According to Social Cognitive Theory (Bandura, 1997), nutritional risk woul
d be predicted by perceptions of nutrition efficacy, which in turn would be
predicted from four sources: modeling, verbal support, physiological condi
tions, and nutrition habits. In telephone interviews with 154 rural elderly
adults (44 men, 110 women; mean age = 74.4 years), nutritional risk was me
asured with Nutritional Risk Index (NRI), Nutritional Screening Initiative
(NSI), and seven-high-risk-nutrients consumed less than 50 percent of RDA (
Diet Plus Analysis). Nutrition-efficacy was measured with a new twenty-five
-item scale of Perceived Nutrition Barriers (PNB). Sources of nutrition-eff
icacy included: mealtime modeling-shared meals, household size; verbal supp
ort-people talk to each day, hours talked, number of confidants; physiologi
cal conditions that may affect nutrition-age, body mass, medications, disab
ility, negative affect, and nutrition habits-daily food variety, use of mea
l services. Path analysis was performed with each measure of nutritional ri
sk (NRI, NSI, 7-high-risk-nutrients) as a criterion variable, nutrition-eff
icacy (PNB) as a possible mediating variable, and sources of nutrition-effi
cacy as predictor variables. Social Cognitive variables accounted for 58 pe
rcent of variance in NRI, 49 percent of variance in NSI, and 29 percent of
variance in seven-high-risk-nutrients. Nutritional risk was directly predic
ted by large households, few shared meals, few confidants, high body mass,
many medications, and few daily foods; it was indirectly predicted (via PNB
) by high levels of negative affect. Perceived Nutrition Barriers (PNB) mos
t often mentioned were food cost, eating alone, food tastelessness, transpo
rtation to the store, and chewing difficulty.