Process-Based Therapy & Idionomic Analysis
Personalized psychology for lasting change.
Idionomic Analysis / Process-Based Therapy (PBT) represents the forefront of personalized clinical psychology. This approach combines deep individual insight with the power of evidence-based strategies. Rather than relying on diagnostic labels alone, idionomic analysis maps the unique patterns of a person's thoughts, emotions, behaviors, and biology — capturing how they dynamically interact over time.
Paired with PBT, the framework pinpoints core psychological processes: cognitive flexibility, emotional regulation, attention, motivation. The result is targeted interventions tailored to the person, not the disorder. Whether you're a clinician, researcher, or client, this precision-based model supports more effective, meaningful mental health care.
Jump to
The key is to start with the individual (idiographic) and then only move to group-level (nomothetic) generalisations when appropriate. Idio-graphic + nom-ethetic = Idionomic.
- Collect sufficient data at the lowest meaningful level.
- This is Level 1 — individual participants, couples, families, or countries.
- May involve intensive longitudinal data to capture meaningful within-unit patterns.
- Build a model for each Level 1 unit.
- For example, use a simple ARIMA model for each individual to estimate the link between a process (X) and an outcome (Y).
- Evaluate whether aggregation is appropriate.
- Use techniques like meta-analysis to assess the pooled effect size and the heterogeneity (e.g., I²).
- Aggregate if appropriate. If the pooled effect accurately represents most individual data.
- Explore functional subgroups if heterogeneity is high. Look for functionally homogeneous subgroups.
The idionomics R package brings idionomic analysis into one place. If you have data where the same people are measured many times (a daily diary, an experience-sampling study, weekly therapy ratings, repeated couple interactions), this package lets you analyse them person-by-person first, and then ask whether it's safe to combine those individual stories into a group average. The traditional approach does the opposite: it averages first and assumes everyone is alike. That assumption hides a lot.
The package is now on CRAN, so installing it takes one line:
Once it's loaded, the main functions you'll reach for are:
i_screener()— flags participants whose responses look like careless or random clicking, so you can decide who to keep before modelling.pmstandardize()— rescales each variable within each person, so a "high" score for one person isn't compared to a "high" score for another. This is the standard first step before any within-person analysis.i_detrender()— removes long-term linear trends from each person's time series, so you can study day-to-day fluctuations cleanly without slow drift confounding the results.iarimax()— fits a per-person time-series model linking a predictor to an outcome, then meta-analyses across people. This is the workhorse: out comes both each person's individual effect and the group summary, with a measure of how variable the effect is.i_pval()— computes individual p-values, so you can see which specific participants showed a reliable effect.sden_test()— tests for the "equisyncratic null" pattern, where the group mean is near zero but only because half the people show a positive effect and the other half show a negative one. This is the test that catches the cases where averaging genuinely hides a story.looping_machine()— identifies feedback loops in person-specific networks (e.g., A drives B drives C drives A), which is useful for case formulation.
A worked vignette walks through the full pipeline with example data: see the idionomics Pipeline vignette on the CRAN page below.
Authors: Cristóbal Hernández (creator and maintainer), Joseph Ciarrochi, Steven Hayes, Baljinder Sahdra. Methods are described in Hernández et al. (2024), Ciarrochi et al. (2024), and Sahdra et al. (2024) — all of which are downloadable in the article library below.
Type any word (author, title, year, topic) to filter. Click a tag to narrow by topic. Click a title to read the PDF in a new tab, or click Download PDF to save it to your computer.