![Page 1: Conceptual Considerations for Analysis of EMA Data Saul Shiffman, Ph.D. University of Pittsburgh ___ Co-Founder, invivodata, inc. Consult to GlaxoSmithKline](https://reader036.vdocuments.mx/reader036/viewer/2022062718/56649e9e5503460f94ba057a/html5/thumbnails/1.jpg)
Conceptual Considerations for Analysis of EMA Data
Saul Shiffman, Ph.D.University of Pittsburgh
___Co-Founder, invivodata, inc.
Consult to GlaxoSmithKline
![Page 2: Conceptual Considerations for Analysis of EMA Data Saul Shiffman, Ph.D. University of Pittsburgh ___ Co-Founder, invivodata, inc. Consult to GlaxoSmithKline](https://reader036.vdocuments.mx/reader036/viewer/2022062718/56649e9e5503460f94ba057a/html5/thumbnails/2.jpg)
Disclaimer
This talk DOES NOT provide any statistical advice and listeners should consult with their own statistician for advice.
This talk provides commentary. The speaker is not a statistician and is not acting as your statistician. The information in the talk is general information and should not be construed as statistical advice to be applied to any specific factual situation.
The Terms and Conditions specify that there is no guarantee or warranty and that the speaker is not responsible for any loss, injury, claim, liability, or damage ("damages") related to your use of the information in the talk, from errors or omissions in the content of the talk.
Use of information in this talk is governed by our Terms and Conditions; refer to our website for more information.
2
![Page 3: Conceptual Considerations for Analysis of EMA Data Saul Shiffman, Ph.D. University of Pittsburgh ___ Co-Founder, invivodata, inc. Consult to GlaxoSmithKline](https://reader036.vdocuments.mx/reader036/viewer/2022062718/56649e9e5503460f94ba057a/html5/thumbnails/3.jpg)
No Stats
3
![Page 4: Conceptual Considerations for Analysis of EMA Data Saul Shiffman, Ph.D. University of Pittsburgh ___ Co-Founder, invivodata, inc. Consult to GlaxoSmithKline](https://reader036.vdocuments.mx/reader036/viewer/2022062718/56649e9e5503460f94ba057a/html5/thumbnails/4.jpg)
4
Self-Report Methods
· Global self-report “Are you the sort of person who…?” “On average….”
· Time-bound recall “In the past month…”
· Episodic recall “When you first used…”
· Momentary assessment
![Page 5: Conceptual Considerations for Analysis of EMA Data Saul Shiffman, Ph.D. University of Pittsburgh ___ Co-Founder, invivodata, inc. Consult to GlaxoSmithKline](https://reader036.vdocuments.mx/reader036/viewer/2022062718/56649e9e5503460f94ba057a/html5/thumbnails/5.jpg)
5
Ecological Momentary Assessment (EMA)
· Ecological Real-world environments & experience Ecological validity
· Momentary Real-time assessment & focus Avoid recall
· Assessment Self-report, psychophysiology, biological samples Repeated, intensive, longitudinal Allow analysis of process over time
Stone & Shiffman, 1994
![Page 6: Conceptual Considerations for Analysis of EMA Data Saul Shiffman, Ph.D. University of Pittsburgh ___ Co-Founder, invivodata, inc. Consult to GlaxoSmithKline](https://reader036.vdocuments.mx/reader036/viewer/2022062718/56649e9e5503460f94ba057a/html5/thumbnails/6.jpg)
6
Characteristics of Ecological Momentary Assessment
· Assesses subjects in the natural environment· Assesses phenomena as they occur· Considers assessments to be samples· Gathers many repeated observations
![Page 7: Conceptual Considerations for Analysis of EMA Data Saul Shiffman, Ph.D. University of Pittsburgh ___ Co-Founder, invivodata, inc. Consult to GlaxoSmithKline](https://reader036.vdocuments.mx/reader036/viewer/2022062718/56649e9e5503460f94ba057a/html5/thumbnails/7.jpg)
7
Sampling Schemes
· Event-based– Record made when event occurs; subject
typically initiates– Event triggers assessment
· Time-based Regular intervals or milestones
– Daily diary; at every meal– Clock or milestone triggers assessment
Time-based schedules controlled by investigator– Random time sampling or other schemes– Need facilities for scheduling and triggering
assessment
![Page 8: Conceptual Considerations for Analysis of EMA Data Saul Shiffman, Ph.D. University of Pittsburgh ___ Co-Founder, invivodata, inc. Consult to GlaxoSmithKline](https://reader036.vdocuments.mx/reader036/viewer/2022062718/56649e9e5503460f94ba057a/html5/thumbnails/8.jpg)
8
Combined Time & Event SamplingSituational Associations with Smoking
![Page 9: Conceptual Considerations for Analysis of EMA Data Saul Shiffman, Ph.D. University of Pittsburgh ___ Co-Founder, invivodata, inc. Consult to GlaxoSmithKline](https://reader036.vdocuments.mx/reader036/viewer/2022062718/56649e9e5503460f94ba057a/html5/thumbnails/9.jpg)
![Page 10: Conceptual Considerations for Analysis of EMA Data Saul Shiffman, Ph.D. University of Pittsburgh ___ Co-Founder, invivodata, inc. Consult to GlaxoSmithKline](https://reader036.vdocuments.mx/reader036/viewer/2022062718/56649e9e5503460f94ba057a/html5/thumbnails/10.jpg)
Why Bother?
