國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
61
policy, the stricter profile privacy settings. Since the results showed positive association between the two variables, H5a was supported. Knowledge accounted for 2.7% of the variance.
H6a proposed that the more negative experiences users had on Facebook, the stricter their profile privacy settings. Although there was a strong association between the two variables, the results showed that users’ negative experiences on Facebook were negatively associated with their strictness of privacy settings on profiles. Therefore H6a was rejected.
Meanwhile, negative experiences accounted for 0.5% of the variance.
Lastly, H7a proposed that the more privacy concerns users had on Facebook, the stricter their profile privacy settings. However, the results showed that there was no significant association between these two variables. Therefore H7a was rejected. Altogether, the model composed of users’ demographics information, years of using Facebook and average hours per day spent on Facebook, value disposition, personal privacy orientation, understanding toward the privacy policy on Facebook, negative experiences and privacy concerns on Facebook accounted for 16.5% of variance in predicting users’ profile privacy settings on Facebook.
5.3 Predicting users’ use of privacy protection strategies on Facebook
In terms of predicting users’ privacy protection strategies on Facebook, results showed that gender (ß=-0.073*), age (ß=-0.069*) and users’ knowledge toward Facebook’s privacy policy (ß=-0.110***) negatively predicted users’ use of privacy protection strategies.
Meanwhile, users’ negative experiences (ß=0.259***) and privacy concerns (ß=0.199***) positively predicted users’ use of privacy protection strategies on Facebook.
Firstly, H1b proposed that female Taiwanese users would use privacy protection strategies more frequently. However, since the results showed a negative association between the two variables, it was actually the males that would use privacy protection strategies more
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
62
frequently than females. Therefore H1b was rejected.
H2b proposed that the younger the users were, the more frequently they would use privacy protection strategies. According to the results, there was indeed a negative association between the variables, and it was significant as well. Therefore H2b was supported.
H3b proposed that the more time the users spent on Facebook, the more frequently they would use privacy protection strategies. However, the results indicated no significant association for neither user’s years of using Facebook, nor the average hours per day for using Facebook. Therefore H3b was rejected.
H4b proposed that the more privacy oriented the users were, the more frequently they would use privacy protection strategies on Facebook. Once again, the results indicated no significant association between the two variables. Therefore H4b was rejected as well.
H5b proposed that the more knowledgeable the users were toward Facebook’s privacy policy, the more frequently they would use privacy protection strategies. However, the results suggested the opposite. According to the results, there was a negative association between the two variables, meaning that the less understanding toward Facebook’s privacy policy, the more frequently users would use privacy protection strategies. Therefore H5b was rejected.
Meanwhile, users’ understanding toward privacy policy on Facebook accounted for 1.3% of the variance.
H6b proposed that the more negative experiences users had on Facebook, the more frequently they would use privacy protection strategies. According to the results, there was a positive and significant association between the two variables. Therefore H6b was supported.
In addition, negative experiences also accounted for 6.7% of variance.
Last but not least, H7b proposed that the more privacy concerns users had on Facebook, the more frequently they would use privacy protection strategies. According to the results, there was also a positive and significant association between the two variables. Therefore H7b was supported, while privacy concerns accounted for 3.1% of the variance. To briefly
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
63
sum up, the model consisted of demographic information, years of using Facebook and average hours per day spent on Facebook, personal privacy orientation, knowledge toward Facebook’s privacy policy, negative experiences and privacy concerns accounted for a total of 13.1% of variance in predicting users’ privacy protection strategies.
According to the results, users’ gender, understanding toward Facebook’s privacy policy and negative experiences on Facebook were significant predictors to both users’ privacy settings on profiles and their privacy protection strategies. On the other hand, users’ years of using Facebook and average hours per day spent on Facebook only served as significant predictors to the strictness of users’ privacy settings on profiles, but not the frequency of use in users’ privacy protection strategies. Also, users’ age and privacy concerns were significant predictors toward users’ privacy protection strategies, but not to users’ privacy settings.
