Unintended pregnancy in young women is one of our country's major public health problems. The purpose of this study was to test the causal relationship between elements of an expanded Health Belief Model and contraceptive use. Desire for control and perceived control constructs as well as three modifying variables age, time sexually active and frequency of intercourse were incorporated into the design. Eight hypotheses were proposed to describe the relationship of the elements of the model with birth control use (BCU).

A questionnaire was mailed to a random sample of 618 single female freshman and sophomore students in a large midwestern university. The questionnaire consists of seven HBM and control scales (seriousness, susceptibility, benefits, barriers, normative belief, perceived control, and desire for control) and a section on relationship and demographic information. A 80% return rate was achieved using the Modified Dillman Protocol. Construct validity was supported by factor analysis. Cronbach alpha reliabilities for the scales ranged from.61 to.95. Hypotheses testing was done using structural equations modeling via the LISREL V computer program. The LISREL V program allowed for the control of measurement error based on reliability estimates. A qualitative index of BCU based on consistency and effectiveness of use was employed as the primary dependent variable.

Of the HBM constructs, the barrier variable alone was significantly related to BCU. Perceived control, time sexually active, and frequency of intercourse also had significant direct paths to BCU. The model predicted 52% of the variance in BCU. Further, eight non-causal correlations were identified in the total model. Deletion of the modifying and control variables significantly decreased the predictive power of the model.

This study would suggest that the HBM constructs, when combined with the control variables and the modifying variables in an integrated model, were less predictive than were perceived control, time sexually active and frequency of intercourse in predicting contraceptive use. Further, Structural equations modeling was felt to be a promising technique for identifying the complex relationships between variables.