中文摘要：

赌徒谬误指当某一独立随机事件发生后，人们倾向认为这一结果再次出现的概率降低。如果一连串的随机事件呈现出一定的趋势，人们倾向于认为随机事件将呈现系统性反转。证券市场中的赌徒谬误指在股票上涨（下跌）序列中做出股价将要下跌（上涨）的判断。本研究探讨股票市场特征（趋势长度及方向）及投资者的人格特征（自我效能感）对赌徒谬误的影响。以83名股票投资者为对象，采用多层线性模型进行分析，结果发现：趋势长度主效应显著，短线情境下赌徒谬误频次更高；趋势方向主效应显著，下跌情境下赌徒谬误频次更高；二者交互作用显著，在短线下跌情境下，赌徒谬误频次更高；投资者的自我效能感对股票趋势长度具有调节作用，高自我效能的投资者在短线情境下更容易出现赌徒谬误。

英文摘要：

Gambler＇s fallacy is the belief that if deviations from the expected behaviors are observed in repeated independent trials of some random processes, future deviations in the opposite direction are then more likely~ When people judge the probability of an uncertain event, gambler＇s fallacy often occurs. Stock performance is also uncertain. So gambler＇s fallacy appears in the stock market investment. The purpose of this research is to investigate when the gambler＇s fallacy occurs in stock investment and what affects its frequency. Besides, how the investors＇ personality, especially their self-efficacy （SE）, moderates the relationship between stocks＇ factors and gambler＇s fallacy is also discussed. A total of 83 stock investors in China participated in this research. In order to measure the frequency of gambler＇s fallacy, stocks＇ K-lines with trends （up or down） were used in this study. According to four kinds of trends （short-up, short-down, long-up, long- down）, we chose 10 pictures for each of these trends. Subjects were asked to judge the current trend （up or down） and the future price （ increase or decrease） of each of the stocks. When they suggested that the price wound increase in a down-trend or decrease in an uptrend, gambler＇s fallacy occurred. Besides, Schwarze＇s general self-efficacy scale was used to measure investors＇ self-efficacy （α = . 86）. The hierarchical linear modeling （HLM） was used in data analysis because of the nested strueture, i. e. , the trend duration and direction as the intra-personal variable was nested in self-efficacy （personal variable）. The results indicated that the duration （β = -. 65, t = -2.44, p 〈. 05） and direetion （β = -. 66, t = -3. 424, p 〈. 01 ） of the trend have both main effeet and interaction effect （β= . 80 ,t = 2. 80, p 〈. 01 ）. Specifically, in the down trend, the frequency of gambler＇s fallacy was more in short-trend than in long-trend （β down trend = -. 65 ,t = -2.44, p 〈 . 05