Analysis
This is the landing page for everything related to the analysis of your data. The analysis you use should be tailored to the characteristics of your measurements, and should help to answer the research question that you have formulated based on your theory.
Analyses can consist of obtaining summary statistics and visualizations from your data. Typical summary statistics are means, proportions, variances, auto-correlations, cross-lagged correlations, or mean squared successive differences. These quantities are often reported as initial descriptives of the data, but may also form the primary focus of the analyses. Visualizations can be used to represent some of these quantities, and to get an impression of the patterns that are present in the data.
Additionally, you can also chose to fit statistical models to your data. Numerical result from a statistical model can be used for descriptive, predictive, and causal inference purposes, depending on your research question. There are many statistical models for ILD, and these can be categorized in various ways. MATILDA uses several categorizations that should help you navigate the many options that exist, and find the models that seem most useful for your data and research question. A first fundamental categorization is based on the types of change that a model can account for. A closely related categorization that is used in the time series literature is into stationarity versus non-stationarity. A third useful categorization is based on whether you have data from a single person or from multiple persons, and thus concerns N=1 versus N>1 models. Other important categorizations are based on continuous versus discrete time models, depending on whether we want to focus on the process evolving continuously over time, or in a discrete time steps manner; and the outcome type, that is, whether we have a continuous valued variable or categorical variable as the outcome and how many outcome variables we want to focus on.
The numerical quantity that you obtain from your analyses (e.g., a model parameter or a summary statistic) should inform you about your research question. Hence, ideally you know exactly what quantities you will eventually compute even before you start gathering your data. This implies that decisions about what you want to learn from your data and what kind of data you need for this are intrinsically connected to the analysis of the data.
Themes
Summary statistics
Measures that summarize specific features of the data.
Visualizations
Techniques to visualize specific features of the data.
Statistical model overview
Overview of all the statistical models that are covered by MATILDA.
Types of inference
Many statistical models can be used with each of these foci, but this often requires specific additional analyses.
Types of change
Distinguishing between lasting changes, reversible changes, or no changes (i.e., mere fluctuations).
Stationarity versus non-stationarity
Categorization of models that is often made in the time series literature and is therefore useful to be aware of.
N=1 versus N>1 models
Categorization based on whether you want to analyze the data of a single person, or of multiple persons.
Continuous time versus discrete time models
Models based on assuming the process evolves continuously over time, versus models that treat time as happening in discrete steps.
Outcome types
Categorization based on whether the outcome is univariate or multivariate, and a continuous-valued variable (e.g., using a slide-ruler), a discrete-valued variable (e.g., using a Likert scale), or a categorical variable (e.g., using distinct, mutually exclusive answer categories).