Individual Submission Summary
Share...

Direct link:

Poster #219 - Using Longitudinal Data Analysis to Analyze the Still Face Paradigm

Fri, March 22, 12:45 to 2:00pm, Baltimore Convention Center, Floor: Level 1, Exhibit Hall B

Integrative Statement

Early life experiences, especially parental sensitivity, influence the regulatory and relational capacities of a developing human (e.g., Fleming, O’Day, & Kraemer, 1999; Heim & Nemeroff, 2001; Uvnas-Moberg, 1997). Maternal sensitivity provides external regulation of both psychological and biological development, facilitating stress regulation (Hofer, 1994; Schore, 2001). The Still Face Paradigm (SFP; Tronick, Als, Adamson, Wise & Brazelton, 1978) is used frequently as a lab-controlled situation for studying self-regulation in infants, designed to systematically but temporarily interrupt communication from the mother to the infant using three segments of interaction, a baseline free-play segment, a still-face segment where parental response is inhibited, and a reunion segment where interaction resumes.
In nearly all research currently available using the SFP, averages of infant affect by interval in the three segments (baseline, still-face, reunion) have been compared to draw conclusions about infant development generally. What has not been examined is change of affect within segments. To obtain more detailed information about infant affective response during the SFP, we compared shorter time intervals (15, 30 seconds) with the usual (90 seconds). This allowed for the examination of response patterns over time, a key component of self-regulation (Feldman, 2008), specifically, the comparison of intercepts and slopes of affect within each segment.
Method
During SFP, positive and negative affect were coded each second (Braungart-Rieker et al., 1998) for a diverse sample of 120 eight-month old infants. Baseline and reunion episodes of the SFP were coded for parent sensitivity (sensitivity and intrusiveness, using a 5-pt. Likert scale) in 10-second increments (Braungart-Rieker, Garwood, Powers, & Wang, 2001). Sensitivity and intrusiveness within each episode were averaged to create a composite sensitivity score in which higher scores reflect more sensitive and less intrusive behavior.
Results and Discussion
Several analyses were performed. First, unconditional multilevel models tested intercepts and slopes of affect for each segment, demonstrating the usefulness of the new intervals for detecting changes in affect within and between segments. Compared to 30-second intervals, 15-second intervals were more reliable and better at detecting change within segments. Next, the traditional 90-second intervals were compared to the new intervals and maternal sensitivity was tested as a predictor of affect during each segment. Though affect at previous segments was predictive of affect at later segments, sensitivity had no effect on affect beyond baseline using this approach. Third, conditional multilevel modeling incorporated comparisons of intercepts and slopes for each segment, testing sensitivity during the baseline as a time-invariant predictor. Several effects will be presented but only a sampling is reported here. Sensitivity was predictive of change in affect across baseline (fixed effect=0.34, p=.04) and of differences in slopes of affect between segments (difference between baseline and still-face slopes fixed effect =-0.62, p=0.1). Overall the 15- and 30-second models provided similar results, and both provided more information than the 90-second models. Analyzing the SFP using a smaller interval and longitudinal analysis techniques was useful in providing information regarding how infants regulate within the SFP segments, expanding the type of data SFP protocols can provide.

Authors