在一般調查問卷裡，里程碑事件常被用來當作詢問題項時間的參照點。不過，目前的文獻對於里程碑事件性質的探討還很欠缺。本研究針對全民健保開辦的里程碑事件，以臺北市民為對象，用運續兩年所蒐集的電話訪問資料，探討時間效應、人口變項和受訪者答題正確性以及答題時間指稱語之間的關係。研究結果發現：（1）雖然幾乎所有的民眾都知道全民健保開辦這回事，但是只有四分之三的人聲稱記得開辦日，而實際上記憶正確的民眾比例更低。（2）就全民健保開辦日之記憶謬誤而言，存在前進遠望的比例高於後退遠望的現象。（3）隨著事件區間越來越小甚而趨近於點時，由於記憶難度逐漸增高，會產生受訪者答題正確比例越來越低的情形。（4）答題的正確性存在有時間效應的現象。訪問時間距離事件發生時間越長，則答題正確性越低。此狀況對年區間、年月區間和年月日點而言皆然。（5）各人口變項對答題正確性的影響，隨著答題正確性指標的不同而有差異。能正確說出開辦年月日的民眾，其平均年齡較記憶謬誤之民眾為低。但是在年區間和年月區間上，年齡無差異現象存在。一般來講，女性對時間記憶的表現優於男性。（6）受訪者所使用的時間指稱語詞存在時間效應的現象。距離事件發生日越久，完全運用詳細數字來指稱時間的傾向也就越明顯，雖然其答案不見得正確。（7）女性比男性、年輕者比年長者更傾向於完全使用詳細數字方式來回答時間記憶題目。（8）完全使用詳細數字來答題者，其答題正確性相對地比運用至少一個概詞者來得高。最後在文末，筆者對於這些研究結果在調查研究方法上的意義作討論，並提出未來研究里程碑事件的若干建議。

　　　　　Landmark event refers to events significant to individuals or the society such as one's birthday, wedding day, or the first presidential election day, etc. These events are often used as reference points when asking survey questions related to time. The survey methodology literature, however, rarely discussed the nature of landmark events. This paper aims at analyzing the relationships among demographic variables, time effect, term usage referring to time, and response accuracy by choosing the implementation of the National Health Insurance (NHI) as a landmark event. The subjects of the study are residents in Taipei area. Data are collected from a two-year cross-sectional telephone survey. Three different indices of time accuracy are employed: (1) day-month-year accuracy, (b) month-year accuracy, and (c) year accuracy only. There are several major findings: (1) Although almost all respondents know the NHI, only three-quarter of them claim that they remember the beginning date of the NHI. The proportion of respondents who can correctly report the date is even lower. (2) There are more forward telescoping than backward telescoping among respondents who made recall errors. (3) Response accuracy decreases as the difficulty of time recall increase. For example, compared to reporting the day-month-year of NHI occurring, higher rates of accuracy are found in indicating the month-year reporting. (4) Larger interval since the landmark event leads to the lower response accuracy. Such a time effect appears no matter which kind of time accuracy measure is used. (5) More young respondents correctly report the NHI starting date than older respondents. (6) The longer the recall date since the event occurring, the more likely the respondents use specific numbers to refer the event-happening time. (7) Female and younger respondents tend to use exact numbers when locating event time. (8) Respondents who use exact numbers to indicate time show higher rate of response accuracy than those who use at least one general term to indicate time.