About intensive longitudinal data (ILD)

MATILDA focuses on the use of intensive longitudinal data (ILD) to study how processes unfold over time.

On this page, you will find: 1) a definition of ILD; 2) a description of different kinds of ILD; and 3) a discussion on the benefits of ILD.

1 What is ILD?

The term intensive longitudinal data (ILD) has been attributed to Walls & Schafer (2006), who edited a volume entitled Models for Intensive Longitudinal Data that helped establish it in psychology and related disciplines (cf Bolger & Laurenceau, 2013). In their introduction, Walls and Schafer define ILD as arising “[…] in any situation where quantitative or qualitative characteristics of multiple individuals or study units are recorded at more than a handful of time points. The number of occasions may be in the tens, hundreds, or thousands. The frequency of spacing of measurements in time may be regular or irregular, fixed or random, and the variables measured at each occasion may be few or any.” (p.xiii Walls & Schafer, 2006).

Hence, ILD differ from panel data in terms of the number of occasions: While panel data consist of as few as two and typically has (far) less than ten occasions, ILD often consists of 20 or more occasions. Note however that there is some gray area in between these two categories, and that the cut-off is arbitrary; for instance, Bolger & Laurenceau (2013) consider as few as five enough to use the term ILD.

While the description of ILD by Walls & Schafer (2006) refers to having data of multiple cases (e.g., individuals or dyads), many of the analysis techniques that were covered in their book are actually for N=1 data. Such data are also known as time series data. In MATILDA articles the term ILD is used to refer to both N=1 and N>1 data; the reason is that many of the N=1 analysis techniques form the basis for N>1 analysis techniques. When imperative, it will be indicate whether ILD refers to N=1 or N>1 data.

2 Kinds of ILD

Traditionally, ILD were gathered through paper and pencil diary studies, where participants were asked to report their experiences once a day (Walls & Schafer, 2006). This approach was already used in the forties of the last century (Baldwin, 1946; Cattell et al., 1947). Another form of ILD developed somewhat later, uses video recordings of for instance mother-infant or spousal interactions in a lab setting, which are then rated frame-by-frame or second-by-second by experts afterwards resulting in long sequences of verbal and nonverbal behaviors (Gottman & Ringland, 1981). These studies provided a unique perspective on processes unfolding over time at a micro-level timescale. However, gathering and processing these kinds of dense data required serious commitment and considerable resources, and therefore it remained a somewhat niche research form for many decades.

But the popularity of ILD research has increased rapidly in the past two decades (Hamaker & Wichers, 2017), and this can be directly attributed to technological developments such as mobile phones, activity trackers, and other wearable devices that have made it much easier to gather ILD. Nowadays, ILD can be obtained with a wide variety of techniques, including both [self-report] and [passive sensing]. Self-reports can be obtained using daily diaries at the end of the day, but also throughout the day with techniques that have been referred to as [experience sampling], [ecological momentary assessments], and [event contingency measurements]. Passive sensing includes physiological measures such as heart rate, movement, and breathing, but also GPS tracking and behavioral measures such as app use, or automated scoring of emotions based on facial or vocal recordings.

ILD is used as an umbrella term for all these forms of data, the common denominator being the large number of occasions per case. Other terms that are often used in a general way are ambulatory assessment (see for instance the Society for Ambulatory Assessment: https://ambulatory-assessment.org/), and real-time data capture (Stone et al., 2007). MATILDA has chosen the term ILD because it seems more encompassing than ambulatory assessments, which may suggest that lab research is not part of it, and more widely known than real-time data capture.

3 Unique strengths of ILD

A unique strength of ILD is that you can use it to study the unfolding of within-person temporal processes. Unlike other data types, such as cross-sectional data that provide only a single snapshot per individual, or panel data that involve just a few repeated measurements typically over long intervals (e.g., months or years), ILD capture fine-grained high resolution information from individual cases. This makes ILD especially well-suited for studying phenomena that are expected to fluctuate from moment to moment, or day to day, such as affect, stress, and motivation, but also cognitions, physiology, and behaviors, and the interaction between all of these and the environment of the individual.

Due to the high temporal resolution of ILD and by taking the temporal order of observations into account, you can study the dynamic patterns that underlie the fluctuations and changes within a person over time, as well as individual differences in those patterns. This opens the door to a wide range of new research questions; questions that have been associated with a particular theory for a long time, but that could not be directly studied before. Moreover, the opportunities and insights that ILD research offers also fuel the development of new theories, and the adjustment and fine-tuning of existing ones.

With the many new opportunities that are offered by ILD methodology also come major challenges: How do you plan your study in such a way that it allows you to investigate the specific process that you are interested in? It requires decisions on how to measure a process effectively (e.g., when, how often, with what instrument, and in whom), and how to analyze the resulting ILD in such a way that you extract from the data the information that is relevant for your theory and your research questions.

MATILDA is here to help you align your theory, measurement, and analysis in the study of processes.

Acknowledgments

This work was supported by the European Research Council (ERC) Consolidator Grant awarded to E. L. Hamaker (ERC-2019-COG-865468).

Citation

BibTeX citation:
@article{hoekstra2025,
  author = {Hoekstra, Ria H. A. and Schuurman, Noémi K. and Hamaker, Ellen L.},
  title = {About intensive longitudinal data (ILD)},
  journal = {MATILDA},
  number = {2025-05-23},
  date = {2025-05-23},
  url = {https://matilda.fss.uu.nl/intensive-longitudinal-data.html},
  langid = {en}
}
For attribution, please cite this work as:
Hoekstra, R. H. A., Schuurman, N. K., & Hamaker, E. L. (2025). About intensive longitudinal data (ILD). MATILDA, 2025-05-23. https://matilda.fss.uu.nl/intensive-longitudinal-data.html

References

Baldwin, A. L. (1946). The study of individual personality by means of the intraindividual correlation. Journal of Personality, 14, 151–168. https://doi.org/10.1111/j.1467-6494.1946.tb01044.x
Bolger, N., & Laurenceau, J.-P. (2013). Intensive longitudinal methods: An introduction to diary and experience sampling research. Guilford press.
Cattell, R. B., Cattell, A. K. S., & Rhymer, R. D. (1947). P-technique demonstrated in determining psycho-physiological source traits in a normal individual. Psychometrika, 12(4), 267–288. https://doi.org/10.1007/BF02288941
Gottman, J. M., & Ringland, J. T. (1981). The analysis of dominance and bidirectionality in social development. Child Development, 52, 393–412. https://doi.org/10.2307/1129157
Hamaker, E. L., & Wichers, M. (2017). No time like the present: Discovering the hidden dynamics in intensive longitudinal data. Current Directions in Psychological Science, 10–15. https://doi.org/10.1177/0963721416666518
Stone, A. A., Shiffman, S., Atienza, A. A., & Nebeling, L. (2007). The science of real-time data capture: Self-reports in health research. Oxford Academic. https://doi.org/10.1093/oso/9780195178715.001.0001
Walls, T. A., & Schafer, J. L. (2006). Models for intensive longitudinal data. Oxford University Press. https://doi.org/10.1093/acprof:oso/9780195173444.001.0001