Analyzing Intensive Longitudinal Data

The overall goal of the workshop is to enable participants to conduct basic analyses of intensive longitudinal data (potentially including their own data), to interpret the results appropriately, and to know what details of analyses and results should be reported in journal articles. Instruction will also be given in how to conduct power analyses for study planning and for grant applications.

A full-cycle treatment will take workshop participants through five stages of answering example research questions: (1) Design study & collect data, (2) Visualize, (3) Analyze, (4) Write up results, and (5) Power the next study.

The workshop is intended for graduate students, post-doctoral fellows, faculty, and research scientists from the social, behavioral, and health sciences. An emphasis will be placed on interpretation and presentation of intensive longitudinal findings for publication purposes and planning a future intensive longitudinal study for replication and/or grant applications.

We will assume that participants have taken a basic linear regression course and that they have collected or are interested in collecting intensive longitudinal data (i.e., data from daily diary, experience sampling, and ecological momentary assessment methods). Participants who would like a refresher on linear regression prior to attending may wish to watch our (no cost) Linear Regression Playlist on YouTube.

Chapter 1: Overview of Intensive Longitudinal Methods (ILMs)
1.1 Definition and strengths of intensive longitudinal (IL) designs
1.2 Two prototypical research questions
1.3 Structure of IL data and how they differ from multilevel data
1.4 Full cycle approach to answering research questions using ILMs

Chapter 2: Modeling the Time Course (Continuous Outcomes)
2.1 Visualize Intimacy Treatment example using panel plots
2.2 Analyze using multilevel model for longitudinal data
2.3 Write up Intimacy Treatment example for publication
2.4 Power Intimacy Treatment example for a replication

Chapter 3: Modeling Within-Person Processes (Continuous Outcomes)
3.1 Daily conflict and intimacy example
3.2 Visualize using scatterplots and panel plots
3.3 Analyze: distinguish between- and within-person levels in multilevel model
3.4 Write up for publication with example tables and figures
3.5 Power a replication study: revise the number of persons and/or time points
3.6 Handling missing data
3.7 Unequal time intervals and lagged effects

Chapter 4: Essential Guidelines for Modeling IL Data
4.1 Why you should always consider random effects
4.2 Multiple ways to take the influence of time into account
4.3 Within- and between-person levels of analysis should always be distinguished
4.4 Choose the right degrees of freedom for inferential tests
4.5 Choose interpretable zero points for all independent variables

Chapter 5: Psychometrics of IL Data
5.1 Why traditional conception of reliability is not helpful for longitudinal measures
5.2 Definition and conceptualization of reliability of change
5.3 Two approaches to calculating reliability of change
5.4 Daily positive affect in breast cancer patients example

Chapter 6: Modeling the Within-Person Process (Categorical Outcomes)
6.1 Generalized linear mixed modeling with random effects
6.2 Different distributions and different link functions
6.3 Morning anger and evening conflict example
6.4 Visualizing results and interpretation back in the original scale of Y

Chapter 7: Modeling Dyadic Intensive Longitudinal Data
7.1 Dyadic designs and what makes dyadic data unique
7.2 Two types of dyads and “dyadic” analysis approaches to avoid
7.3 Three-level dyadic data structure but two-level dyadic statistical model
7.4 Daily work-to-home stress spillover example
7.5 Extension to Actor-Partner Interdependence Modeling

Chapter 8: Multilevel Within-Person Mediation
8.1 Mediation: Conceptualization, definition, and assumptions
8.2 Single level versus multilevel mediation
8.3 Multilevel (1➞1➞1) within-person mediation
8.4 Daily work-to-home stress spillover example
8.5 Possible extensions of modeling within-person mediation

Chapter 9: Some Future Modeling Extensions and Directions
9.1 Dynamic Structural Equation Modeling (DSEM)
9.2 Modeling random within-person residual variance
9.3 Latent centering of predictors
9.4 Standardizing slopes using individual within-person standard deviations

Live-stream lectures will begin at 10:00 a.m. and finish at approximately 4:30 p.m. Eastern Time (US) each day. There will be a one-hour lunch break as well as 15 minute breaks during the morning and afternoon sessions, the times of which may vary somewhat from day to day to coincide with natural breaks in the material. Local time zones within which the participant is connected must be adjusted to correspond to Eastern Time (US).

Throughout the workshop and interspersed between lecture material, we will illustrate, where possible, how analyses can be conducted on example data sets using all three of the following packages: R, SPSS and Mplus software. While it is helpful to have some familiarity with SPSS, R, and Mplus, it will not be essential.

R can be downloaded for free here. A trial version of SPSS can be downloaded here. This is a 30-day trial version containing all features and is available for both PCs and MACs. A demo version of Mplus can be download here. This is a demo version that is fully-functional, available for both PCs and MACs, but limited in the number of variables used in an analysis.

If you wish to follow along and run the example data files and syntax, these software packages should be installed on your computer prior to the start of the workshop.

Participants receive access to the livestream as well as lifetime access to all course materials, including complete course notes, lecture recordings, software demonstration notes, software demonstration recordings, and data and code for all examples. You can revisit these materials any time you like, without worrying about expiration dates. Pretty awesome, huh?