影响生育价值观因素的研究－传统回归分析与多层线性模型分析结果的比较

Based on the research of predictors of VOC, this study explores the predictive effect of factors, such as generation, urban/rural context, collectivism/individualism orientation, family value, independent/interdependent self, adult attachment, on the Emotional and Traditional factors of VOC. Considering the hierarchical data structure of the VOC study, which resulted from the original research design, this dissertation applies Hierarchical Linear Model (HLM) after using traditional regression. A comparison between the results from the tow statistical methods is made, and the results are as follows: 1) Reliability coefficients of questionnaires used in this study are satisfactory, and most of them can be used in further research. 2) Samples from different generation and urban/rural context show significant differences on the score of collectivism/individualism orientation, family value, independent/interdependent self, adult attachment, and VOC. 3) Regression equations with VOC as outcome variable differ from each other when using data from sample with restricted generation or urban/rural context. 4) Results by HLM shows that interdependent self and mother identity have positive effect on emotional factor of VOC. Emotional factor’s variation on family level is not significant. 5) Results by HLM shows that Individualism, Interdependent Self and Grandmother Identity can predict Traditional factor of VOC. Traditional factor’s variation is significant on family level, which can be explained by family income and it’s area-urban or rural. Based on the results above, the researcher concludes that a) generation identity and urban/rural context have important effect on VOC; b) Interdependent Self is an important predictive factor of VOC’s Emotional factor, which is nearly subjective to other factors; d) VOC’s traditional factor varies with other factors, which show its strong relation with culture and tradition; e) more exact results can be gotten from HLM analysis, which beyond tradition regression.