本文旨在以貝氏分析法來探討誤差項具有自我相關的迴歸模型。全文一冊約三萬兩仟 字，共分為六章，十二節。內容如下： 第一章 緒論：說明迴歸模型，自我相關誤差的意義，及貝氏分析法之理論體系。 第二章 誤差項具有一階自我相關的簡單迴歸模型：分析以傳統抽樣理論法及貝氏分 析法對模型作分析並比較其結果。 第三章 多元迴歸模型：以貝氏法分析自我相關誤差之多元迴歸模型。 第四章 事前分配及其他假設的考慮。 第五章 我國民間消費與個人可支用所得迴歸模型的分析。 第六章 結論。

台灣金融機構得以至中國大陸地區設置分支機構的法源主要來自於立法院在於2006年11月時修正准許台灣銀行業赴中國大陸地區設立分行。本研究針對39家台灣銀行業在中國大陸營業據點之經營策略與績效進行分析，並採採多元迴歸模型檢定其經營策略與績效之關係。 根據本文實證後的主要研究結果如下： 一、無論是以資產報酬率 (ROA) 或是股東權益報酬率 (ROE) 等兩項獲利指標所檢定的結果，均未有足夠的證據顯示台灣銀行業在中國大陸地區積極設置分行等據點的經營策略或方向是可行的。 二、由股東權益報酬率 (ROE) 分析的結果，支持台灣銀行業若採行提早投入中國大陸地區設置實體據點的發展策略，對於其股東權益報酬率 (ROE) 會呈現顯著的正向關係。 三、無論是以資產報酬率 (ROA) 或是股東權益報酬率 (ROE) 來做為被解釋變數，結果皆無法支持台灣銀行業在中國大陸地區設置分行等實體據點的數量規模大小與銀行的資產報酬率 (股東權益報酬率) 有關。 / Because of the Legislature amended the Act in November 2006, Taiwan’s Banking had been permitted and set up branches in mainland China since 2006. This paper analyzed the business strategy and performance of 39 Taiwanese banks in mainland China. The multiple regression models were adopted and we test the relationship between their business strategy and performance. The main results of this paper were as follows: First, whether the rate of return on assets (ROA) or return on equity (ROE) profitability indicators, there is no sufficient evidence that the Taiwan’s banks actively set in mainland China Branches and other positions of the business strategy or direction is feasible according to the other two test result. Second, as a result of the ROE analysis, it is significant to support the Taiwan’s banking industry to adopt the development strategy of setting up an entity base in mainland China, which has a significant positive effect on the ROE. Third, whether the return on assets (ROA) or the return on equity (ROE) as an explanatory variable, the results are unable to support that the the scale of Taiwan’s banking in mainland China have a relation with the ROE or ROA.

本研究以TEPS資料庫第二、三波追蹤樣本為資料來源，使用多元迴歸分析中的OLS（ordinal least squares）最小平方法，分析不同管教方式之父母比較行為如何影響子女之綜合能力。刪除作答不全的缺失值後，共計取得有效樣本為2,595人。 研究結果顯示：父母管教方式若以雙向度作為分類，以高反應（74.6％）與高要求（69.2％）者居多，綜合兩大面向後，有52.40％的父母採取開明權威型的管教方式，有22.20％的父母為專制權威型，有16.90％的父母為寬鬆放任型，有8.50％的父母是忽視冷漠型。此外，父母最常把子女與其同班同學（45.5％）進行比較，其次依序為鄰居小孩、同事或朋友小孩（42.0％）、同年齡的親戚（37.0％）、兄弟姊妹（26.9％），父母最少拿自己小時候（18.8％）與子女相比。 經由多元迴歸分析之後得知，無論是開明權威、專制權威、寬鬆放任或忽視冷漠型的父母，凡拿子女與其「同班同學」比較者，子女的綜合能力越高，達到統計顯著水準；但是當父母持專制權威的管教方式時，無論有無加入控制變項（子女性別、父母教育程度和家庭月收入），有與「同班同學」和「父母小時候」比較者，其綜合能力顯著越高，但與「同年齡親戚」比較者，其綜合能力顯著越低。 / The research was to explore the effects of parenting style, parental comparison, and the interaction of them on children’s general ability. Data were collected from Taiwan Education Panel Survey (TEPS), the core panel in the second and the third wave. Total number of the students from high school reduced to 2,595 people after listwise deletion. We explored the effects by using ordinal least squares in the multiple regression analysis. The results showed that there were 74.6% high-respective parents, and 69.2% high-demanding ones. Mixed with the two dimensions, demanding and responsive, there were 52.40% authoritative parents, 22.20% authoritarian ones, 16.90% indulgent ones, and 8.50% neglectful ones. Besides, parents compared their children with their classmates (45.5%), neighbors (42.0%), cousins (37.0%), siblings (26.9%), and parents’ childhood (18.8%). After regression, the results showed in the following ways. First, no matter what the parenting style was, children who were compared with classmates had significantly higher ability. Whether adding the control variables like children’ gender, parents’ education level and family monthly income or not, when parents were authoritarian, children compared with classmates and parents’ childhood had significantly higher ability. Meanwhile, if they were compared with cousins, then they had significantly lower ability. Keywords: Taiwan Education Panel Survey (TEPS), multiple regression analysis, parenting style, social comparison, general ability

