本研究以选择螺旋理论为指导，以今日头条27个月中43,662条贴文作为数据，以城市品牌作为选择因素，以粤港澳大湾区11个城市名作为选项，以发布、点读、转发作为用户行为，检测分析了大湾区城市的品牌效应和发、读、转三个群体的不同偏好及其相互协调。本研究发现，发布行为的选择度最强(s_l=0.47)，点读次之(s_l=0.255)，转发最弱 (s_l=0.04)；与此对应，对选择螺旋首轮之末转发内容的分布，发布者影响最大 (c_t=62.74%) ，点读者次之(c_t=33.92%)，转发者影响最小 (c_t=3.34%)。这不是“议程设置”，而是“议程协调”；螺旋首轮的协调被发布群体主导 ； 但“主导”不等于“设置”更不是“垄断”；读者群体超过三分之一的影响亟应重视，转发群体超过3%的影响也非子虚乌有。更重要的是，发布是一次性行为，发布完成，意味着发布群体停止参与；首轮之后，信息螺旋由发、读、转三国斗（trio）演化为读转二人转（duet）。所以，两群体在首轮以后各轮中的预测影响上升为91.04% 和8.96%。

        首轮中，三群体关于城市品牌的具体选择也各不相同。 发布者最强调广州、深圳，其次香港；点读者最关注香港、澳门，其次东莞、中山；而转发者的选择近于平均，略偏向深圳、中山、佛山。若发布者的目标是高点读，其选择及其结果与目标相距甚远。

        同时，本研究也对不同身份的发布者所发布的帖文、不同时段发布的帖文进行了分析，结果与总数据相近。本研究还进行了共词分析。

        Importing and implementing elements of agenda setting, negotiated reality and selective spiral, this study adapted, advanced, and applied the theories of negotiated agenda and negotiated spiral. Under the integrated theories, city brand (city name) was chosen as a selecting factor; 11 cities in the Guangdong Hong Kong Macau Greater Bay Area (GBA) were chosen as selections; release, read and relay were chosen as user reactions; and 43,662 postings in 27 months from Toutiao.com were crawled as the main data. A main purpose of the study was to dissect and depict the selective behaviors of the users, including releasers, receivers (readers), and relayers, and the negotiations between the groups regarding the brand names.

        It was discovered that, of the three groups, releasers were the most selective (s_l=0.47 on 0-1 scale) and contributed the largest share (c_t=62.74%) to the distribution of the content relayed at the end of the first round of the selective spiral. The readers were the second most selective (s_l=0.255) and contributed the second largest share (c_t=33.92%). While the relayers were minimally selective (s_l=0.04), they still contributed (c_t=3.34%). The findings suggest that, where online communication dominates, some classical dichotomies may be giving way to multiplicities and continuums. The dichotomy of the media versus the audience, for example, may expand to the multiplicity of users, including releasers, receivers, relayers, respondents, reactors, etc, in addition to platforms such as Toutiao. Strong versus weak effects of the media may evolve into contributions, ranging on continuums of 0~100%, by various groups. The question about agenda setting by the media versus the audience might be more informatively addressed by analyzing agenda negotiations between users playing various roles in selective spirals.

        The data was also analyzed in terms of identities of releasers, and time periods. The results were consistent with our analysis of the whole dataset. This study also conducted a co-word analysis.