Marina FORTES – BOURBOUSSON
Physical activity is an important contributor in the treatment of obesity. Even if weight loss can be directly explained by energetic costs in physical activity participation, associate psychological benefits are induced as well. Physical activity, sport or exercises further enhance psychological well-being, improve self-esteem and reduce feelings of anxiety and depression. These psychological mechanisms are major in explaining physical activity programm efficacy. However, most researches displaying these conclusions are based on nomothetic approach that constitutes a limit when considering time-evolutionary properties of psychological dimensions (Slifkin and Newell, 1998). In this perspective, the purpose of this study was to model and characterize the psychological processes that underlie the dynamics of psychological dimensions over time in the aim to understand the role of physical activity on obese psychological functioning.
Most researches are based on nomothetic approaches : it signifies they are interested in the average change of psychological determinants or generic functioning of the self-concept. Considering that self-esteem presents long-term fluctuations (Rosenberg, 1986) and appears to be sensitive to daily (bad or good) events (Nezlek & Plesko, 2001), one objection to these classical approaches of psychological constructs (i.e. self, anxiety, depression) is that they only pay attention on changes of magnitude and not on the processes that underlie the emergence over time. According to Slifkin and Newell (1998), variability in living systems cannot be understood without considering the time-evolutionary properties of behavior.
In this perspective, the purpose of this study was to model and characterize the psychological processes that underlie the dynamics of self-esteem, depression, anxiety and physical self over time in the aim to understand the role of physical activity on obese adults psychological functioning.
Method
Participants. Six obese adults (5 woman, 1 man; mean age of 46.8 ±9.3) were recruited from a larger pool of 30 obese adults who participated in a 6-month physical activity. Participants were volunteered for this study. They gave informed written consent to participate. Participants were included if they were: (a) obese (BMI>30 kg/m2), (b) between the ages of 18 and 75 years and (c) not submitted to any insulinotherapy treatment or sulfamides or any other invalidating pathology. All participants received their physician’s medical clearance.
Procedures. From the beginning of the program, each participant completed a booklet in which each page contained six 10-cm visual analog scale measuring depression, anxiety, self-esteem and social, familial, and physical self. The subjects were asked to respond twice a day, between 7:00 and 9:00 (AM and PM), over a three-month period. The six items were presented successively in random order, and participants had to respond on a visual analog scale anchored by “not at all” at the left extremity and “absolutely” at the right. We obtained 168-point time series for each scale and each participant.
Time series analyses. The ARIMA procedures (Box & Jenkins, 1976) were realized for each individual scale and allowed to model construct evolution. These analyses are based on the study of the autocorrelation (ACF) and partial autocorrelation functions (PACF), which inform about the temporal dependence of the series. An ARIMA model is composed of the potential association of three kinds of mathematical processes: auto-regressive (AR), integrated (I) and moving average (MA) processes. The model is labeled as (p, d, q) where p indicates the number of auto-regressive terms, d the number of differentiations, and q the number of moving average terms. The models resulting from analysis are constituted from the combination of these three processes. For example, an ARIMA model (0,1,1) obeys the following equation:
yt = m + y(t-1) – qe(t-1) + et (1)
where m is the mean of the series, q1 the moving average coefficient and et the error at time t.
A representative time series is presented in Figure 1. As can be seen, the series appeared generally non-stationary, with constantly marked evolutions.
Figure 1: A representative time series (subject 2, global self-esteem)
The results showed that autoregressive models without differentiation and with a significant constant provided a satisfactory support accounting for the dynamics of 93.3% of the series (p < .01). Surprisingly, results revealed seasonality components necessary to give an account of 46.7% of the series (p < .01).
As such, the resulting models could be presented as follows:
y t = f1y(t-1) + et .17 < f < .54 (16 series) (2)
yt = μ + fy(t-1)+ Fy(t-14) + et .19 < F < .42 (14 series) (3)
Autoregressive models suggest that past value (past self-evaluation) entirely determines current value. It reveals short-time dependencies and the absence of stable reference like personality trait.
Discussion.
The aim of this paper was to study the dynamics of the self in obese adults. The results of this study are opposed to Fortes et al. (2004) conclusions. These authors have shown that healthy self-concept dynamics was fitted by a differenced first-order moving average model. It suggested a kind of individual disposition related to the stability of self-esteem and its resistance to the influence of daily events. The dynamics of the series seem organized around a locally stable reference value, a kind of transient trait that evolves progressively under the influence of life events.
Surprisingly, the results of this study emphasize the importance of daily events independently to exercise sessions and highlight constant changes of depression, anxiety and self. Obese adults psychological functioning can be characterized by strong and permanent fluctuations. This great variability and short-term historicity (time dependence) show that obese adults have not stable self-evaluations or representations but rather “build” themselves everyday or continuously, they “adapt” themselves to daily perturbations or displaying chronic psychological patterns which were revealed by seasonality.
Besides, physical self that was supposed to be more affected or influenced by physical activity program shares similar dynamics than other dimensions for a same subject. In this way, this study reveals the general extreme vulnerability of obese adults when considering psychological dimensions.
It underlines the limits to realize nomothetic or group studies and can explain the lack of consensus when considering results of researches having measured the effect of different program (diet, exercise…) on psychological changes from few evaluations.
This exploratory research first emphasizes the relevance of an idiographic protocol design to provide a potent means to more fully understand the functioning of psychological constructs such as the self. Second, as advocated by Slifkin & Newell (1998), it emphasizes the need to consider the assessment of both the amount and the dynamic structure of self-evaluative variability.
