EmaCalc 1.1.1

Package EmaCalc implements probabilistic Bayesian analysis of Ecological Momentary Assessment (EMA) data. The EMA methodology can be used to evaluate the effect of any kind of psycho-socio-medical intervention, for example, the subjective performance of hearing aids, in the everyday life of the user or client,

EMA Experiments

In an EMA study, each participant is requested to respond to a questionnaire during normal everyday life, typically several times per day. Some questions may address the current real-life Situation, i.e., the physical environment and the user's activity and intentions. The participant may also be asked to rate any perceptual Attribute of interest in the study, for example, the ease of speech understanding in the current situation.

Thus, EMA data usually include both nominal and ordinal results. Typically, many records are collected from each participant, but the number of records may vary a lot among respondents.

This causes some interesting challenges for the statistical analysis (Oleson et al., 2021). The present analysis model estimates Attribute values numerically on an objective interval scale, given the Situation, although the raw input data are subjective and indicate only ordinal judgments for each Attribute, and nominal categories for the Situations.

This package does not include functions to handle the data collection; it can only use existing files with data recorded earlier. The package can analyze data from simple or rather complex experimental designs, including the following features:

1. The complete EMA study may include one or more test Phases, for example, before and after some kind of intervention.

2. Each EMA record may characterize the current situation in one or more pre-defined Situation Dimensions. For example, one dimension may be specified by the Common Sound Scenarios (CoSS) (Wolters et al., 2016), which is a list of broad categories of situational intentions and tasks. Other dimensions may specify the Importance of the situation, and/or the Hearing-Aid Program currently used.

3. Each EMA record may also include discrete ratings for one or more perceptual Attributes. For example, one Attribute may be Speech Understanding, with ordinal grades Bad, Fair, Good, Perfect. Another attribute may be Comfort, with grades Bad, Good.

4. For each Situation Dimension, a list of allowed Situation Categories must be pre-defined. An assessment event is defined by a combination of exactly one selected Category from each Dimension.

5. For each perceptual Attribute, a list of discrete ordinal Attribute Grades must be pre-defined. Ordinal scales may be unique for each attribute, or shared by more than one attribute.

6. An EMA study may involve one or more distinct Populations, from which separate groups of participants have been recruited.

7. Populations are distinguished by a combination of categories from one or more Group Dimensions. For example, one dimension may be Age, with categories Young, Middle, or Old. Another dimension may be, e.g.,Gender, with categories Female, Male, or Unknown.

8. The analysis model does not require anything about the number of participants from each population, or the number of assessments by each participant. Of course, the reliability is improved if there are many participants from each population, each reporting a large number of EMA records.

EMA Data Analysis

The analysis uses the recorded data to learn a probabilistic model, representing the statistically most relevant aspects of the data. The analysis includes a regression model to show how the Attribute values vary across Situations.

1. The analysis results will show predictive Situation Probabilities for each population, credible differences between situation probabilities within each population, and credible differences between populations.

2. The analysis results will also show perceptual Attribute Values for each population, credible differences between Attribute Values in separate situations, and credible Attribute Differences between populations.

3. Differences between populations can be shown as main effects in only one or a few of the pre-defined Group Dimensions, or as complete interaction effects of categories in all Group Dimensions.

The Bayesian analysis automatically estimates the statistical credibility of all analysis results, given the amount of collected data. The Bayesian model is hierarchical. The package can estimate results for

1. an unseen random individual in the population from which the participants were recruited,
2. the population mean,
3. each individual participant.

The first of these probability measures may be most important in a study designed, e.g., by a hearing-aid manufacturer to predict the future marketing success of some new hearing aid feature. The second probability measure is most closely related to the statistical significance as estimated by conventional hypothesis tests. The third measure might be most important in a clinical study to quantify the benefit of an intervention for individual clients.

The package can also quantify individual rating differences by the Non-overlap of All Pairs (NAP) effect measure (Parker & Vannest, 2009), including approximate confidence intervals calculated by the "MW-N" method recommended by Feng et al. (2017).

Package Documentation

General information and version history is given in the package doc-string that may be accessed by commands import EmaCalc, help(EmaCalc) in an interactive Python environment.

Input EMA data can be accessed from files in several of the formats that package Pandas can handle, e.g., .csv, .xlsx. Specific information about the organization of input data files is presented in the doc-string of module ema_data, accessible by commands import EmaCalc.ema_data, help(EmaCalc.ema_data).

After running an analysis, the logging output briefly explains the analysis results presented in figures and tables. Figures can be saved in any file format allowed by Matplotlib. Result tables can be saved in many of the file formats that Pandas can handle, e.g., .csv, .txt, .tex, as defined in module ema_file. Thus, the results can easily be imported to a word-processing document or to other statistical packages.

Usage

1. Install the most recent package version: python3 -m pip install --upgrade EmaCalc

2. For an introduction to the analysis results and the input data format, you may want to study and run the included simulation script: python3 run_sim.py.

3. Copy the template script run_ema.py, rename it, and edit the copy as suggested in the template, to specify

   • your experimental layout,
   • the top input data directory,
   • a directory where all output result files will be stored.

4. Run your edited script: python3 run_my_ema.py

5. In the planning phase, complete analysis results may also be calculated for synthetic data generated from simulated experiments. Simulated experiments allow the same design variants as real experiments. Copy the template script run_sim.py, rename it, edit the copy, and run your own EMA simulation.

Requirements

This package requires Python 3.9 or newer, with recent versions of Numpy, Scipy, and Matplotlib, as well as a support package samppy. Package Pandas is used to handle input data and result tables and requires openpyxl to access Excel (.xlsx) files. The pip installer will check and install these packages if needed.

Pandas can also access input files in other formats, but may then need other support packages that must be installed manually.

New in version 1.1

Flexible user control of result display styles.

Population differences can be shown in one and/or several of the predefined Group Dimensions, as selected by the user.

References

A. Leijon, P. von Gablenz, I. Holube, J. Taghia, and K. Smeds (2023). Bayesian analysis of ecological momentary assessment (EMA) data collected in adults before and after hearing rehabilitation. Frontiers in Digital Health, 5(1100705). download

A. Leijon et al. (2023). Bayesian Analysis of Ecological Momentary Assessment (EMA) Data. Documentation: Theory, Code usage, Validation, and Computational Details. Contact the author for a copy.

D. Feng, G. Cortese, and R. Baumgartner (2017). A comparison of confidence/credible interval methods for the area under the ROC curve for continuous diagnostic tests with small sample size. Statistical Methods in Medical Research, 26(6):2603–2621. download

J. J. Oleson, M. A. Jones, E. J. Jorgensen, and Y.-H. Wu (2021). Statistical considerations for analyzing ecological momentary assessment data. J Sp Lang Hear Res, ePub:1–17. download

R. I. Parker and K. Vannest (2009). An improved effect size for single-case research: Nonoverlap of all pairs. Behavior Therapy, 40(4):357–367. download

K. Smeds, F. Wolters, J. Larsson, P. Herrlin, and M. Dahlquist (2018). Ecological momentary assessments for evaluation of hearing-aid preference. J Acoust Soc Amer 143(3):1742–1742. download

F. Wolters, K. Smeds, E. Schmidt, and C. Norup (2016). Common sound scenarios: A context-driven categorization of everyday sound environments for application in hearing-device research. J Amer Acad Audiol, 27(7):527–540. download