· Ecological validity To study and understand the real world
· Self-report validity To avoid recall error and bias
· Reliability through aggregation To get many observations to achieve reliability, replication
· Temporal ordering and resolution To study how events and processes unfold over time
10
![Page 11: Conceptual Considerations for Analysis of EMA Data Saul Shiffman, Ph.D. University of Pittsburgh ___ Co-Founder, invivodata, inc. Consult to GlaxoSmithKline](https://reader036.vdocuments.mx/reader036/viewer/2022062718/56649e9e5503460f94ba057a/html5/thumbnails/11.jpg)
Craving and Smoking…and Craving…
11
Time
Cra
ving
Time
Cra
ving
![Page 12: Conceptual Considerations for Analysis of EMA Data Saul Shiffman, Ph.D. University of Pittsburgh ___ Co-Founder, invivodata, inc. Consult to GlaxoSmithKline](https://reader036.vdocuments.mx/reader036/viewer/2022062718/56649e9e5503460f94ba057a/html5/thumbnails/12.jpg)
Downward Spiral of Self-Efficacy as Lapses Lead to Relapse
12
Time
Sel
f-E
ffic
acy
![Page 13: Conceptual Considerations for Analysis of EMA Data Saul Shiffman, Ph.D. University of Pittsburgh ___ Co-Founder, invivodata, inc. Consult to GlaxoSmithKline](https://reader036.vdocuments.mx/reader036/viewer/2022062718/56649e9e5503460f94ba057a/html5/thumbnails/13.jpg)
Time is a Crucial Element in EMA Analysis
13
![Page 14: Conceptual Considerations for Analysis of EMA Data Saul Shiffman, Ph.D. University of Pittsburgh ___ Co-Founder, invivodata, inc. Consult to GlaxoSmithKline](https://reader036.vdocuments.mx/reader036/viewer/2022062718/56649e9e5503460f94ba057a/html5/thumbnails/14.jpg)
Collapsing Time:Between-Subject Analyses
14
Craving reportedby abstinent smokerstreated withnicotine patch vsplacebo
Shiffman, S. & Ferguson, S.G. (2008). The effect of nicotine patch on cigarette craving over the course of the day: Results from two randomized clinical trials. Current Medical Research and Opinion, 24, 2795-2804
![Page 15: Conceptual Considerations for Analysis of EMA Data Saul Shiffman, Ph.D. University of Pittsburgh ___ Co-Founder, invivodata, inc. Consult to GlaxoSmithKline](https://reader036.vdocuments.mx/reader036/viewer/2022062718/56649e9e5503460f94ba057a/html5/thumbnails/15.jpg)
Blenderizing Time:Between-Occasion Analyses
15
% of occasions w/ alcohol consumption,when smoking vsnot smoking,among non-dailysmokers identified as“social smokers”
Shiffman, S., Li, X., Dunbar, M., Scholl, S., & Tindle, H. (2012, March). Non-daily smokers = Social smokers? In a symposium on Increasing our understanding of nondaily smoking: Individual patterns, smoking trajectories, and cultural influences (Jasjit Ahluwalia & Saul Shiffman, chairs), presented at the annual meeting of the Society for Research on Nicotine and Tobacco (SRNT), Houston, TX
THURSDAY 1:00 p.m.–2:30 p.m.........Grand Ballroom C, Level 4 INCREASING OUR UNDERSTANDING OF NONDAILY SMOKING
![Page 16: Conceptual Considerations for Analysis of EMA Data Saul Shiffman, Ph.D. University of Pittsburgh ___ Co-Founder, invivodata, inc. Consult to GlaxoSmithKline](https://reader036.vdocuments.mx/reader036/viewer/2022062718/56649e9e5503460f94ba057a/html5/thumbnails/16.jpg)
16
![Page 17: Conceptual Considerations for Analysis of EMA Data Saul Shiffman, Ph.D. University of Pittsburgh ___ Co-Founder, invivodata, inc. Consult to GlaxoSmithKline](https://reader036.vdocuments.mx/reader036/viewer/2022062718/56649e9e5503460f94ba057a/html5/thumbnails/17.jpg)
PromptedAssessment
SubjectEntries
R ***
T L
PrecedingDay
LapseDay
SucceedingDay