(Table 5.11)
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
64
Table 5.11 Hierarchical linear regression on Facebook users’ privacy settings on profiles and privacy protection strategies in Taiwan (Final standardized regression coefficients)
Privacy settings Privacy protection strategies Block 1: Demographics
Gender 0.105*** -0.073*
Age 0.006 -0.069*
Educational level 0.017 0.008
Employment status -0.033 0.008
Incremental R2 (percent) 1.5** 0.8
Block 2: Time spent on Facebook
Years of using Facebook -0.066* 0.050
Hours of using Facebook -0.206*** 0.048
Incremental R2 (percent) 5.0*** 0.9*
Block 3: Value disposition
Personal privacy orientation 0.233*** -0.014
Incremental R2 (percent) 6.6*** 0.3
Block 4: Understanding toward privacy policy on Facebook
Knowledge 0.163*** -0.110***
Incremental R2 (percent) 2.7*** 1.3***
Block 5: Negative experience
Previous negative experiences on Facebook -0.074* 0.259***
Incremental R2 (percent) 0.5* 6.7***
Block 6: Privacy concerns
Privacy concerns on Facebook 0.041 0.199***
Incremental R2 (percent) 0.1 3.1***
Total R2 (percent) 16.5 13.1
N=1102. (p < 0.05; p <0 .01; p <0 .001.)
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
65
CHAPTER SIX: DISCUSSION
There were quite a few notable trends in Facebook users’ privacy management in Taiwan. This thesis included gender, age, years of using Facebook and average hours per day spent on Facebook, value disposition, knowledge toward Facebook’s privacy policy, negative experiences and privacy concerns as independent variables to predict the strictness of users’
privacy settings and the frequency of their privacy protection strategies. In terms of the correlation between independent and dependent variables, gender, hours of using of Facebook, knowledge toward Facebook’s privacy policy, previous negative experiences on Facebook and privacy concerns on Facebook were significantly correlated to both users’ profile privacy settings and privacy protection strategies. Meanwhile, years of using Facebook and personal privacy orientation were only correlated to users’ profile privacy settings. However, correlation only meant the independent and dependent variables were linearly related.
Therefore this section would focus on the discussion of major findings, results of predictions and their implications.
Gender
Firstly, the male and female ratio of respondents in this study was relatively equal in portion. To be more specific, among the respondents, 48.6% were males, while 51.4% were females. Similar to the findings of comScore Media Metrix released by InsightXplorer in May 2013, the gender distribution for the online population of 11.87 million users, male accounted for 51.5% of population while female accounted for 48.5% of the population. In terms of predicting users’ privacy management, H1a predicted that females were more likely to manage their privacy on Facebook through stricter profile settings and was supported in this thesis. The results in this thesis were in line with previous findings by Stutzman &
Kramer-Duffield (2010), where gender was found to have an impact on users’ privacy settings. In addition, Madden (2012) found that female tended to have a more restricted profile than male. Similarly, female Facebook users in Taiwan also tended to have more
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
66
restrictive profile settings since their profiles were more private. On the other hand, H1b predicted that females would also use the privacy protection strategies more frequently as a mean of privacy management. However, results rejected H1b. Instead, it was suggested that males would use privacy protection strategies more frequently than females. The implication regarding gender as a factor to predict users’ privacy management suggested that females and males perhaps had different preferences in their means of doing privacy management.
Age
Furthermore, this study found that the age distribution among the respondents echoed the global trend of Facebook users. In this study, respondents between the age 25-34 occupied 31.9% of all the age groups. It further reinforced the findings of Global Wed Index, in which the greatest population of users came from the age group of 25-34 (Mander, 2014).
This could be due to the fact that the majority of the users had been using Facebook since it first came out, hence the increasing in age of users. Furthermore, H2a predicted that the younger the users were, the more likely they would have stricter privacy settings. However, the results showed that age had no association with users’ privacy settings. Therefore H2a was rejected. This finding thus suggested a different story than Madden (2012), in which the research found that younger users in the States were more likely to have a private profile. As a result, it may suggest that the use of privacy setting was an act regardless of the users’ age.
