以往認為學障學生因學習的困難與挫折，容易使其心理健康低落，他們能否和一般學生一樣，透過家長支持和自我效能，提升心理健康，仍待瞭解。本研究在106到107學年間進行三次調查，探討學障學生家長支持、自我效能與心理健康等因素長時間下的發展關係，利用一般學生為參照群體，瞭解兩個群體發展軌跡是否不同，採用潛在成長模式進行資料分析，納入國小一般學生459名和學障學生205名。結果顯示，一般學生自我效能一直都高於學障學生，兩個群體家長支持未隨時間有明顯差異，心理健康在一開始時，一般學生高於學障學生，最後反而學障學生為佳。發展關係上，都可見一般學生與學障學生的自我效能不僅在一開始會影響心理健康，成長速率之間也呈現正相關，但一開始有較高自我效能的一般學生，後續心理健康則越顯負向；兩個群體一開始的家長支持與同時間的心理健康能有效預測，學障學生家長支持與心理健康間的後續發展成正相關，而一般學生一開始家長支持對於後續心理健康呈現負相關。三者間的關係上，學障學生一開始的家長支持透過自我效能對個體心理健康產生影響，這與一般學生不同；學障學生家長支持與自我效能之間的發展變化則息息相關。最後依據研究結果，提出未來研究與實務上之建議。

Students with learning disabilities (LDs) are at high risk of experiencing social, emotional, and behavioral difficulties. However, protective factors, such as parental support and self-efficacy, can mitigate the negative effects of LDs on their mental health. In this field of research, literature findings are limited by a narrow focus on school-related factors, a lack of longitudinal examinations, and a poor understanding of the mechanisms underlying these factors. Furthermore, few studies have investigated students with LDs. This study employed latent growth modeling to explore the developmental trajectories and relationships among self-efficacy, parental support, and mental health over an extended period to clarify the psychological and social changes experienced by students with LDs. Students without LDs were used as a basis for clarifying whether students with and without LDs differ in terms of the aforementioned factors. Specifically, this study (1) determined the initial status of and changes in individual factors (i.e., self-efficacy, parental support, and mental health) among students with and without LDs; (2) determined whether significant differences exist between students with and without LDs in terms of their parental support, self-efficacy, and mental health; and (3) determined whether the initial status of and changes in parental support and self-efficacy among students with and without LDs influence the initial status of and changes in their mental health. To form the control group, that is, the group comprising students without LDs, 280 fourth-grade students and 280 fifth-grade students without LDs were sampled from elementary schools in Kaohsiung City; these schools were classified into four urbanization levels, and sampling proportional to the population distribution across urbanization levels was performed. After parental consent was obtained, 529 students without LDs were enrolled in this study; they comprised 268 fourth graders and 261 fifth graders from 13 elementary schools. To form the experimental group comprising students with LDs, 140 fourth-grade students and 140 fifth-grade students with LDs were selected from the general population , and the same form of proportional sampling was conducted. After parental consent was obtained, 257 students were enrolled in this study; they comprised 131 fourth graders and 126 fifth graders from 55 elementary school resource classes. If a participating student without LDs withdrew from this study because of a school transfer or if a participating student with an LD withdrew because of a school transfer, reclassification as an individual without an LD, or other reasons, then their survey response was excluded from subsequent analysis. The analysis included only students who completed the survey thrice. In total, 459 students without LDs and 205 students with LDs met this criterion. For data collection, the students without LDs were surveyed by regular classroom teachers during early morning self-study sessions, whereas the students with LDs were surveyed by resource teachers during resource classes. If a student experienced difficulty with reading, the surveying teacher would read the questions of the survey to them. This study developed and used its own assessment tools (i.e., a self-efficacy scale, parental support scale, and mental health scale) rather than used an existing tool. These tools, which have a low number of items, were developed on the basis of the literature definitions for self-efficacy, parental support, and mental health, and they were developed primarily to reduce the testing time and minimize the effect of repeated testing on the learning experience of the students. The self-efficacy scale comprised four items for measuring an individual's belief in their ability to perform the tasks required to attain various achievements; the items' reliability ranged from .69 to .73, and the Cronbach's α was .87. The parental support scale comprised four items with a reliability range of .83 to .85 and had a Cronbach's α of .95. The mental health scale comprised three items for measuring the degree to which students did not experience emotional or psychological problems; the items' reliability ranged from .60 to .67, and the Cronbach's α was .78. The items of all three scales were rated on a 4-point Likert scale with endpoints ranging from 4 to 1 (4 points, strongly agree; 3 points, agree; 2 points, disagree; and 1 point, strongly disagree). A higher score indicated superior performance. The researcher mailed the scales to the schools, and regular and resource teachers separately administered the scales to the students without and with LDs, respectively. To ensure the consistency and validity of the results, the researcher provided clear instructions for administering the scales. If a student with an LD experienced difficulty with reading the questions, their resource teacher assisted by reading the questions. The testing duration was 20 min. This study employed a longitudinal design and collected data thrice over 2 years. The first round of data collection was conducted between November and December 2017 (first semester of the 2017/2018 academic