The findings demonstrated that the air pollution self efficacy measure developed by this study obtained good validity values. In addition we found that the pattern of observed level of self efficacy across the stages of change was consistent with health education theory’s prediction . Self-efficacy was lower among participants who were in pre action stages and was higher among those who were in the action stages. The findings from current study were consistent with previous findings on the topic [16, 40–43] indicating that targeting interventions that focus on self-efficacy as a theoretical framework for a desirable behavior (that was reduced time spent outdoors) might lead to acceptance of a behavior (that was prevention of air pollution hazards).
Few studies have investigated the validity of self-efficacy scales using confirmatory factor analysis. In line with our study, Latimer et al.  reported a unidimensional scale for weight loss among women with a sedentary lifestyle while Barkley and Burnes  reported a three-dimensional tool for condom use self-efficacy. One possible explanation for such differences in construct of these questionnaires is the fact that these tools were used for different types of behaviors. For instance while weight loss is a matter of personal attempt, condom use self-efficacy is a unique behavior that requires interaction between two partners.
As Bandura  advocated a behavior-specific approach to the study of self-efficacy, he argues that a measure of general self-efficacy in overall ability for tapping an individual’s efficacy in managing tasks associated with a specific behavior would be inadequate. Thus, to assess air pollution exposure self-efficacy, an instrument specific to tasks could lead to more concise values of self-efficacy as compared to using a general self-efficacy measure as reported by Schwarzer .
The validity of the scale was strengthened further by estimates of how well the observed indicators (each item on the measure) served as a measurement tool for the construct of self-efficacy related to the prevention of air pollution hazards behaviors. These estimates all were suitable, providing strong evidence that each item reflected self-efficacy. The Cronbach’s alpha coefficient was 0.74 and seems very satisfying for a 4-item scale. It is argued that using large scales are not necessary for predicting a health behavior and rather using rigorous theory-based item wording is more important than the number of statements of an scale . Thus, one might conclude that our short scale could be useful to measure self-efficacy. Self-efficacy scores are very important for setting priorities when developing an specific intervention. One goal of developing a scale is to construct parsimonious measures that can be integrated into a more comprehensive questionnaire . As such we feel our new scale could be integrated into interventions based on many health education and promotion theories and models such as: TransTheoretical Model, Social Cognitive Theory, Theory of Planned Behavior and Health Belief Model; where all have the construct of self-efficacy and are among the most used theories for health behavior change .
In summary, one of the most important millennium development goals is to improve maternal health. Also millennium goal 7 and 8 state that the study of air pollution and its potential public health impact on the general population and highly susceptible groups such as pregnant women should be a priority . In addition it is recommended that one goal of any program is to be measured correctly . Yet, we thought developing a measure of self-efficacy for prevention of air pollution hazards for pregnant women might cover these goals in particular and help to improve women’s health in general.
The current study, however, had some limitations. Almost all participants (96%) were housewives. Perhaps further testing of the measure is needed with employed pregnant women. In addition since our analysis was not based on the maximum required sample size, its replication in a larger sample is warranted to confirm the factor structure of the measure.