---
title: "Factor analysis"
url: "https://helpdesk.telmarhelixa.com/en/article/factor-analysis"
language: "en"
last_updated: "2026-10-27T15:24:44.671Z"
---

# Factor analysis

Reduces large batteries of correlated questions into foundational latent dimensions (FA).

-   Evaluates **factor loadings** (±1 represents a strong association).
-   Factor scores can be saved directly as custom audiences, for profiling in Crosstabs or as clustering inputs.
-   Variable reduction automatically prunes attributes below defined correlation thresholds.

## **Core purpose and mechanism**

-   **Simplification:** instead of analyzing 50 separate attitude questions, group them into 5 to 10 factors (for example "Price Sensitivity", "Innovation Seeker", "Brand Loyalist").
-   **Identifying underlying constructs:** assumes that observed correlations between variables come from shared underlying factors.
-   **Data input:** works best with continuous or interval data, such as Likert-scale responses or volumetric data, where the variables are correlated.

## **Use cases and benefits**

-   **Attitudinal segmentation input (pre-clustering step):** FA is frequently run before Cluster Analysis. It reduces the dimensionality of psychographic / attitudinal data, so clustering uses 5 to 10 composite factor scores instead of 50 raw variables, giving a more robust, less noisy input.
-   **Scale construction and validation:** confirm that survey items designed to measure one construct (for example "Trust in a Brand") really measure a single factor.
-   **Identifying key drivers:** shows which groups of statements or attributes move together, pointing to a shared psychological or behavioral driver.
-   **Data interpretation:** makes large datasets easier to handle and lets researchers name and characterize the latent factors.

## **Interpreting factor analysis**

-   **Factors:** the new, unobserved variables the analysis creates, ordered by how much variance they explain.
-   **Factor loadings:** correlation coefficients between the original variables and the factors.
-   -   **High loadings (close to 1 or -1)** mean a strong relationship. Variables with high loadings on the same factor are conceptually grouped.
    -   **Naming factors:** read the variables that load highly on a factor and give it a descriptive name (for example "Health Consciousness").

## **Relationship to crosstab tools**

Each respondent's **factor scores** can be treated as new variables.

-   Save them as a custom audience in Explore and use them as column or row variables (see Explore - Custom Audiences and Snapshots).
-   This lets researchers profile factors (for example "Which demographics are high on the 'Innovation Seeker' factor?") or feed further analysis such as regression or Cluster Analysis.

## **Settings**

The settings modal (settings button) allows **variable reduction** based on a factor loading threshold. It removes variables with negligible correlation to keep the data clean and preserve useful variables. Highlight colors and factor loading thresholds can be set here to visualize Large, Moderate and Small variances in the Factor Analysis table.