicate better fit, while fit indices closer to 1 are better. These fit measures are described as adjusted for free parameters, and local fit indices are also available for later discussion.

To interpret the fit numerically, the session points to a variance-explained framing. Fit values are treated as ranging from 0 to 1 and interpreted as the variance accounted for by the model specification. The example results include a fit around 0.646, described as explaining roughly 64.6% of the variance, and an additional output indicating about 50% of the variance of all variables is explained by the model. Together, these are used to label the model fit as “reasonable.”

Finally, hypothesis testing is carried out through bootstrapping. The session recommends running bootstrapping with 5,000 resamples, using bias-corrected and a one-tailed test setting. The resulting path coefficients show statistical significance, leading to the conclusion that assurance has a significant and positive impact on organizational performance. The walkthrough ends by noting that the same GSCA approach can be extended later into more complex models.

Cornell Notes

GSCA in SmartPLS provides an alternative structural equation modeling approach that estimates relationships between observed variables and components while also producing model fit statistics. In a basic example, assurance is linked to organizational performance, and the workflow mirrors earlier SmartPLS modeling: connect paths, calculate, and review outer loadings plus reliability and validity outputs. The session highlights GSCA’s fit reporting—especially SRMR (values < 0.08 are better) and fit indices closer to 1—along with variance-explained interpretations (e.g., about 64.6% variance accounted for). Hypotheses are tested using bootstrapping with 5,000 resamples, bias-corrected settings, and a one-tailed test, yielding a significant positive path from assurance to organizational performance.

What does GSCA add to a SmartPLS workflow compared with earlier SEM approaches?

GSCA’s standout feature in this walkthrough is model fit assessment. After calculating the model, it provides fit statistics such as SRMR and fit indices (with guidance that SRMR closer to zero and fit indices closer to 1 indicate better fit). It also offers variance-explained style interpretation of fit values, letting users quantify how much of the observed variance the model accounts for.

How is the measurement side checked after running a basic GSCA model?

The session reports outer loadings for the measurement model and then checks reliability and validity outputs. Outer loadings are shown as acceptable (“all good”), reliability/validity results are reviewed (including composite reliability, noted as not visible in that specific view but expected), and discriminant validity is also presented in graphical form.

How should SRMR and fit indices be interpreted in this GSCA example?

SRMR is interpreted using a threshold: values less than 0.08 (closer to zero) indicate good fit. Fit indices are interpreted as better when closer to 1. The session also notes these fit measures are adjusted for free parameters, and local fit indices exist as additional diagnostics.

What variance-explained interpretation is used for GSCA fit values?

Fit values are treated as ranging from 0 to 1 and interpreted as the variance accounted for by the model specification. In the example, a fit value around 0.646 is described as explaining about 64.6% of the variance, and another output indicates roughly 50% of the variance of all variables is explained by the model.

How are hypotheses tested after estimating a GSCA model in SmartPLS?

Hypotheses are tested via bootstrapping. The walkthrough uses 5,000 bootstrap samples, with bias-corrected settings and a one-tailed test. The resulting path coefficients are then checked for significance; the example concludes that assurance has a significant positive impact on organizational performance.

Review Questions

  1. In a GSCA model, which fit statistic is used here to judge absolute fit, and what threshold is given for it?
  2. How does the session interpret fit values numerically in terms of variance explained?
  3. What bootstrapping settings are used for hypothesis testing, and how do they affect the path significance check?

Key Points

  1. 1

    GSCA in SmartPLS estimates path relationships between observed variables and components while producing dedicated model fit statistics.

  2. 2

    A basic GSCA workflow can reuse an existing SmartPLS project: set model type to GSCA, save, connect paths, and calculate.

  3. 3

    Outer loadings and reliability/validity outputs (including discriminant validity) are reviewed after calculation, similar to earlier SEM approaches.

  4. 4

    Model fit is assessed using SRMR (values < 0.08 are treated as good) and fit indices (closer to 1 is treated as better).

  5. 5

    Fit indices are interpreted as variance accounted for by the model specification, enabling percentage-style statements about explained variance.

  6. 6

    Hypothesis testing is performed with bootstrapping (5,000 resamples, bias-corrected, one-tailed), and significant path coefficients support directional claims.

Highlights

GSCA’s practical value in SmartPLS is the built-in fit reporting—especially SRMR and fit indices—used to judge whether the model accounts for observed data.
The example translates fit into variance-explained language, citing about 64.6% variance accounted for and roughly 50% explained across variables.
Bootstrapping with 5,000 bias-corrected resamples and a one-tailed test yields a significant positive path from assurance to organizational performance.

Topics

Mentioned

Get summaries like this for any content

Videos, articles, research papers — all summarized by AI and searchable in one place.

Start building your library