On the other hand, in terms of predicting users’ frequency of using privacy protection strategies, H2b suggested that the younger the users were, the more frequently they would use privacy protection strategies. This was confirmed by this thesis. This finding was in fact more in line with what Madden & Smith (2010) suggested, that profile “pruning,” namely deleting, un-tagging and removing content was on the rise for the younger audience. In addition, the younger users were also the ones that were more likely to “un-friend,” hereby deleting someone from their friend lists as a part of their reputation management. This implied that younger users were more prone to using privacy protection strategies aside from their
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
67
pre-existing privacy settings.
Years of using Facebook and average hours per day spent on Facebook
First of all, the respondents had spent an average of 4.71 years on Facebook at the time of the study, which meant they became Facebook members after the traditional Chinese interface was introduced. It also coincided with the fact that Facebook gained its tremendous popularity around 2010 from the famous game “Farmville,” or commonly known as “Happy Farm” in Taiwan (Penny, 2010). In addition, respondents reported that they spent an average of 3.06 hours on Facebook per day.
Aside from gender and age, the amount of time users spent on Facebook may also contribute to the data they provided to the social networking site (Debatin et al., 2009;
Madden & Smith, 2010). H3a thus predicted that the more time users spent on Facebook, the stricter their privacy settings on profiles. However, H3a was rejected since the results showed that in fact, the less time users had spent on Facebook, the stricter the profile settings.
Meanwhile, H3b predicted that the more time users’ spent on Facebook, the more frequently they would use privacy protection strategies. However, the results depicted that there was no significant association between users’ time on Facebook and their use of privacy protection strategies. As a result, H3b was rejected, and it was unclear why there was such an association between users’ time on Facebook and their privacy management.
Value disposition
In terms of Facebook users’ privacy orientation in Taiwan, they were predominately privacy-oriented. The respondents found their privacy on Facebook important, disliked sharing personal information with people other than their close friends, found it rude when their acquaintances inquired them regarding their personal matters, and did not usually share personal information with my acquaintances. This finding further reinforced the ground blocks of CPM, where individuals had the right and ownership to their private information (Petronio, 2002). H4a predicted that the more privacy oriented users were, the stricter the
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
68
privacy settings. As a result, H4a was supported. This confirmed the findings of Hesse &
Rauscher (2013), in which individuals’ personal privacy orientation also played a role in determining their privacy tendencies in an online environment. On the other hand, H4b predicted that the more privacy oriented the users were, the more frequently they would use privacy protection strategies. However, H4b was rejected due to the lack of significant association between the two variables. Therefore, it was implied that privacy protection strategies were not related to users’ personal privacy orientation.
Understanding toward privacy policy on Facebook
As Debatin et al. (2009) revealed in their study, while the majority of Facebook users reported having an knowledge toward privacy settings to make use of their privacy settings, users may have a skewed sense of what that exactly entailed. The findings also echoed this proposition. As it turned out, only a little bit over one-third of respondents had actually read the privacy policy of Facebook. Unsurprisingly, the knowledge-type of questions also revealed that users in Taiwan had partial knowledge toward Facebook’s privacy policy. In more concrete terms, over 90% of users were not aware of all the possible options they had with the posts they generated on Facebook. As of now, Facebook already allowed users to select their intended audience, in which the combination could vary from all the Facebook friends, acquaintances, network to a specific list of audience. However, the knowledge-type of question revealed that the majority of the users were not aware of the extent of these privacy settings. On the other hand, results showed that majority of the respondents were aware of the differences between hiding a post, and deleting it from their timelines. It depicted that users knew the effect of these privacy protection strategies.
Ou (2011) also stressed that an individual’s knowledge toward the current privacy policy was crucial to determine how it may affect users’ decisions in the privacy settings. While H5a predicted that the more knowledge users had regarding the privacy policy on Facebook, the stricter the profile settings, it was supported. This finding was in line with Butler et al. (2011),
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
69
in which the researchers found that it required adequate knowledge regarding Facebook’s privacy policy in order to know how to manage one’s privacy accordingly. However, while H5b predicted that the more knowledgeable users’ were, the more frequently users would use privacy protection strategies, it was rejected due to the significantly negative association between the variables. As a result, it implied that the less understanding users had toward Facebook’s privacy policy, the more frequently they would use privacy protection strategies.
It could be that users’ were unaware of the possible privacy settings for them, so they had to use additional privacy protection strategies to mitigate their privacy concerns.
