The experience sampling method (ESM) is an intensive data collection technique, where individuals are prompted to respond to a questionnaire (e.g., containing questions about affect and various contextual variables) multiple times throughout the day and typically for multiple days. By assessing thoughts, emotions, perceptions, and behaviors repeatedly and 'in the moment', we can gain deeper and more valid insights into individual differences and processes of change and examine their relationship to contextual variables (instead of having to rely on cross-sectional data and/or data that is possibly distorted by recall biases). Other terms to describe this and related data collection methods include ecological momentary assessment, ambulatory assessment, event sampling, the structured diary method, intensive longitudinal assessment, and real-time data capture.
Such data collection techniques yield a large number of repeated measurements on multiple variables per individual. The analysis of such data therefore poses particular challenges for researchers. Classical analysis techniques (such as repeated measures ANOVA or MANOVA) are not typically useful in this context, as they cannot easily handle the complexities involved (e.g., missing data, unequally spaced time points, time-varying covariates, residual correlation). Instead, multilevel (mixed-effects) models are typically the method of choice for the analysis for such data.
The purpose of this course is therefore to describe how intensive longitudinal data can be analyzed with appropriate mixed-effects models. After a brief introduction to ESM (and related methods) and the data structures that arise from studies using such data collection techniques, we will examine models that can account for the multilevel/hierarchical structure of the data (i.e., repeated observations nested within individuals). This includes models with random intercepts, models with random intercepts and slopes, and models allowing for remaining correlation in the residuals (serial/autocorrelation). Models with more than two levels (e.g., repeated observations nested within days, which in turn are nested within individuals) will also be covered.
An important issue in any modeling process are the model assumptions. We will therefore examine methods for checking the model assumptions (e.g., examining the distribution of residuals and random effects; checking for conditional independence, linearity, and outliers). Additional time will also be devoted to issues such as model selection, model comparison, and testing (e.g., estimation methods, Wald-type tests and LRTs, information criteria).
The emphasis up to this point in the course will have been on models with quantitative/continuous outcomes. However, other types of outcomes (e.g., dichotomous, multinomial, ordinal) arise in practice. We will therefore take a look at appropriate models for such outcomes (e.g., mixed-effects logistic regression). Next, some psychometric topics (reliability, validity, and factor analysis with multilevel data) will be discussed. The last set of topics to be covered include power analysis, multivariate multilevel models, missing data, allowing variance components to differ across groups, modeling of heteroscedasticity, nonlinear models, and some other advanced topics.
The course consists of a mixture of lectures and computer exercises to cover not only the theoretical background, but also provide practical experience with analyzing real ESM datasets using one or more software packages. Emphasis throughout the course is on the application of the various methods and the interpretation of the results obtained (supplementary references can be provided to those interested in the mathematical/statistical details).
Note: Despite the level of detail, this schedule is tentative. The starting and ending times of the course are definite, but everything in between is subject to change. Also, breaks are not explicitly indicated in the schedule below, but are planned in throughout the days.
|10:20-12:00||Lecture 1: An overview of experience sampling and related methods|
|14:30-16:30||Lecture 2: From regression to the random intercepts model|
|09:00-10:00||Lecture 3: Models with random intercepts and slopes|
|11:00-12:00||Lecture 4: Models with residual serial correlation|
|14:00-15:00||Lecture 5: Models with more than two levels|
|16:00-17:00||Lecture 6: Model assumptions & model selection/comparison/testing|
|09:00-10:00||Lecture 7: Models for other outcome types|
|11:00-12:00||Lecture 8: Psychometrics (reliability, validity, factor analysis)|
|14:00-16:00||Lecture 9: Miscellaneous topics|
In general, all efforts will be made to make the course as self-contained as possible. However, basic knowledge of statistical methods (e.g., regression, analysis of variance, hypothesis testing, factor analysis) will be extremely beneficial to better appreciate the application of these concepts and their extension to the multilevel context.
Also, this course is focused on the statistical methods used for analyzing intensive longitudinal data. There will be a brief introduction to experience sampling methodology at the beginning of the course, but for a more in-depth coverage of the background and the methodological/practical aspects of implementing an ESM study, you could consider taking the general ESM course first, which is offered by members of my department once or twice per year (check here for course offerings).
Finally, the primary software package to be used during the course for the analyses will be R (see below for more details on R and how other software packages will be accommodated). Although everything that you need to know to do the computer exercises will be explained in the course, it would be useful to familiarize yourself with R a little bit ahead of time (here are some notes on preparing to use R). You could also consider following the Introduction to R Course first.
We will conduct a number of practical exercises throughout the course that require a computer. Therefore, please bring a laptop to the course. Moreover, you should have the current version of R installed on the laptop (R is a software environment for statistical computing and graphics). More information about R can be found at the R Project Website. You can download R from the Comprehensive R Archive Network (CRAN) (if you use Windows, choose "Download R for Windows", then "base", and then download the setup program; versions for MacOS and Linux are also available).
One of the big advantages of R is that it is freely available and runs on a large variety of operating systems. Moreover, R has also become the 'lingua franca of statistics' and the software of choice for analyzing data in various disciplines. However, course participants familiar with and with access to other statistical software capable of conducting multilevel analyses are of course welcome to use it. We will in fact accommodate (i.e., provide support and make datasets/scripts available) to users of SPSS and Stata (note that you should have at least version 19 of SPSS). Users of other software (e.g., SAS, Mplus, HLM, MLwiN, SuperMix, BUGS/JAGS/STAN) are welcome to use it, but support will be limited.
Software to open/view PDF files should also be installed on the laptop (e.g., Adobe Reader or see Wikipedia for a comprehensive list of PDF software). In all likelihood, such software is already installed on your laptop anyway.
You can also pair up with another course participant for the computer exercises, so technically not everybody needs to bring a laptop. However, if at all possible, please bring your own laptop so we don't end up with the unfortunate situation that nobody brings one!
The course fee is 600 Euros and includes lunch on all three days, refreshments (e.g., coffee, tea, water) during the breaks, but not dinner or accommodations.
To register for the course, please complete the registration form (to be posted later) and send it by regular mail or e-mail to:
Department of Psychiatry and Neuropsychology
P.O. Box 616 (VIJV1)
6200 MD Maastricht
Tel: +31 (43) 388-3511
E-mail: jolanda.koch (at) maastrichtuniversity.nl
The course will be held at Hotel van der Valk, Maastricht, the Netherlands.
6227 AL Maastricht
Google Maps Link
If you arrive by car, just use the Google Maps Link to obtain directions depending on your starting point. There is free parking at the hotel.
If you arrive by train, you have a couple options. If you leave the train at the main station in Maastricht, then you can take bus number 350 in the direction Aachen (the bus leaves every 15 minutes the last time I checked). Take the bus to the Akersteenweg stop (takes about 10 minutes). From there, it's about a 5 minute walk to the hotel (walk back a bit on the Akersteenweg back from where the bus came, then turn left into the 1 Juliweg, then right into the Nijverheidsweg).
Alternatively, if you leave the train at Randwyck train station, it takes about 10 minutes by foot to get to the hotel (just cross the bridge, follow the Joseph Bechlaan, then turn left into the Demertstraat).
|Min/Max Number of Participants||10/40|
|Certificate for Participation||Yes|
|Number of European Credit Points||1|