4.3. Partisan Slant of News Media: Network Correlation Analysis Results To answer RQ2, this research conducted QAP correlation analysis, applying Facebook posts from two political party actors, to measure the partisan slant of news media. In addition, this study used posts from environmental groups as the non-partisan parameter, and measure the correlation between news media as a reference. All the results showed significant values (p<.001). Table 7 depicted the analysis results.
Table 7
QAP Correlation Analysis between News Media Networks and Parameter Networks
News Media Network Facebook Parameter Network Partisan Slant Attribution
Note: all of the correlation results were significant (p<0.001)
Generally, the QAP correlation of each news organization indicated the higher scores were correlated to the environment network, compared to the scores correlated to DPP or KMT
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
actors. It suggested that the semantic networks of all the news media were relatively similar to the networks composed by posts from environmental group, compared to DPP and KMT networks. The results suggested that in this case of Shenao power plant, what highlights in news media reports were similar with what environmental groups discussed about, rather than political actors’ concerns. That is, generally news media set the agenda closer to the environmentally-related perspectives, instead of political disputes.
To answer the RQ2a, this research conducted the correlation analysis and compared the correlation score of the two types of news media with KMT and DPP networks separately.
The type of news media would be marked on the partisanship leaning as it got higher correlation scores with either KMT or DPP network. The results indicated that traditional news media (TN-ALL) got higher correlation analysis score to KMT (.660) actors, rather than to DPP (.659) actors. Similarly, digital online news media (DN-ALL) also got higher correlation analysis score to KMT (.682) than to DPP (.663) actors. That is, the semantic network of traditional news media (TN-ALL) and digital online news media (DN-ALL) were more correlated to KMT’s semantic network. Generally, the semantic structure of traditional news media and digital native news media were more similar to KMT political actors. To delve into more comparison between two types of news media, the correlation score of digital native news media to KMT (.682) was higher than the score of traditional news media to KMT (.660). Also, the result score of digital native news media indicated more correlated relationship on DPP (.663) than the relationship of traditional news media to DPP (.659).
These result suggested that the semantic network of digital native media was more similar to the partisan networks on both KMT and DPP, compared to the one of traditional news media.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
To answer RQ2b, this research conducted the QAP analysis on each news media brand selected in traditional news media or digital native news media. As the result suggested, among six traditional news media brands, one news media brand revealed higher correlation score to DPP parameter network, and other five traditional news media brands got higher score to KMT network. It suggested that the semantic structure of most of the traditional news media brands were more similar to KMT actors than DPP actors in Shenao power plant issue. Only Sanlih E-Television online (SETN) got higher correlation score on DPP (.603) than on KMT (.600) network, which was the only traditional media brand marked as pro-DPP upon this issue. On the other hand, five other news media brands, Apple Daily online (APPL), ETTV Financial news online (ETTV), Liberty Times online (LIB), TVBS News online (TVBS), United Daily News online (UDN), of which the semantic networks were more correlated to KMT parameter, compared to DPP actors. These five news media brands would be marked as pro-KMT in Shenao’s issue.
As for the digital native news media, one news media brands revealed higher correlation score to DPP parameter network, and other three traditional news media brands got higher score to KMT network. Similar to the case of traditional news media, it indicated the more similar semantic structure of most of the digital news media brands to KMT actors than DPP actors in Shenao power plant issue. Only Storm Media (STOM) got higher correlation score on DPP (.649) than on KMT (.648) network, which was the only digital native media brand marked as pro-DPP upon this issue. On the other hand, five other news media brands, Apple Daily online (APPL), ETTV Financial news online (ETTV), Liberty Times online (LIB), TVBS News online (TVBS), United Daily News online (UDN), of which the semantic networks were
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
more correlated to KMT parameter, compared to DPP parameter. These three news media brands would be marked as pro-KMT in Shenao’s issue.
Totally, two news media were marked pro-DPP, as the other eight news media were KMT leaning, labeled as pro-KMT. This result would bring out the classification of news media in RQ3, which would analyze the salient concepts of Shenao power plant issue shown in labelled pro-KMT and pro-DPP news media.
As for the partisan slant, the results of RQ2a and RQ2b suggested that most news media were more correlated to KMT political actors. It might be due to the KMT’s political position as an opposition party. As the previous paragraph suggested, contents of news media were more relevant to what environmental groups focused, and also the issue-based concerns. On top of this, based on the issue-based or environmental information, the news media might cover more on what KMT actors focused, such as the referendum, countersign, slogans of
“anti-Shenao power plant”, and so on, which led to similar semantic networks between news media and KMT actors.
Under this condition, some news media (e.g., Liberty Times online (LIB) or Apple Daily online (APPL)) that were commonly considered as pro-DPP among Taiwanese readers surprisingly revealed their slant towards KMT. From the trend analysis of KMT and DPP in the previous section, the posts from KMT actors hugely outnumbered the posts from DPP actors. The Shenao referendum in the 2018 election was considered a prelude for the presidential election in 2020, the incumbent party (DPP) was facing tough challenges from the opposition party (KMT) (Aspinwall, 2018). Additionally, KMT politicians initiated the anti-Shenao power plant referendum which built the consensus from the public. Under this
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
context, KMT actors dominated the discussion of Shenao power plant issue, compared to DPP actors. In addition, the Storm Media (STOM), which was commonly considered as pro-KMT among Taiwanese readers, revealed its slant towards DPP. In fact, the result suggested smaller differences between the value of correlation to be pro-KMT or pro-DPP. It might be due to the higher rate of review articles (13.3%) that might influence the constructions of word and matrix, and affect result of QAP analysis.
4.4. Semantic Network Analysis: Pro-DPP/KMT News Media in Shenao’s