Uncover latent factors and simplify complex patterns in the data.
Factor analysis is a statistical technique that is used to reduce many variables, or factors, into a smaller number of variables. Often in research data, we have a dependent variable (such as overall satisfaction, or a NPS score), and we want to understand what can be done to improve this metric, or we may want to predict this metric for individual customers. There may be a lot of variables to consider. Many of these will be ‘co-correlated’ – meaning that a person who rates one aspect highly will also likely rate a second or third aspect highly.
Factor analysis can help you reduce the number of factors (reduce the dimensionality) of the data, and which can simplify data interpretation and provide more powerful insights. The ‘factors’ which are uncovered are generally more powerful predictors and hidden or ‘latent’ factors can also be identified. At Lonergan Research we utilise the key insights from factor analysis as an input to carry out further analysis for predictive modelling and multivariate analysis.
HOW FACTOR ANALYSIS WORKS
Our factor analysis methodology:
- Data Collection: We include a range of relevant survey questions that capture a range of information under investigation.
- Factor Analysis: We employ advanced statistical techniques such as principal component analysis and factor rotation to identify the latent factors among observed variables.
- Factor Interpretation: We interpret the newly created factors based on their correlation with the initial variables, and group them into a new set of insightful factors.
- Insights & Further Analysis: The identified factors provide valuable insights into relationships between variables, which we use in further analysis for predictive modelling and multivariate analysis.
- Data visualisation: By reducing the number of variables in the data, while also retaining the most important information, factor analysis allows you to visualise highly complex relationships. This enables you uncover hidden patterns or clusters.
- Variable selection: Identify key variables or indicators that best represent the underlying factors, which can then be used for predictive modelling or analysis.
- Predictive modelling: Factor analysis can help create powerful predictive models, with a limited number of variables, to predict a real world outcome.
Need expert help? Contact Lonergan Research