While there has been considerable research effort expended on determining correlates of breastfeeding initiation and duration, less systematic testing of theory has been done. For health care professionals to provide interventions to increase breastfeeding incidence and duration, it is essential that substantive theory be developed and tested.

A correlational descriptive design with causal modeling methodology was utilized. The purposes of the research were three-fold: (1) to examine differences between women who breastfed and bottle-fed; (2) to test Ajzen's theory of planned behavior with a group of 138 breast and bottle-feeding mothers; and (3) to test a model based on the theory to increase explanation in behavior in terms of early experiences and duration of breastfeeding (n = 148).

Measures based on the theory, Cuson's Attitudes on Breastfeeding Scale, the Breastfeeding Experience Scale, Hughes Breastfeeding Support Scale, and a demographic questionnaire were used for data collection. Women were contacted in their final weeks of pregnancy, shortly following birth, and for those breastfeeding, four to six weeks postpartum. Data analysis included: use of descriptive, parametric, and nonparametric statistics to examine group differences; psychometric testing; and multiple linear and logistic regression, as well as residual analysis, to test the models.

Results indicated significant differences between the two groups of women on major demographic variables and some model variables. The theory of planned behavior reduced to its predecessor, the theory of reasoned action, and with respecification resulted in a model containing additional variables; previous breastfeeding experience and family income. Fifty-two percent of the variance in prenatal intentions was explained by the first stage variables and twelve percent of the variance in actual behavior was predicted by intentions.

The test of the intention and outcome model was partially supportive of the theoretical hypotheses. Model respecification resulted in a model that explained twenty-three percent of the variance in intentions, ten percent of the variance in breastfeeding experience perceptions, and nine percent of the variance in breastfeeding duration. Theoretical and methodological issues, suggestions for clinical practice, and recommendations for future research are presented.