Undergraduates’ Knowledge Attitude and Behavior (KAB) towards the Disclosure of Personal Data Online in China

For reliability data for this questionnaire, Cronbach's alpha was employed. The statistical analysis's findings are presented in this section, and the aforementioned research questions are put to the test. We investigated college students' understanding of online privacy breaches. The average inter-item correlation and average total item correlation were used to assess Cronbach's alpha (.905). In this preliminary analysis, after removing certain questions, the valid questions were Qs: 1, 3, 4, 5, 7, 10, 12, 14, 18, 21, 22, and 27. We analyzed college students' attitudes toward online personal information leakage. The average inter-item correlation and average total item correlation were used to assess Cronbach's alpha (.896). After removing certain questions, the following questions were determined to be valid in this preliminary analysis Qs: 32, 33, 34, 35, 36, 38, 41, 42, 43, 45, 46, 47, 49, 53, 54, 55, 56, 57, and 58. We analyzed the behavior of college students' online personal information leakage. The average inter-item correlation and average total item correlation were used to assess Cronbach's alpha (.857). Through preliminary examination, it was determined that the following questions were valid Qs: 60, 61, 62, 63, 65, 66, 67, 68, 69, 70, 71, 72, 76, 77, 79, 80, 81, 82, and 85. The variables' normality was examined. Because there were more than 50 participants, predetermined tests like Kolmogorov-Smirnov and Shapiro-Wilk were inappropriate. A manual evaluation was used to determine the skewness and kurtosis values. The term "normal distribution" was used to describe samples with skewness between -2 and +2 and kurtosis between -7 and 7. Average knowledge (skewness = -.237, kurtosis = 1.307), attitude (skewness = -.394, kurtosis =-.657), and behavior (skewness = -.166, kurtosis =.351) were all within the acceptable threshold range and were normally distributed. Figures 1-3 show that the data develops roughly along the normal distribution line in the quantile plot, with some deviations in the tail. Based on these figures, we can safely assume that this set of data is normally distributed. Because equal variance is not achieved, Pillai's Trace can only be used for analysis, (Anderson and Walsh, 2013) where the significance test shows that the significance value of gender is 017 [41]. If the value is 0.01P0.05, the difference is significant. As a result, the analysis reveals that only gender can detect the difference. Another test of between-subjects effects analysis revealed that in the significance test, only gender and mean knowledge had an effect (<.001), indicating that there was a difference between males and females. Differences between grades, mean knowledge, mean attitude, and mean behavior of different years are shown in Figure 4. Figure 4 depicts the average values and differences in knowledge, attitude, and behavior of different grades regarding the disclosure of personal information online. The mean values of the freshman's knowledge, attitude, and behavior are 2.330, 2.768, and 2.776, respectively. Sophomores’ mean values for knowledge, attitude, and behavior were 2.875, 3.090, and 2.901, respectively. Juniors' mean knowledge, attitude, and behavior scores were 2.974, 3.075, and 2.898, respectively. Seniors’ mean values for knowledge, attitude, and behavior were 2.542, 2.974, and 2.925, respectively. In terms of knowledge, the difference between freshmen and juniors was the largest. In terms of attitude, the difference was greatest between freshmen and sophomores. In terms of behavioral averages, the freshman and senior years differed the most. Less understanding to understanding, less agreement to mostly agreement, and less agreement to mostly agreement are the mean values for knowledge, attitude, and behavior, respectively. Figure 5 shows the average values and professional differences in the three aspects of knowledge, attitude, and behavior for various majors. In general, the mean values of knowledge, attitude, and behavior in the liberal arts are 2.599, 2.926, and 2.806, respectively. The mean values of knowledge, attitude, and behavior in engineering are 2.442, 2.800, and 2.895, respectively. The mean values of knowledge, attitude, and behavior in science were 3.000, 3.204, and 2.904, respectively. In terms of knowledge, the difference between engineering and science was the largest. In terms of attitude, the difference was greatest between engineering and science. In terms of behavior, the difference was greatest between science and engineering. Less understanding of understanding, less agreement to mostly agreement, and less agreement to mostly agreement are the mean values of knowledge, attitude, and behavior, respectively. Figure 6 depicts the average values and gender differences in knowledge, attitudes, and behaviors. In general, the mean values of boys' knowledge, attitude, and behavior were 2.922, 2.958, and 2.968, respectively. Girls' mean knowledge, attitude, and behavior scores were 2.298, 2.858, and 2.719, respectively. Less understanding of understanding, less agreement to basic agreement, and less agreement to basic agreement were associated with lower mean values of knowledge, attitude, and behavior, respectively.