Negative experiences
As for negative experiences users had on Facebook, over three quarters of them reported the negative experience of being invited into the spammed shopping groups. The percentage was twice as much to the second most common negative experience, indicating the prevalence of those spammed shopping groups. For the second and third most common negative experiences, a little more than one third of them reported having their Facebook accounts taken over, and being targeted by Facebook and other advertisers with ads. On the other hand, for negative experiences related to the content they generated being misused, or the interpersonal, financial, physically and reputational damages were drastically less common among the respondents. However, one reason to this could be that respondents were reluctant to report those negative experiences since its private nature, and thus they were leaning towards giving socially desirable answers and would not indicate having those troubles.
As Petronio (2002) has suggested, in the face of boundary turbulence, such as negative experiences on Facebook, an individual would try to restore to its original boundary. As a result, H6a predicted that the more negative experiences users had, the more likely they would have a stricter profile setting. However, according to the results, H6a was rejected due to a negative association between the two variables. It meant in fact, the fewer negative
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
70
experiences users had, the more likely they would have a stricter privacy settings on profiles.
On the other hand, H6b predicted that the more negative experiences the users had, the more frequently they would use the privacy protection strategies. As a result, H6b was supported.
The implication of the findings suggested that in the face of a privacy turbulence, users would to privacy protection strategies to mitigate the boundary.
Privacy concerns
In terms of privacy concerns, the concern levels were very similar regarding how Facebook made use of their information, how the content about them or generated by them were being used, and their log-in details such as location and timestamp. What this suggested was that users were becoming predominantly aware of the potential privacy breaches on Facebook, similar to the findings by University of Vienna on Facebook users in Europe (Woollaston, 2013). In addition, this study revealed that the concerns regarding how Facebook made use of users’ information were among the top concern for respondents.
Similarly, according to the report done by Taiwan Network Information Center (TWNIC) on broadband Internet usage in Taiwan, the study conducted between April 3 - May 21, 2013 showed that 73.78% of users were concerned about the leak of their personal information when using the Internet (TWNIC, 2013). It showed how users were very concerned with how their information was being handled.
As for predicting users’ privacy settings, H7a predicted that the more privacy concerns users had, the stricter the privacy settings. As it turned out, H7a was rejected due to the lack of significant association between the variables. Meanwhile, H7b predicted that the more privacy concerns the user had, the more frequently they would use privacy protection strategies. As a result, H7b was supported. This finding suggested a similar implication that in the face of a privacy turbulence such as users’ privacy concerns, the users would turn to privacy protection strategies to mitigate the concerns.
Privacy settings on profiles
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
71
In addition, this thesis found that respondents would utilized different privacy settings based on the nature of the information on their profiles. For instance, educational information, work information and placed they had lived before, as well as their interests were considered as less sensitive. Therefore, over half of the users set them to be visible by the public, or by their Facebook friends. In addition, photos, videos and tags of the users were also set to be visible by the public and their friends. This reinforced the findings from previous research.
Previous studies have found that similar to the behavior of bloggers, users of social networking sites mostly used it for keeping up with trends, storing information, and sharing information (Lee et al., 2008; Waters & Ackerman, 2011).
On the other hand, for the more sensitive information on their profiles, users were less likely to set it to be visible by the public or Facebook friends. For instance, unlike the other user generated content such as photos, videos and tags, notes were considered as more sensitive and private. Therefore less than half of the respondents set it to be visible by the public and friends. Less than half of the respondents set their blood type, relationships status and with whom they were in a relationship with to be visible by their friends as well. In addition, less than one third of them set their religious views and political views. Without a doubt, users’ mobile phones and addresses were very sensitive information. They were hardly
On the other hand, for the more sensitive information on their profiles, users were less likely to set it to be visible by the public or Facebook friends. For instance, unlike the other user generated content such as photos, videos and tags, notes were considered as more sensitive and private. Therefore less than half of the respondents set it to be visible by the public and friends. Less than half of the respondents set their blood type, relationships status and with whom they were in a relationship with to be visible by their friends as well. In addition, less than one third of them set their religious views and political views. Without a doubt, users’ mobile phones and addresses were very sensitive information. They were hardly