R - Random Prompt T - TemptationL - Lapse
Time as Sequence
within subject
![Page 18: Conceptual Considerations for Analysis of EMA Data Saul Shiffman, Ph.D. University of Pittsburgh ___ Co-Founder, invivodata, inc. Consult to GlaxoSmithKline](https://reader036.vdocuments.mx/reader036/viewer/2022062718/56649e9e5503460f94ba057a/html5/thumbnails/18.jpg)
Negative Affect in Background, Temptations & First Lapses
Series10
25
50
75
100
125
150
Random
Tempts
Lapses
Neg
ativ
e A
ffec
t (T
sco
re)
Shiffman, S., Paty, J.A., Gnys, M., Kassel, J.D., & Hickcox, M. (1996). First lapses to smoking: Within-subjects analyses of real-time reports. Journal of Consulting and Clinical Psychology, 64, 366-379
![Page 19: Conceptual Considerations for Analysis of EMA Data Saul Shiffman, Ph.D. University of Pittsburgh ___ Co-Founder, invivodata, inc. Consult to GlaxoSmithKline](https://reader036.vdocuments.mx/reader036/viewer/2022062718/56649e9e5503460f94ba057a/html5/thumbnails/19.jpg)
Pre-Post Event
19
Self-Efficacy,before and after a Temptation vsa Lapse episode
Shiffman, S., Hickcox, M., Paty, J.A., Gnys, M., Kassel, J.D., & Richards, T. (1997). The Abstinence Violation Effect following smoking lapses and temptations. Cognitive Therapy and Research, 21 (5), 497-523
![Page 20: Conceptual Considerations for Analysis of EMA Data Saul Shiffman, Ph.D. University of Pittsburgh ___ Co-Founder, invivodata, inc. Consult to GlaxoSmithKline](https://reader036.vdocuments.mx/reader036/viewer/2022062718/56649e9e5503460f94ba057a/html5/thumbnails/20.jpg)
Event-Anchored Calendar
Time
20
Craving intensity amongabstinent smokers,temptation episodes vsrandom moments, over days since quitting
Shiffman, S., Engberg, J., Paty, J.A., Perz, W., Gnys, M., Kassel, J.D., & Hickcox, M. (1997). A day at a time: Predicting smoking lapse from daily urge. Journal of Abnormal Psychology, 106, 104-116
![Page 21: Conceptual Considerations for Analysis of EMA Data Saul Shiffman, Ph.D. University of Pittsburgh ___ Co-Founder, invivodata, inc. Consult to GlaxoSmithKline](https://reader036.vdocuments.mx/reader036/viewer/2022062718/56649e9e5503460f94ba057a/html5/thumbnails/21.jpg)
Event-Anchored ReverseCalendar & Clock Time
21
Negative affectamong abstinentsmokers, in the days and hours preceding a first lapse, by lapse trigger
Shiffman, S. & Waters, A. J. (2004). Negative affect and smoking lapses: A prospective analysis. Journal of Consulting and Clinical Psychology, 72 (2), 192-201
![Page 22: Conceptual Considerations for Analysis of EMA Data Saul Shiffman, Ph.D. University of Pittsburgh ___ Co-Founder, invivodata, inc. Consult to GlaxoSmithKline](https://reader036.vdocuments.mx/reader036/viewer/2022062718/56649e9e5503460f94ba057a/html5/thumbnails/22.jpg)
![Page 23: Conceptual Considerations for Analysis of EMA Data Saul Shiffman, Ph.D. University of Pittsburgh ___ Co-Founder, invivodata, inc. Consult to GlaxoSmithKline](https://reader036.vdocuments.mx/reader036/viewer/2022062718/56649e9e5503460f94ba057a/html5/thumbnails/23.jpg)
Time as Risk
23
Time to relapse, after a first lapse, by pleasantness ofsmoking in the lapse
Shiffman, S., Hickcox, M., Paty, J.A., Gnys, M., Kassel, J.D., & Richards, T. (1996). Progression from a smoking lapse to relapse: Prediction from abstinence violation effects and nicotine dependence. Journal of Consulting and Clinical Psychology, 64, 993-1002
![Page 24: Conceptual Considerations for Analysis of EMA Data Saul Shiffman, Ph.D. University of Pittsburgh ___ Co-Founder, invivodata, inc. Consult to GlaxoSmithKline](https://reader036.vdocuments.mx/reader036/viewer/2022062718/56649e9e5503460f94ba057a/html5/thumbnails/24.jpg)