为获取本问卷的信度数据，本研究采用了克朗巴哈系数（Cronbach's alpha）。本节将呈现统计分析结果，并对前述研究问题进行检验。本研究针对大学生对网络隐私泄露的认知情况展开了调查。我们以题项间平均相关系数与题项-总分平均相关系数作为克朗巴哈系数的评估指标，其值为0.905。在本次初步分析中，经剔除部分题项后，有效题项为Q1、Q3、Q4、Q5、Q7、Q10、Q12、Q14、Q18、Q21、Q22及Q27。

本研究同时分析了大学生对网络个人信息泄露的态度。同样采用题项间平均相关系数与题项-总分平均相关系数评估克朗巴哈系数，其值为0.896。经初步分析剔除无效题项后，有效题项为Q32、Q33、Q34、Q35、Q36、Q38、Q41、Q42、Q43、Q45、Q46、Q47、Q49、Q53、Q54、Q55、Q56、Q57及Q58。

本研究还针对大学生网络个人信息泄露相关行为展开了分析。本次分析的克朗巴哈系数采用相同指标进行评估，值为0.857。经初步筛选后，有效题项为Q60、Q61、Q62、Q63、Q65、Q66、Q67、Q68、Q69、Q70、Q71、Q72、Q76、Q77、Q79、Q80、Q81、Q82及Q85。

随后对各变量进行正态性检验。由于参与者数量超过50例，柯尔莫哥洛夫-斯米尔诺夫（Kolmogorov-Smirnov）检验与夏皮罗-威尔克（Shapiro-Wilk）检验等预设检验不再适用，因此通过手动计算获取偏度与峰度值。本研究将偏度介于-2至+2、峰度介于-7至7的样本视为符合正态分布。本次研究中，认知维度（偏度=-0.237，峰度=1.307）、态度维度（偏度=-0.394，峰度=-0.657）与行为维度（偏度=-0.166，峰度=0.351）的数值均处于可接受阈值范围内，符合正态分布。图1至图3的分位数图显示，数据大致沿正态分布趋势分布，仅尾部存在少量偏差。基于上述图表，可判定本数据集符合正态分布。

由于未满足方差齐性要求，仅可采用皮莱迹统计量（Pillai's Trace）进行分析（Anderson & Walsh, 2013）。显著性检验结果显示，性别的显著性值为0.017[41]。当显著性值介于0.01至0.05之间时，差异具有统计学意义。因此，本次分析表明仅性别可检测到显著差异。另一项被试间效应检验结果显示，在显著性检验中，仅性别与平均认知维度存在显著影响（p<0.001），表明男女群体间存在显著差异。

图4展示了不同年级学生在平均认知、平均态度与平均行为维度上的差异。图4呈现了不同年级大学生在网络个人信息披露方面的认知、态度与行为的均值及差异情况。其中，大一学生的认知、态度与行为均值分别为2.330、2.768与2.776；大二学生的对应均值分别为2.875、3.090与2.901；大三学生的认知、态度与行为均值分别为2.974、3.075与2.898；大四学生的对应均值分别为2.542、2.974与2.925。在认知维度上，大一与大三学生的差异最大；在态度维度上，大一与大二学生的差异最为显著；在行为维度均值上，大一与大四学生的差异最大。认知维度的均值从「不太了解」至「了解」，态度与行为维度的均值均从「不太同意」至「基本同意」。

图5展示了不同专业学生在认知、态度与行为三个维度上的均值及专业差异。整体而言，文科学生的认知、态度与行为均值分别为2.599、2.926与2.806；工科学生的对应均值分别为2.442、2.800与2.895；理科学生的认知、态度与行为均值分别为3.000、3.204与2.904。在认知维度上，工科与理科学生的差异最大；在态度维度上，工科与理科学生的差异最为显著；在行为维度上，理科与工科学生的差异最大。认知维度的均值从「不太了解」至「了解」，态度与行为维度的均值均从「不太同意」至「基本同意」。

图6展示了不同性别学生在认知、态度与行为维度上的均值及性别差异。整体而言，男生的认知、态度与行为均值分别为2.922、2.958与2.968；女生的对应均值分别为2.298、2.858与2.719。认知维度的均值从「不太了解」至「了解」，态度与行为维度的均值均从「不太同意」至「基本同意」，且女生在各维度的均值均低于男生。