Application of Classification and Regression Trees in Iranian Women’s Childbearing Desire

Socio-economic factors, in addition to value and attitude changes, have had a major impact on fertility behavior of women toward childbearing. The main purpose of this study is to investigate childbearing desire of 6183 women in provinces with total fertility rate (TFR) lower (TFR ≤ 2) and upper (TFR > 2) than replacement level by classification and regression trees (CART) algorithm, as one of the most applicable classification trees. In this study women’s job, place of residence, age, opinion, marriage age, educational level, and children ever born were investigated as predictors. The accuracy of fitted classification trees for CD in provinces with TFR ≤ 2 and TFR > 2 were 0.68 and 0.72, respectively. As a result, women’s children ever born and age had significant roles on their desire to have a child in all provinces while their opinion toward childbearing had just effect on provinces with TFR ≤ 2. In provinces with TFR ≤ 2, 10 - 39 years old women with 2 children and less CEB were willing to have another child. Women with 2 or less children at the age of 10 - 29 as well as women with 3 or more than 3 children at the age of 20 - 29 had a desire for more children in provinces with TFR > 2. If the women have a positive attitude toward childbearing and their age is not high, they will desire to give birth to a child until they reach the ideal number of their children. Thus, policies in changing their attitudes toward childbearing and creating the necessary facilities to prevent delays in marriage could be an important step in changing fertility rates.