Repeated Events over Time
24
Accelerating time-to-re - lapse times oversuccessive lapses,initially slowed bynicotine patch treatment
Kirchner, T.R., Shiffman, S., Wileyto, P. (2012). Relapse dynamics during smoking cessation: Recurrent abstinence violation effects and lapse-relapse progression. Journal of Abnormal Psychology, 121, 187-197
![Page 25: Conceptual Considerations for Analysis of EMA Data Saul Shiffman, Ph.D. University of Pittsburgh ___ Co-Founder, invivodata, inc. Consult to GlaxoSmithKline](https://reader036.vdocuments.mx/reader036/viewer/2022062718/56649e9e5503460f94ba057a/html5/thumbnails/25.jpg)
Even More Ways to Think About Time in EMA Data
· Reciprocal effects e.g., smoking reduces self-efficacy, which increases
smoking, which reduces self-efficacy, which …..
· Cumulative effects e.g., cumulative effort of coping eventually exhausts
quitters, leading to relapse
25
![Page 26: Conceptual Considerations for Analysis of EMA Data Saul Shiffman, Ph.D. University of Pittsburgh ___ Co-Founder, invivodata, inc. Consult to GlaxoSmithKline](https://reader036.vdocuments.mx/reader036/viewer/2022062718/56649e9e5503460f94ba057a/html5/thumbnails/26.jpg)
Data Analysis
· Effort: 50% thinking about theory and question 30% organizing data to address question 20% statistical analysis (now easier)
· Design envy: Experiments: structure dictates analyses EMA: Not much structure… Question dictates
analysis
26
![Page 27: Conceptual Considerations for Analysis of EMA Data Saul Shiffman, Ph.D. University of Pittsburgh ___ Co-Founder, invivodata, inc. Consult to GlaxoSmithKline](https://reader036.vdocuments.mx/reader036/viewer/2022062718/56649e9e5503460f94ba057a/html5/thumbnails/27.jpg)
5 Subjects’ EMA Data
27
![Page 28: Conceptual Considerations for Analysis of EMA Data Saul Shiffman, Ph.D. University of Pittsburgh ___ Co-Founder, invivodata, inc. Consult to GlaxoSmithKline](https://reader036.vdocuments.mx/reader036/viewer/2022062718/56649e9e5503460f94ba057a/html5/thumbnails/28.jpg)
5 Subjects’ EMA Data
28
![Page 29: Conceptual Considerations for Analysis of EMA Data Saul Shiffman, Ph.D. University of Pittsburgh ___ Co-Founder, invivodata, inc. Consult to GlaxoSmithKline](https://reader036.vdocuments.mx/reader036/viewer/2022062718/56649e9e5503460f94ba057a/html5/thumbnails/29.jpg)
304 Subjects’ EMA Data
29N=304 subjects, 191,841 observations
![Page 30: Conceptual Considerations for Analysis of EMA Data Saul Shiffman, Ph.D. University of Pittsburgh ___ Co-Founder, invivodata, inc. Consult to GlaxoSmithKline](https://reader036.vdocuments.mx/reader036/viewer/2022062718/56649e9e5503460f94ba057a/html5/thumbnails/30.jpg)
Design Envy
· In traditional design, design dictates analysis· 1 or n observations / person· Confounds are limited by design
· EMA: We have to work harder to select, arrange, structure data to fit question & analysis
30
Active Placebo
Men
Women
![Page 31: Conceptual Considerations for Analysis of EMA Data Saul Shiffman, Ph.D. University of Pittsburgh ___ Co-Founder, invivodata, inc. Consult to GlaxoSmithKline](https://reader036.vdocuments.mx/reader036/viewer/2022062718/56649e9e5503460f94ba057a/html5/thumbnails/31.jpg)
Summary
· EMA data unstructured+ Can address many different questions
- Require hard thinking & effort to shape for analysis
· Find structure and statistics to match question(not vice versa)
· Consider treatment of time in analysis
31