Materials and Methods
Selecting of research subjects and survey practising methods:
The source population in this study was those residents dwelling in Jiayi county with age above 65 to 74 years old in April 2001, and two stage sampling mode was adopted to do sampling. In the first stage, the unit was township city, and four township cities were selected by the manner of conception sampling according to the geography characteristic of the township city, which selected by sequence were: Liujiao region with highest senior citizens’ ratio, Zhuqi region adjacent to mountains, Budai region neighboring with oceans, and Minxiong region with the tendency of urbanization. In the second stage, the study unit was individual person, and random sampling was done based on the data in the register of senior citizens in the population office of Jiayi county. Totally 1000 members of researching subjects were sampled to do the investigation, and totally 900 copies were collected, and after the eliminating of those with insufficient information to do analysis, the valid questionnaires summed up to 896 copies.
Design of the questionnaire:
The questionnaire in this study was in the style of self-made structured type, and had passed the exterior checking test of three experts in related fields, and a small pre-study was done to confirm its feasibility and suitability after obliterating or rectifying of unfit items. Besides the social anthropology items, the included contents also had two big parts: one part for social capital in questionnaire of neighbor living quality (30), and one part for depression screening questionnaire in Taiwan area (TDQ) (31).
Statistical analysis:
Firstly the description analysis was done by percentage and frequency distribution, and Chi-square test was used to do single factor analysis between melancholia, personal information and community social capital, and logistic regression analysis was made to estimate the effect of personal information and community social capital on melancholia.
Results:
1. Basic information for the samples studied:
There were totally 896 copies of valid questionnaires, in which there were 24.2% of people had the symptom of depression in the previous week. The male number of interviewees was 493 (55.1%), and for female it was 44.9%. Their mean age was 69.4 (SD 2.7) years old. There were 47.9% interviewees had no schooling experience or in illiteracy. There were 72.1% who married. 253 (28.2%) of them still had jobs at the time of interviewing. To living conditions, the majority of them lived with their children or grandchildren, the ratio was 36.4%. 76.1% of those aged people had no independent economy sources at present. There were 676 (75.5%) senior citizens with the mean family income lower than 30, 000 Yuan. And there were 69% of interviewees participated no social community activities in the past six months. About 64% senior citizens had physical diseases (table 1).
2. Relation analysis between personal information, social capital and melancholia:
Table 1 showed that higher melancholia ratio existed in those of female, no schooling experience or in illiteracy, no job, not married, no independent economy sources, mean family income lower than 30, 000 Yuan, participated no social community activities in the past six months, and worse social capital, and in statistics the difference was significant (x2=22.92, p=0.00; x2=29.83, p=0.00; x2=8.95, p=0.00; x2=6.36, p=0.01; x2=7.04, p=0.01; x2=8.70, p=0.00; x2=27.89, p=0.00; t=4.95, p=0.00. For different geographical areas higher melancholia prevalence rate existed in Zhuqi region, however, in statistics the difference was not significant（x2=5.66, p=0.13）. (Table 1)
Further logistic regression analysis was done to the variance of personal information and social capital etc. and the result showed that the prediction factors correlated with hypochondria ratio were sex, educational level, mean family income, whether existed other physical diseases, whether participated any social community activities in the past six months and social capital. In other words, higher ratio of hypochondria might exist in female, no schooling experience or in illiteracy, lower mean family income, existed other physical diseases, participated no social community activities in the past six months, and lower social capital (OR=1.53, 95%CI 1.06  2.22; OR=1.70, 95%CI 1.17  2.48; OR=1.72, 95%CI 1.13  2.62; OR=2.74, 95%CI 1.85  4.07; OR=1.91, 95%CI 1.23  2.98; OR=0.91；95%CI 0.88  0.94). (Table 2)

材料与方法
研究对象选取与调查实施方法
于2001年4月以年龄65-74岁以上且居住在嘉义县之居民为研究之母群体，并采二阶段取样方式取样。第一阶段以乡镇市为单位，依乡镇市地理特质以立意取样之方式选取四个乡镇市，依序选出的有：老人比率最多的『六脚乡』、依山的乡镇『竹崎乡』、靠海的乡镇『布袋镇』、以及较趋向都市型态的『民雄乡』。第二阶段则以人为单位，根据嘉义县户政机构老人资料名册，进行随机抽样。总计选取1000名左右研究对象，进行调查研究。总计回收900份，剔除资料不足无法分析者后，共计得896份有效问卷。
问卷之设计
本研究是以自制结构式问卷，且经由三位相关领域专家进行问卷表面校度之检测。将不适之题目剔除或修正后，进行一小型前驱研究，以确定其可行性与适切性。内容除了基本社会人口学变项外，尚包括两大部分：邻里生活品质问卷中之社会资本的部份[30]、台湾人忧郁筛选问卷(TDQ) [31]。
统计分析
先以百分比及次数分配进行描述分析，并以卡方检定 (Chi square) 作为个人属性、社区社会资本与忧郁症间之单变项分析；以逻辑回归分析估算个人属性、个人社区社会资本对于忧郁症之效应。
结果
一、研究样本之基本属性
有效问卷共为896份，过去一周有忧郁症状者有24.2%。受访者中有男性493名(55.1%)，女性44.9%，平均年龄为69.4(SD 2.7)岁；47.9%的受访者未就学或不识字；已婚占72.1%；有253人(28.2%)的受访者目前仍然有工作；居住状况则以与子或孙同住者为较多，有36.4%；有76.1%的老人目前并无独立的经济来源；而家庭平均收入三万以下者有676名(75.5%)，有69%受访者在过去六个月中未曾参与任何社会团体活动。约64%的老人有身体上的疾病。（表一）
二、个人属性、社会资本与忧郁症关系分析：
由表一显示，女性、不识字或未就学、无工作、非已婚者、无经济独立来源者、家庭平均收入在30,000元以下、最近六个月并无参加任何社会团体活动、社会资本较不佳则会有较高的忧郁症比例，且统计学上也达显著的意义（x2=22.92, p=0.00; x2=29.83, p=0.00; x2=8.95, p=0.00; x2=6.36, p=0.01; x2=7.04, p=0.01; x2=8.70, p=0.00; x2=27.89, p=0.00; t=4.95, p=0.00）。在地理区位差异上以竹崎乡有较高的忧郁症盛行率，但统计学上并未有显著差异（x2=5.66, p=0.13）。（表一）
    进一步将个人属性与社会资本等变项以逻辑回归分析，结果发现，与忧郁症发生率相关之预测因素为性别、教育程度、家庭平均收入、是否有其它身体疾病、最近六个月是否参与社会团体活动、社会资本。换言之，女性、未就学或不识字、家庭平均收入愈低、身体上有其它疾病、最近六个月无参与任何社会团体活动、社会资本越低者则会有较高的忧郁症发生率(OR=1.53, 95%CI 1.06  2.22; OR=1.70, 95%CI 1.17  2.48; OR=1.72, 95%CI 1.13  2.62; OR=2.74, 95%CI 1.85  4.07; OR=1.91, 95%CI 1.23  2.98; OR=0.91；95%CI 0.88  0.94)。(表二)