Files and representations
QuickPLS accepts tabular CSV and XLSX sources for ordinary observation-level analyses. Methods based on correlations or covariances require the matching matrix representation and metadata. Keep one row per observation and one variable per column unless the selected method says otherwise.
Two-step import
- Choose the file and data representation.
- Review the automatically generated preview and choose Import.
Parser controls, delimiter and decimal options, encodings, and column metadata remain under Advanced import settings. Open them only when the preview or an error requires correction.
Identifiers and analysis roles
Keep case identifiers out of measurement blocks. Retain group variables, sampling weights, mediators, moderators, controls, outcomes and predictors as separate columns. Do not recode an ordinal or binary variable merely to make it look continuous.
- Group: stable authored labels with adequate observations in each group.
- Weight: finite positive values with a documented population meaning.
- Binary outcome: document which value represents the event.
- Moderator: inspect its scale and any centring decision before creating an interaction.
Data-quality checks
- Inspect missing-value codes and do not confuse sentinel values with real observations.
- Confirm numeric parsing, decimal separators and date/text exclusions.
- Look for constant columns, impossible ranges, duplicates and very small groups.
- Record every deletion, transformation, imputation or recoding outside the raw source.
Source protection and provenance
Import into a project copy and keep the source file unchanged. Record the source filename, checksum, row and column counts, role decisions, preprocessing choices and the QuickPLS version used for the analysis.