台灣資本市場目前已實行競價拍賣制度有十餘年之久，相關研究資料累積已具相當之規模，因此許多學者進行競價拍賣制度對於承銷制度改變影響之研究，已具相當之成果，大多針對競價拍賣制度實施後是否可以對於改善上市掛牌後股票折價幅度過大以及蜜月期過長等不效率現象進行相關探討及研究。 因此，本研究針對投資人標單資訊內涵進行對報酬率及價格變動率影響之研究，於建立兩大模型研究結果分析後，我們可以清楚了解台灣競價拍賣市場中，個人投資者標單資訊內涵可以提供蜜月期報酬率及變動率的正向解釋能力與預測效果，也因此證明台灣股票市場報酬率及價格波動程度容易受一般散戶投資人情緒所影響；而法人投資者標單資訊內涵則可能由於與美國市場具有結構性差異，使得台灣市場的法人投資者標單資訊內涵無法提供模型的顯著解釋能力，肇因於台灣市場主要參與者為散戶投資人。另外一原因可能是台灣市場中的法人投資者標單資訊內涵已反映在IPO股票價格之中，因此無法對蜜月期及上市後報酬率和價格變動率有所影響。加入峰態係數進行研究後發現，投資者標單需求曲線型態對於報酬率以及價格變動率皆沒有顯著影響，顯示投資人出價集中的高狹峰型態不能解釋蜜月期及上市後的報酬率和價格變動率，可能是由於資訊早已暴露在市場中，價格也早已反映此項資訊，因此無法解釋之後的報酬率以及價格變動率型態。

在可預見之未來無線點對點的世界裡，資訊交易就如同現實世界人與人間的交易模式，需考量到交易資訊的成本、價值與人際關係衡量之因素；傳統廣告行銷的效率問題也待創新的資訊交易平台來解決。本研究提出的資訊交易模型，以人性考量為基礎，在各種不同情境下，動態衡量資訊成本、價值與使用者間關係，透過助理軟體，協助加入資訊交易平台之個體，以更便利方式進行資訊交易，預期讓交易結果更貼近使用者的需求；而企業方面也能透過點與點之間快速傳遞資訊的特性，預期以更低成本、更高效率，完成商務行銷目的，達到破壞性行銷之目標。 / This paper presents a novel ambient e-service aiming at distributed marketing through sensible bartering in foreseeable wireless Peer-to-Peer (WP2P) environments. A variety of influential factors (e.g., cost, value, relationship) are proposed and formalized for empowering the bartering mechanism, unfolding a rich arena of ambient distributed trading and a disruptive paradigm of e-marketing.

網際網路的發展讓人們的生活起了變化，Web2.0的概念更是增加了人們對網際網路的依賴性，我們成為網路內容的生產者、我們在社交網站上發表、追縱朋友的動態，以及取得全球世界各地的資訊。在這無限的虛擬空間中隱含的巨大商機，讓各大企業紛紛而至，因而加速了Web2.0社群網站的發展，維持與增加網站流量更是成為社群網站生存的關鍵與重要的績效指標。但社群網站該如何從流量指標之變化來評斷社群網站之績效呢?這是令我們最好奇之處。 藉由Google Analytics提供的流量分析工具，本研究蒐集了台灣四間社群網站1-3年間的流量資訊進行分析，考量蒐集之資訊具時間序列性質特性，本研究首先採用移動視窗法重新進行資料的整理，並據此概念應用在後續的統計分析。此外，本就以指數加權平均法及多元迴歸分析進行流量異常值之偵測，最後，對照各網站重大事件里程碑並與各網站業主進行一對一深訪。故本研究實際上包含質、量化之分析結果。 本篇研究四間個案網站為例，並依網站創造的服務與使用者互動情形流量將其區分為社交互動型與資訊交換型網站，並歸納其在網站流量指標上不同特徵表現及各自可參考之績效評估指標。同時，本研究採用多元迴歸分析做為社群網站績效評估模型，並企圖建構一績效評估分析流程期以做為後續研究者針對網站流量相關研究之參考。 / The development of Internet makes a great influence on human society and the development of Web2.0 enhances human’s dependence on the internet and becomes a channel of social connections. Currently, most contents of the Internet are generated by common users who could retrieve information through the entire network and trace their friends’ actions over the Social Network Sites (SNSs).Owing to the potential business opportunities on the internet, companies try to enter the market causing the prosperities of SNSs. Maintaining or even increasing traffic flows become a critical issue for SNSs to survive in the competitive market. However, how to evaluate the performance of SNSs based on traffic flow indices remains unsolved.This study collected Google Analytics data for 1-3 years from four SNSs’, respectively.Consider the time series charactics, this study applied “Moving Windows“ to organize the data for further statistical analysis.In addition, Exponentially Weighted Moving Average and Multiple Regression Analysis were used to detect the abnormal traffic flows. Finally, these abnormal records were compared with the important events and one-on-one interviewings with the SNSs operators were conducted. The results of this study are based on qualitative and quentitative analysis. This research studiesd four SNSs that were categorized into information-oriented and interaction-oriented services based on their services and users’ interaction. The SNSs at different categories behaved differently following certain characteristics defined previously.A performance evaluation process was developed as a reference for further studies.