本研究的目的，主要在探討精神病患家屬參加座談會的有關因素。以民國七十一年三軍總醫院精神科320個住院病患的家屬進行問卷調查，同收了158份有效問卷，做爲本研究的主要研究對象，發現有107個家屬（家庭）表示願意參加，家屬的參加意願與病人的婚姻狀況有關——未婚的病人，家屬的參加意願較高，其中有59個家庭出席了座談會，家屬的出席與否，與病人的就醫身份、年齡、籍貫、婚姻狀況、社會階層、住院次數等無關，但與病人的性別及臨床診斷有關。病人（家屬出席座談會者）出院後的醫療狀況與病人的就醫身份及主要照顧者的教育程度有關——民眾身份的病人及主要照顧者教育程度愈高的患者，規則門診治療的比例較高。家屬是否得知病人診斷情形，與病人的就醫身份、社會階層有關——民眾身份的病人以及社會階層較高的患者，家屬愈能在病人住院期間或經由座談會的途徑得知病人的診斷情形。以家屬的求醫行為而言，社會階層愈低的家屬，愈認為傳統的求神拜佛、中藥治療等，是治療病人最有效的方式。

The purpose of this study is to understand the related factors which affect psychiatric patients' families to join the family education group. There had been 320 patients at the Psychiatric Department of Tri-Service General Hospital during 1982. After mailing 320 questionnairs, we received 158 of it as the main objects of this study. One hundred and seven out of 158 patients' families are willing to join our family group. The study concludes that the willingness of the patients' families attending the group has not been in relation to the patients' sex, age, domicile, marital status, social economic condition, the frequency of hospitalization and the patients' identity. Those families of male patients and with the clinical diagnosis of affective disorder for their mentally disturbed relatives tend to attend the group actively. Whether the patients (whose families are present at the group) would have had regular treatment or not after discharge has concern with their identity and the education level of families. It means that the civilian patients' families and higher education level families have greater rate to receive regular treatment at OPD. Additionally, whether the family members knew the patients' mental condition or not are correlated with the patients' identity and social class as well. It is to say that the civilian patients' families and families with higher social class tend to get more information about the disease entity. Meanwhile, the families with lower social class prefer to call for medical helps mainly from god worship, practitioner of herb medicine, and some other nonprofessionals.