Download presentation
Presentation is loading. Please wait.
Published byWidyawati Oesman Modified over 6 years ago
1
Introduction to sequential analysis Brian Baucom, Ph.D.
DEPARTMENT OF PSYCHOLOGY | THE UNIVERITY OF UTAH
2
Operationalizing temporal processes in close relationships
Overview Conceptualizing emotional and behavioral processes in close relationships Operationalizing temporal processes in close relationships Important considerations in lagged models Over-time APIM Repeated-measures APIM Cross-lagged APIM
3
Research foci and clinical orientation
Conduct basic and intervention research on overlapping relational, mental, and physical health outcomes Research driven by couple therapy perspective Small couple therapy clinic Karney & Bradbury, 1995; Robles et al., 2013; Whisman & Uebelacker, 2003
4
polarization model Differences between partners
Regulation Experiences and perceptions of differences Strong emotional reactivity & impaired emotion regulation maintain & exacerbate the mutual trap and vilification Mutual trap Vilification Reactivity Relationship distress CLOSE Lab
5
Types of data that could be used to observe emotional, cognitive, and behavioral processes
Audio and video Couple interactions in lab & at-home Psychotherapy sessions Psychophysiology Couple interactions Daily life Self-report Standard questionnaires Daily diary Ecological momentary assessment
6
Operationalizing temporal processes in close relationships
Processes unfold over time Processes involve intrapersonal and interpersonal components Many processes are contextually dependent
7
Cross-sectional Actor-Partner Interdependence Model
X1 Y1 E1 p p X2 Y2 E2 a Outcome = Y Predictor = X Within-partner association = a Actor effect Between-partner association = p Partner effect
8
APIM over time Lagged associations Link between predictor (measured at earlier time point) and outcome (measured at later time point) Major goal is to predict change in outcome Possible with simultaneous measurement of 2 or more variables from both members of a dyad on 2 or more occasions
9
Stability, frequency, & intensity of measurement
Stability of outcome Measures need to change over from one assessment point to the next Frequency and intensity of measurement needs to match phenomenon Determining “correct” time lag is a major on-going area of research Safest approach is to sample over smallest time interval with the greatest frequency possible
10
Measurement invariance
Measures assumed to assess the same underlying construct in the same way at each time point See Brown, 2006 for methods of assessing measurement invariance Want to demonstrate invariance in factor loadings Ensures equivalent rank ordering over time
11
Ex: husband and wife ambulatory heart rate data
We are interested knowing how spouse’s heart rates are related to one another throughout the day. Spouses wore Actiheart biosensors that continuously recorded heart rate during all waking hours over 7 consecutive days 128 Hz ECG waveform data was down-sampled to 1-minute epochs for analysis Case A, 210 minutes of HR data per spouse (N = 420 data points) Case B, 60 minutes of HR data per spouse (N = 120 data points)
12
HRH, time =t-1 HRH, time = t E1 HRW, time = t-1 HRW, time = t E2
Lagged APIM a HRH, time =t-1 HRH, time = t E1 p p HRW, time = t-1 HRW, time = t E2 a
13
Data organization
14
Model results
15
Vector flow field diagram of results
16
Vector flow field diagram of results
17
Comparing vfds
18
Creating A vfd
19
Creating a vfd, cont.
20
Extending the model: does it matter if spouses are in the same place?
21
Extending the model, 2: vocal entrainment and emotional bond
Clientf0 Therapistf0 Therapistf0 Clientf0 Bond Talk turn = t-1 Talk turn = t
22
Results - interview Bond Clientf0 Therapistf0 Therapistf0 Clientf0
n.s. n.s. Clientf0 Therapistf0 .32*** Therapistf0 Clientf0 .3*** .02** -8.55** Bond
23
Results - intervention
Bond -.05*** n.s. Clientf0 Therapistf0 .34*** Therapistf0 Clientf0 .3*** -.07*** n.s. Bond
24
VFDs Low emotional bond High emotional bond Client f0 Therapist f0
Stronger Client f0 Strength of association We currently only have a sample of 13 individuals for whom we have processed complete data, so these results are very preliminary. But really interesting trends seem to be emerging. First, descriptives show that both patient and therapist tend to express more emotion in their voice during the intervention itself, compared with during the assessment. The bottom table shows results from the regression model. We included patient and therapist f0 at both assessment and intervention in the model to predict bond. Note that we excluded age, gender, and type of treatment from this model simply because we don’t have the sample size to include that many predictors. Again these are very preliminary results. However, the first thing to notice is that therapist f0 at assessment still does not seem to predict bond, and patient f0 at assessment still seems to be an important predictor of bond, and a strong one, even when f0 at intervention is included in the model. Again, higher patient vocally encoded emotional arousal at assessment is associated with lower ratings of bond. During the intervention itself, suddenly therapist f0 matters, and seems to matter a lot. The more the therapist expresses emotion in her voice, the strong bond tends to be. But the most interesting trend that seems to be emerging from these data is that the relationship between patient f0 and bond is flipped during intervention. Suddenly, higher vocal expression of emotion predicts higher bond, as we originally hypothesized. So for the patient, higher vocally encoded emotional arousal at assessment is bad for bond, but during intervention it is good. Weaker Therapist f0
25
VFDs are helpful plots for presenting results of lagged APIMs
summary Lagged models assess associations from previous time point to current time point. Lagged APIM results need to be interpreted with respect to time rather than to the mean. VFDs are helpful plots for presenting results of lagged APIMs Helps present information in a combined fashion
26
resources Additional reading about APIM over time
Kashy, D.A., & Donnellan, M.B. (2012). Conceptual and methodological issues in the analysis of data from dyads and groups. In K. Deaux and M. Snyder (Eds.), The Oxford Handbook of Personality and Social Psychology (pp. 209 – 238). New York: Oxford. Assessing measurement invariance over time Brown, T.A. (2006). Confirmatory factor analysis for applied research. New York: Guilford. Interpreting lagged APIMs Perry, N. S., Baucom, K. J., Bourne, S., Butner, J., Crenshaw, A. O., Hogan, J. N., ... & Baucom, B. R. (2017). Graphic methods for interpreting longitudinal dyadic patterns from repeated-measures actor–partner interdependence models. Journal of Family Psychology, 31, 592. VFD on-line tool
28
Biological mechanisms: Surrogate endpoints Clinical endpoints
Enduring vulnerabilities: Behavioral mechanisms Adaptive processes Marital quality Marital stability Cognitive mechanisms Psychological symptoms Psychopathology Stressful events
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.