Application of statistical methods in the analyses of foster family

Date

2023-9-08

Author

Atar, Gizem

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

122
views

51
downloads

Cite This

The foster family system is a service provided for children between the ages of 0- 18, who are not suitable for living with their biological family due to various reasons, and who live in institutional care under state protection. Hereby, an alternative care model created by considering the best interests of the child in the form of staying in a family environment where there is one-to-one care, love and respect, instead of institutional care for the physical, emotional, and psycho-social development of the child. In our study, we deal with the Turkish family structure whether the family accepts to be a foster family. For this purpose, we use the Research on Family Structure in Türkiye (TAYA) 2016 data and analyze which social effects influence this decision significantly. In our assessment, we apply lasso method for dimension reduction approach and various statistical modeling as well as machine learning methods for regression purposes. Furthermore, in our analyses, we implement them as single methods and two-stage modeling approaches in such a way that initially clustering methods are performed to identify the correlated variables with the decision of foster family and then, these selected variables are used to construct mathematical models for the data. Looking at common variables among all models, there are some demographic variables and the others are essentially about people’s thoughts and beliefs about children, mostly adopted or fostered. Different models work well for different types of datasets, such as generalized additive models for categorical data and machine learning models for classification and regression tasks. Spline-based models can also be effective on such datasets. Overall, there are many modelling options that exist for social datasets.

Subject Keywords

Generalized additive models, Statistical modeling, Dimension reduction, Foster family in Türkiye, Research on Family Structure in Türkiye (TAYA)

URI

https://hdl.handle.net/11511/105468

Collections

Graduate School of Natural and Applied Sciences, Thesis