A Longitudinal Study
(16 bit high color and 800 x 600 resolution requested)
The present study was designed to predict nutrition behavior half a year after a medical and psychological screening had taken place. Based on 315 participants with complete data at two measurement points in time, the intention to change ones diet and eating behavior were chosen as dependent variables, whereas risk perception, outcome expectancies, and perceived self-efficacy served as predictors within a structural equatition model.
|Eating a healthy diet with low
levels of sodium and fat is a popular medical recommendation. Such a nutrition may help to
prevent cardiovascular diseases and other aliments. However, most individuals do not
adhere to this health behavior, and many have not even developed an explicit intention to
behavior theories describe and explain how and why individuals refrain from risk behavior
and adopt health behavior instead. One such theory is the Health
Action Process Approach (HAPA; Schwarzer, 1992; Schwarzer & Fuchs, 1996), which
suggests that a distinction should be made between preintentional motivation processes and
postintentional volition processes. Thus, there are at least two distinct phases: one
leading to a goal or intention, and one to actual behavior. With both phases, different
social-cognitive processes influence the way people became motivated and turn to action
(cf. Figure 1).
Figure 1. The Health Action Process Approach.
|The Berlin Risk Appraisal and Health Motivation
Study (BRAHMS) was conducted by André Hahn,
Britta Renner und Thomas von Lengerke
under the supervision of Ralf Schwarzer
(Renner & Lengerke, 1996). It comprised two measurement points in time, one in April
one in October. Participants responded to a set of psychometric scales including the three
predictors, and their blood pressure and cholesterol level were measured. These indices
were used for a brief health counseling session. The treatment consisted of personalized
feedback and individual face-to-face counseling. Half a year later behavioral indices were
The longitudinal sample comprised 621 persons. The present analysis is based on 325 participants with complete data, of whom 38% were employed, 16% jobless, 25% students, and 17% retired. The age ranged from 17 to 76 years, with a mean of 44.5 years. Of the participants, 47% were woman.
|For the assesment of risk
perception, two five-item scales were developed. The absolute scale consited of
items such as "How do you estimate the likelihood of experiencing a heart
attack..." (a = .86). Responses were made on seven-point
Positive outcome expectancies were measured with ten items such as "If I stick to a low-sodium and low-fat diet, then my health will benefit." (a = .85). Responses were made on four-point scales.
Perceived self-efficacy was measured by ten items such as "I am confident that I can stick to a healthy diet (low sodium, low fat), even if its taste does not appeal to me in the beginning." (a = .93).
The intention to adopt a preventive nutrition was assessed by two items, namely "I intend to eat only a very low amount of fat (such as animal fat, cheese, butter) over the next months." and "I intend to eat only a minimal amount of salt over the next months."
There were two different approaches to the assessment of nutrition behavior half a year later, a highly specific one referring to eating sausages ("How frequently do you eat fatty sausages with 40-50% fat content, such as salami or liverwurst?" and "How frequently do you eat salty sausages or smoked ham or bacon?" Six frequency categories were provided for responses. The second approach to the assesment of nutrition behavior was a 12-item nutrition scale with items such as "I take care not to eat much fat." and "I only eat a low sodium diet."
|To examine the associations between the variables, a structural equatition approach was chosen. Six latent variables were specified, with two manifest variables (indicators) each: risk perception, outcome expectancy, perceived self-efficacy, intention, sausages consumption and nutrition habits. The first three were specified as predictors of the intention. The intention itself, along with perceived self-efficacy, was specified as a predictor of the two behavior latent variables at the second measurement point in time.|
|The structural model as outlined in Figure 2 fit
the data well (c ² = 68.27; df = 44; RMSEA =
0.042; AGFI = .94). Within the motivation phase, the
intention was predicted substantially by outcome expectancies (b
= .51), perceived self-efficacy (b = .32), and risk perception
(b = .21), as hypotesized. These three predictors accounted for
52% of the intention variance.
Within the voliton phase, different patterns of prediction
appeared for the two endogenous variables that pertain to nutrition behavior. The
intention was very closely associated with subsequent nutrition habits (b = .76) and sausages consumption (b =
.46). Perceived self-efficacy, on the other hand, was not at all related to sausages
consumption (b = .01), and only weakly to nutrition habits b = .19).
Figure 2. Strutural equation model.
|All popular health theories suggest that an
intention to change is the best predictor of subsequent behavior, unless unexpected
barriers make the long-term adoption of the new behavior unlikely. The present study has
confirmed that assumption. Two measures of preventive nutrition were chosen, and both were
well predicted by the intention that the research participants had expressed half a year
A second assumption within the HAPA model was the differential prediction in the motivation and volition phases. The behavioral intention was well predicted by outcome expectancies, perceived self-efficacy, and, to a lesser extent, by risk perception. The superiority of outcome expectancies over perceived self-efficacy replicates earlier research findings (e.g., Schwarzer & Fuchs, 1996). It is possible that, depending on the particular cirumstances, one or the other construct is more essential for developing an intention. Both are seen as necessary, and neither is seen as being sufficient alone.
Individuals need to know the contingencies of behaviors and outcomes, and they need to be confident that they can perform the behavior in question. The relationship between the two constructs here was b = .32, with a hypothesized path leading from outcome expectancies to perceived self-efficacy. However, the causal direction is very speculative. The genesis of such variables is probaly interactive.
In the postintentional, preactional phase (e.g., action planning), the HAPA model suggests a strong influence of perceived self-efficacy. Self efficacious individuals are expected to develop more optimistic success scenarios for behavioral change. In the present study, no measures were included to tap this process. However, a link between perceived self-efficacy and later nutrition habits emerged, pointing to the fact that individuals who are more self-efficacious can better self-regulate their actual nutrition behavior. The failure to replicate this link for sausages consumption, however, makes this an unresolved issue.
One possible interpretation pertains to the nature of the critical behavior and the sample. If, for example, a group of addicted smokers undergoes treatment for cessation, we expect perceived self-efficacy to become the most important determinant in the volition phase, since such persons face barriers and setbacks. In contrast, in the average population, a minor change in ones diet is less dramatic, and coping self-efficacy is not really required in this case. Thus it is not surprising that intention is the only strong predictor of nutrition changes.
|Renner, B., & Lengerke, T. v. (1996): Risiko kennen, Verhalten
ändern? Berlin: Techniker Krankenkasse.
Satow, L., & Schwarzer, R. (1997): Sozial-kognitive Prädiktoren einer gesunden Ernährungsweise: Eine Längsschnittstudie. Zeitschrift für Gesundheitspsychologie,5,243-257.
Schwarzer, R. (1992). Psychologie des Gesundheitsverhaltens. Göttingen: Hogrefe.
Schwarzer, R., & Fuchs, R. (1996). Self-efficacy and health behaviors. In M. Conner & P. Norman (Eds.), Predicting health behaviour. Research and practice with social-cognitive models (pp. 163 - 196). Buckingham: Open University Press.
* * *