year), the second round was conducted between November and December 2018 (first semester of the 2018/2019 academic year), and the third round was conducted between May and June 2019 (second semester of the 2018/2019 academic year). To address the research questions, three types of latent growth model were used, namely univariate, conditional, and multivariate models. These models were used to analyze the developmental trajectories of self-efficacy, parental support, and mental health in both the general and LD student populations (i.e., control and experimental groups) and to identify the differences between these two populations in terms of these variables. The univariate model was used to describe the developmental trajectory of each group. The conditional model was used to identify differences between the two groups with respect to changes in their self-efficacy, parental support, and mental health. Finally, the multivariate model was used to investigate the interrelationships among these three variables over time. The study revealed that the self-efficacy of the students with LDs gradually increased over time; however, individual differences were also identified. The initial state of self-efficacy was revealed to be negatively correlated with the rate of improvement in self-efficacy; that is, those with a high initial level of self-efficacy tended to experience a slower increase in self-efficacy. By contrast, the self-efficacy of the students with LDs did not change significantly over time, and both groups exhibited individual differences in self-efficacy, with a negative correlation being noted between their initial state and rate of change. During the initial stage, the students without LDs exhibited a higher level of self-efficacy relative to those with LDs, and this difference persisted over time, with both groups exhibiting similar growth trajectories. The parental support received by the students without LDs gradually decreased over time. Although individual differences in the initial level of parental support received were identified among the students in this group, these differences diminished over time. The initial state of parental support was not significantly correlated with the rate of improvement in parental support. The parental support received by the students with LDs did not change significantly during the 2-year study period. Although significant individual differences in the initial state of parental support were identified among the students with LDs, these differences diminished over time; similarly, no significant correlation between the initial state of parental support and the rate of improvement in parental support was identified in this group. Finally, the two groups did not differ significantly in terms of the rate of improvement in parental support during the study period. For mental health, the participant students without LDs experienced a gradual decline in mental health over time, and individual differences were identified. Furthermore, a negative correlation was identified between the initial state of mental health and the subsequent rate of change in mental health. By contrast, the mental health of the students with LDs continued to improve over time. Although persistent individual differences in mental health were identified among these students, no correlation between their initial state of mental health and their subsequent rate of change in mental health was identified. A comparison of the mental health growth trajectories of the two groups revealed that the students without LDs initially exhibited better mental health relative to those with LDs but were subsequently overtaken by them over time. A negative correlation between self-efficacy and mental health was identified in both the students with and those without LDs. For the students without LDs, a high initial level of self-efficacy had a positive effect on their mental health, and the rate of change in self-efficacy was positively correlated with the rate of change in mental health. For the students with LDs, their initial level of self-efficacy predicted their initial mental health status, and the rates of change of both variables was positively correlated. The findings regarding the relationship between parental support and mental health are as follows. For the students without LDs, their initial level of parental support positively and negatively predicted their initial state of mental health and the rate of change in their mental health, respectively. For the students with LDs, their initial level of parental support was positively correlated with their initial mental health status, and the rates of change of both variables were positively correlated. In summary, this study revealed that for the students without LDs, their initial level of parental support was direct correlated with their self-efficacy and mental health and subsequently enhanced their self-efficacy, which improved their mental health. However, for the students with LDs, their initial level of parental support affected their self-efficacy, which in turn affected their mental health. In addition, for the students without LDs, their initial level of parental support was negatively correlated with their subsequent improvement in self-efficacy but positively correlated with the growth relationship between their parental support and self-efficacy. However, for the students with LDs, the growth relationship between their parental support and self-efficacy was the only observed phenomenon. Although the two groups differed in terms of psychological and social mechanisms, for both, parental support had an effect on student self-efficacy during learning.