![Sample our Education journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5932/TOC-Subject-Sample-2021-Education.jpg"})
ABSTRACT
This research set out to investigate how, in a post-conflict area, parental preferences and household characteristics affect school choice for their children. A multinomial logit is used to model the relationship between education preferences and the selection of schools for 954 households in Freetown and neighboring districts, Western Area, Sierra Leone. The increased economic well-being of a family tends to increase the likelihood of choosing a nongovernment school. As a child gets older parents are more likely to select government over nongovernment schools. For girls, parents are twice as likely to select a nongovernmental organization (NGO) school than a government one. Where parental preference for girls is a “safe environment” government is the preferred choice over a private proprietor school. Interestingly, the level of household education does not affect the likelihood of attending any school management type.
Declaration of interest
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received funding from the Sir John Templeton Foundation (Grant ID: 20842) for this research.
Notes
2. The variable “total number in the family” was included as in developing countries, households often include extended family members. Therefore, the number of family members may be an indicator of a family’s willingness to pay for education.
3. Measures show that the model fits the data well, with the likelihood ratio test (χ2(51) = 409.316, p < 0.0001), implying that the model as a whole fits significantly better than an empty model with no predictors. The Pearson statistic for the measure of goodness of fit (χ2 (2739) = 2894.033, p > 0.05) implies that there is no significant difference between the expected and actual values (Long, 1997; Train, Citation2009). Pseudo-R2 likelihood ratio indices: 18.6% (McFadden, Citation1974), 35% (Cox & Snell, Citation1989) to 38.8% (Nagelkerke, Citation1991).
4. For continuous independent variables the odds ratio is exp[SDx Coeff.] This gives an estimated odds ratio for an increase of 1 SD. Where a 1 standard deviation is a meaningful change in the respective continuous variable. Within this definition the dichotomous variables were taken to have an standard deviation of 1, giving the odds ratio of exp[Coeff.]. Each of the coefficients indicates the change in the odds that a parent selects a given type of school instead of a government school for a 1 SD increase in his or her importance rating of the respective school characteristic.
5. Diagnostic checks using tolerance and variance inflation factor methods were used to test for collinearity between the variables. It was found that no colinearity was present. The variance inflation factor check for “affordability,” “income,” and “expenditure” gave values of 1.418, 1.439, 1.435 respectively, all below 10 and close to 1. For the tolerance check, the values were 0.705, 0.695, and 0.697 (respectively), all above 0.2, again showing no collinearity. Collinearity diagnostics checks to see if the matrix was ill-conditioned, gave low eigenvalues, also indicating that the values were not dependent.
6. See Appendix for the estimates of the model for boys and girls separately
7. When running the MNL separately for boys and girls