Reaching out: Exploring the use of a competitor’s hashtag on Twitter

By Marissa P. Liponhay and Christian M. Alis

(This analysis was first published in Rappler.com)

Hashtags are a way to label a tweet which makes it more easily discoverable by users interested to read tweets related to the hashtag. This discoverability is further enhanced when a surge in a hashtag’s usage pushes it into the list of trending topics, which greatly multiplies a tweet’s potential audience.

Not surprisingly, groups and individuals use hashtags to promote their cause, especially for the upcoming Philippine elections. Indeed, some hashtags mirror the competition in the election itself. Take for example #LeniDuwag. We could surmise that this hashtag labels tweets against presidential candidate Leni Robredo and, for simplicity, benefits her main rival presidential candidate Bongbong Marcos. This hashtag, however, is not isolated and seems to be a jibe against the pre-existing and longer-lived hashtag #MarcosDuwag, which can be surmised is against Marcos and in favor of Robredo. The hashtags #LeniDuwag and #MarcosDuwag can therefore be considered as a pair of competing hashtags.

In this work for #FactsFirstPH, we investigate public tweets that contain a hashtag from a pair of

competing hashtags. These paired hashtags are #LeniDuwag and #MarcosDuwag, #LetLeniLeave and #LetLeniLead, and #SolidNorth and #SolidSnort. For each hashtag, we identify by looking at the hashtag itself whether it is Pro-Marcos or Pro-Robredo. We then assign a user to the Pro-Marcos group if all their tweets in the dataset contain only Pro-Marcos hashtags. Similarly, a user is assigned to the Pro-Robredo group if all their tweets in the dataset contain only Pro-Robredo hashtags. If a user has a tweet that has a Pro-Marcos hashtag but also has a tweet that has a Pro-Robredo hashtag, then that user is assigned to the Mixed group. Table 1 shows the groupings of hashtags. Note that a tweet may contain several hashtags so it could have both Pro-Marcos and Pro-Robredo hashtags.

Table 1. Assignment of hashtag pairs into groups

Groups	Hashtags
Pro-Robrero	#MarcosDuwag, #LetLeniLead, #SolidSnort
Pro-Marcos	#LeniDuwag, #LetLeniLeave, #SolidNorth
Mixed	Pro-Robredo + Pro-Marcos

Data Analysis

We were able to gather 386,814 public tweets from January 15 to March 16, 2022, and identified 79.29% of the users in the Pro-Robredo group, 11.76% are in the Pro-Marcos group, and 8.95% are in the Mixed group, as shown in Figure 1. This implies that among the users in this dataset, most do not use a hashtag of the “other” side.

Figure 1. Number of users in different groups.

Despite the lowest number of authors or users in the Mixed group, the daily volume of tweets from this group is comparable to the volume of tweets coming from the other two groups as shown in Figure 2. The users in the Mixed group have more tweets on average (18.3 tweets/user) compared to the two other groups (3.9 tweets/user for Pro-Marcos, 2.4 tweets/user for

Pro-Robredo). It is also noticeable that the volume and lifetimes of tweets with hashtags that are pro-Marcos are smaller and shorter, respectively.

Figure 2. Volume of tweets per day.

In Figure 3, we further separate the tweets into specific hashtags to show the ratio contained in each group. It is evident that #LetLeniLead dominates the hashtags in the Pro-Robredo group while #LeniDuwag dominates the hashtags in the Pro-Marcos group in terms of volume.

Figure 3. Total volume per hashtag.

In Table 2, we present the different parameters that describe the users belonging to each group. We investigate a common speculation that most of the Twitter accounts involved were only created recently to boost the engagement of partisan tweets by exploring the age of accounts, in terms of the number of years since account creation. We find that the average age of the accounts of users that are in Pro-Robredo is 6 years (2016) while it is 4 years for the Pro-Marcos group (2018). Although the average age of the accounts for both groups is less than 10 years, the oldest accounts were created in 2006 and 2007 for Pro-Robredo and Pro-Marcos groups, respectively. These findings are still inconclusive in terms of disproving the speculation and thus, require a more in-depth study. We also looked at the number of influencers or users with more than 100,000 followers and find that the Pro-Robredo group has about 12,000 influencers while the Pro-Marcos group has only 104 influencers.

