Guide to designing a research model

A model canvas that runs in your browser: build constructs and paths, choose the analysis method, simulate data and read the familiar result tables.

Step 1 — Open the canvas and pick a method

Open the Model designer page. You don’t need to sign in to draw a model and save it in your browser. Under 01 Analysis method, choose SmartPLS, SPSS or AMOS — the result tables follow the method you pick.

Want a quick look? Click Load sample model in the Model file section to open a ready-made model.

Step 2 — Build the model

Step 3 — Configure the data

SettingOptions
Sample size (n)Up to 5,000 cases per run.
Item scaleLikert 1–5 (default), Likert 1–7 or continuous (z-score).
Outer loading strengthHigh (0.80–0.92) or Medium (0.72–0.86).

Step 4 — Simulate and read the results

Click Simulate & analyze (sign-in required), rank the strength of the independent constructs, then click Confirm & generate data. The result tables include:

MethodKey statistics
SmartPLSOuter loadings, VIF, Cronbach’s Alpha, rho_A, CR, AVE, HTMT, path coefficients with t and p values, R², f², Q², SRMR, indirect effects.
SPSSCronbach’s Alpha, KMO and Bartlett’s test, EFA, Pearson correlation, regression (VIF, Durbin–Watson).
AMOSCFA, CR, AVE, Fornell–Larcker, HTMT, CMIN/df, GFI, CFI, TLI, RMSEA, PCLOSE, indirect effects (bootstrap).

Click ↻ Resample to draw a different sample, or ⬇ Download DATA to download the full simulated dataset (CSV for SmartPLS, Excel for SPSS; charged per row, see Pricing).

Step 5 — Save and move on to the questionnaire

About simulated data

The data the canvas generates is simulated data based on the model you draw — use it to learn how to read results, test your design and prepare your analysis workflow, not as a substitute for real survey data.

Do I need to install SmartPLS, SPSS or AMOS?

No. The canvas and the result tables run in your browser; the statistics are computed on The Light Data’s servers.

How many items can a construct have?

Each latent construct can have up to 30 observed items; a model can have up to 50 constructs.

Why do I need to sign in to run the analysis?

Sampling and computation run on the server; signing in lets you save your models and keeps the load fair for every user.