先前关于留守儿童依恋的研究大多属于横断面研究，难以考察留守儿童的依恋随时间发展的特征和变化趋势。本研究分别于2015, 2016和2017年三次采用自编背景资料调查问卷和修改版的父母与同伴依恋问卷(Inventory of Parent and Peer Attachment, IPPA，对安徽省合肥市肥东县某中小学四至九年级1061名学生的父母或照料人依恋进行了追踪调查。旨在为留守儿童的相关研究和心理干预丁作捍撰科学的理讼佑据。
本研究分为两部分。首先，计算三次追踪调查中儿童与父亲、母亲或照料人之间的依恋得分并对其进行潜类别增长模型分析(Latent Class Growth Analysis,LCGA)，根据分析结果确定最佳轨迹类型数量，并刻画依恋发展轨迹，考察不同个体所遵循的不同依恋发展轨迹以及遵循不同依恋发展轨迹的个体类别;同时探讨了留守儿童父亲、母亲和照料人依恋水平随时间变化的趋势和依恋发展的稳定性。其次，采用多元逻辑回归模型分析了留守儿童依恋发展轨迹的影响因素。
(1)总样本的母亲依恋发展轨迹存在三种类型:Gl C6.1%、G2 C34.4%和G3 C59. 5%;而父亲依恋发展轨迹也存在三种类型:G1 C3.3%、G2 (30.1%)和G3 C66.6%)。其中96.7%的儿童的父亲依恋和93.9%的儿童的母亲依恋相对稳定发展，而3.3%的儿童的父亲依恋和6.1%的儿童的母亲依恋发生了明显改变。由此说明，依恋是相对稳定的，但也会根据环境和经历发生改变。
(2)留守儿童的母亲依恋发展轨迹存在三种类型:G1C5.8%), G2C33.4%)和G3 C60.8 %);父亲依恋发展轨迹存在两种类型:Gl(26.5%)和G2 (73.5%);照料人依恋发展轨迹存在四种类型:Gl C4.2、G2 C12.1、G3 C37.6 %)和G4 C46.0 %)。观察各轨迹发现，留守儿童与父亲的依恋关系较好，且父亲依恋相对稳定发展。而相比父亲，留守儿童的母亲和照料人依恋相对不稳定发展。
Attachment is a special emotional bond between the caregiver and the infant,
which is relatively stable,but still changes according to the different living environment and experiences. Left-behind children refers to the children who are under 18 and left in their rural hometown when one or both of their parents migrated to urban areas for work. Left-behind children' s attachment is relatively lower than non-left-behind children because they separate with parents perennially or repeatedly and growing up in a situation of parental absence and unstable caregiving.
Previous studies mostly focus on cross-sectional studies, which are less possible to examine the characteristics and trends of the attachment of left-behind children over time. In this study, we measured 1061 children three times in 2015, 2016 and 2017 using the modified version of the Inventory of Parent and Peer Attachment (IPPA) scale, aiming to provide scientific theoretical basis for the research and intervention of left-behind children This study consists of two parts. In part 1，we calculated the attachment scores of children with their father, mother, or caregivers in each survey and analyzed by latent class growth analysis. According to the results of analysis, the number of optimal trajectory types was determined and the developmental trajectories of attachment was characterized. Part 1 aims at examining the different developmental trajectories of attachment followed by different individuals and the individual categories of different developmental trajectories of attachment. It also aims at exploring the trends of the attachment levels of left-behind children with fathers, mothers, and caregivers over time and the stability of attachment development. In part 2, we analyzed the factors which affect the developmental trajectories of attachment among left-behind children by multiple logistic regression.
The main conclusions are as follows:
(1) There are three developmental trajectories of maternal attachment are found in all students: Gl(6.1%), G2(34.4%), G3(59.5%); there are also three developmental trajectories of paternal attachment: Gl(3.3%), G2(30.1%), G3(66.6%). In maternal and paternal attachment trajectories, 96.7% and 93.9% of children shows relatively stable maternal and paternal attachment. However, 3.3% and 6.1% of children shows notable variability in maternal and paternal attachment. These results explains that attachment is relatively stable, but it also changes because of environment and experiences.
(2) There are three developmental trajectories of maternal attachment among left-behind children: Gl(5.8%), G2(33.4%), G3(60.8%); there are two developmental trajectories of paternal attachment: Gl(26.5%), G2(73.5%);there are four developmental trajectories of caregiver attachment: Gl(4.2%), G2(12.1%),G3(37.6%), G4(46.0%). By observing the trajectories, we find that the left-behind children have an better attachment relationship with their father, and their paternal attachment development is relatively stable. Comparing with paternal attachment, the maternal and caregiver' s attachment is relatively unstable. This shows that the “left-behind”condition has a greater impact on the maternal and caregiver' s attachment.
(3) There is a difference in the developmental trajectories of attachment between left-behind children and non-left-behind children. The developmental trajectories of maternal attachment are more unstable. One of the trajectories of maternal attachment shows a trend of increasing first and decreasing later, and it is unstable. In this study,we believe that the instability of this trend meant a negative development of attachment. In addition, the level of maternal and paternal attachment is lower than that of non-left-behind children. These results show that the“left-behind”condition had a negative effect on children' s attachment development.
(4) The grade, the left-behind type, the left-behind age and the frequency of
telephone contact with parents are important effect factors of developmental trajectories of maternal and paternal attachment. No matter fathers and mothers go out to work, the attachment relationship between left-behind children and their parents would be significantly effected. At the age of their parents' left, the younger the child was, the more negative impact he would experienced. However, a higher frequency of telephone contact may play a moderating role on children's parental attachment relationship. The results bring theoretical and practical implications to the intervention of left-behind children's attachment problems.