Table 2.

	Pro-Robredo	Pro-Marcos
Average Age of Account	6 years	4 years
Influencers (>100000 followers)	12220/210003 users	104/43938 users
Unique referenced tweets as sources for viral posts	11/17910	1/1018

Engagement

We also look at the level of engagement of the tweets in each group and find that the supporters of Marcos are most actively engaged in terms of the number of replies and likes, while the supporters of Robredo are most actively engaged in terms of the numbers of retweets and quotes as shown in Table 3. These numbers are visualized in Figure 4.

Table 3. Metrics of Engagements in Twitter

Figure 4. Visualizing the level of engagements of Twitter users per group.

Referenced Tweets

While it has been observed that most users only use hashtags aligned toward a candidate, the existence of a Mixed group where they tweeted hashtags from opposing groups is intriguing. Whether intentional or not, the usage of two opposing hashtags is an indication of an attempt to be seen by the members of the other group. To investigate this further, we look into the tweets that are being retweeted or quoted by users in each group. We refer to these as reference tweets and consider them as the users’ source of information. Retweeted tweets are tweets that are shared without adding anything, while quoted tweets are tweets that are shared but with additional comments from the user.

We find that 8,559 tweets were shared by both members of the Pro-Robredo and Mixed groups but not by members of the Pro-Marcos group. On the other hand, 4,125 referenced tweets were shared by both Pro-Marcos and Mixed group members (see Figure 5). These numbers are equivalent to 23.55% of the total referenced tweets in the Pro-Robredo group being mixed with hashtags in favor of Marcos, while 32.05% of the total reference tweets in the Pro-Marcos group are being mixed with hashtags in favor of Robredo.

Figure 5. Venn Diagram showing the share of referenced tweets of the two groups with the Mixed group.

Information Source of Viral Posts

For further analysis, we analyzed the referenced tweets of viral posts defined as tweets with more than 4,000 retweets. In the Pro-Leni group, there are only eleven (11) referenced tweets of the viral posts, while in the Pro–Marcos group, only one (1) referenced tweet (see Table 1) serves as the source of information on viral posts. These findings imply that viral posts point to very few unique referenced tweets as their source of information.

We explore the purpose of mixing positive hashtags of competing political sides by tracing the referenced tweets in the Mixed group. We find that the most referenced tweets common to Pro-Robredo and Mixed groups are all advocating for Robredo and nothing negative about Marcos. This suggests that Robredo’s supporters seem to attempt to spread the positive aspects about her to the supporters of Marcos by using hashtags in favor of Marcos. Sample screenshots of top references are shown below. The account names are intentionally removed for privacy.

Figure 6. Screenshots of top referenced tweets of viral posts for Robredo that are being mixed with hashtags that are in favor of Marcos.

On the other hand, the most referenced tweet with a hashtag in favor of Marcos that is sometimes mixed with hashtags in favor of Robredo is using a hashtag against Robredo (#LeniDuwag) but whose content is in favor of Robredo. The following is the content of the tweet:

‘mfs really using #LENIDUWAG when their candidate backed out on all major presidential debates 😴‘

Other highly referenced tweets with hashtags in favor of Marcos that are in the Mixed group are found to be advocating for Robredo or Leody De Guzman, another presidential candidate, instead. Screenshots of a few sources are shown below where the account names are intentionally removed for privacy.

Figure 7. Screenshots of a few most referenced tweets of viral posts for Marcos that are mixed with hashtags that are in favor of Robredo.

Conclusions

In this work, we explored the mixing of competing hashtags on Twitter. Our main finding shows that supporters of both camps are actively mixing hashtags.  Interestingly viral posts in the Pro-Robredo group have only 11 reference tweets while viral posts in Pro-Marcos have only 1 reference tweet being retweeted. Among these sources, 9 reference tweets in favor of Robredo are being mixed with the hashtags in favor of Marcos, and all of these advocate positively for Robredo. On the other hand, the top one reference tweet with a hashtag in favor of Marcos that is being mixed with hashtags for Robredo has content that is expressing support for Robredo, instead. The positive contents of most common reference tweets for Robredo possibly imply a strategy to reach out to the exclusive ‘echo chambers’ of Marcos supporters.