CAPÍTULO 5. “AN APPLICATION OF GAME THEORY: COST ALLOCATION”
5.3 Estudio de un problema de asignación de interés
Introduction
This chapter provides demographic information about the fans and users of the TFWCH page. This chapter is a descriptive analysis of the data supplied by Facebook, accessed through the company Locowise, relating to who was using the page, such as: their age, gender, location, and if they were accessing the page from a mobile or desktop device. It also explores other metrics that give an indication of community engagement with the page, such as how many users were reached and how users were engaging with the content. These data are presented because they give an initial overview of the fans and activity on the page, thus providing context for the upcoming data chapters. Thus this chapter is more descriptive than analytical.
Method
Research design
This analysis was part of a retrospective case study.
Participants
The participants in this study are the individuals who used the TFWCH Facebook page during January 2013 to January 2014, and whose usage of the page could therefore be analysed via aggregated engagement statistics, 41 accessed via Locowise, which is described below.
41 Facebook Insights also provides information about how a page is performing, such as how people are
discovering a page and how they are responding to posts. Facebook’s Page Insights are available to the Administrators of pages. Locowise was selected as it delivers more comprehensive metrics in a more user- friendly platform.
Materials
No new data had to be gathered as such; the data were simply downloaded and organised. Locowise was used to download the statistical data that had been created during the life of the TFWCH Facebook page.
Locowise is a commercial provider of analytics tools and social media management for a range of platforms such as Facebook, Twitter and Instagram. It is a relatively new company based in the United Kingdom, and it is a useful software tool for assessing and accessing Facebook analytics. Locowise software enables the user to download all previous posts and comments from a social networking site source into a database, and then examine key metrics and demographic data.
Procedure
Locowise was informed about the current research project, and on request, downloaded the TFWCH Facebook page archives.42 After initial technical
difficulties with accessing data from the first three days of the TFWCH Facebook page’s life, the full data set from the first year of TFWCH was made available. In addition to exploring Facebook’s definitions and utility of the metrics available, Thinktank Social43 and Locowise advised on the most meaningful statistics worthy of investigation. Only the most relevant metrics are explored.44
42 Historical archives are not usually available for commercial licensees of the product. While Facebook data is
publically available, being able to utilise Locowise rather than the less sophisticated Facebook Insights function was preferable. The authors are grateful to Locowise for their support.
43 Thinktank Social is an Australian social media marketing and analysis agency, www.thinktanksocial.com.au 44 Some metrics were excluded from this analysis that would be useful for other purposes. For example, many
people came to TFWCH from within Facebook – a metric known as ‘page like sources’. These are not included here in this thesis, but are defined in Table 9 in Appendix C; and Table 10 lists the results for this metric. This metric is useful for analysing which parts of Facebook most successfully draw people to a page. Emergency management and community groups could capitalise on this information in order to have a greater reach.
Data analysis
Disasters have distinct timelines. There is usually an initial flurry of activity by formal and informal responders that occurs as the response takes place, and then activity slows as the recovery continues. This timeline can be tracked by examining social media archives. Thus the statistics of the TFWCH page were analysed in separate blocks of time, presented in Table 5.1 below. January was the busiest month and it made sense to separate it into four time blocks of one week apiece.
Table 4.1.1. TFWCH: The nine key time periods under investigation
Time period Description
Week 1 January 4 – January 10,2013
Week 2 January 11 – January 17,2013
Week 3 January 18 – January 24, 2013
Week 4 January 25 – January 31, 2013
First Quarter (1st Qtr.) January, February, March 2013 Second Quarter (2nd Qtr.) April, May, June 2013
Third Quarter (3rd Qtr.) July, August, September 2013 Fourth Quarter (4th Qtr.) October, November, December 2013
Year 1 January 4, 2013 – January 3, 2014
A detailed description of the basic Facebook terms appears in the Appendix (Table 1). Table 5.2 below describes the different metrics that are explored in this chapter, as well as categorising how each statistic was calculated.45
45 Although it is not critical, it is useful to briefly reflect on how Locowise calculates each of the metrics, in order
to be aware of the types of statistical analyses that can and cannot be completed with this dataset. A statistic such as “daily page stories” is an average of the number of stories shared on all the days in that particular time period. A statistic such as “sex of fans sharing stories” is an accrual or accumulation of the number of females who have shared stories, and the number of males who have shared stories, in that particular time period. A statistic such as “total page likes” is simply a measure of the number ‘likes’ on the final day of the time period under
Table 4.1.2. Facebook term glossary: 22 key definitions
Variable Description Statistic calculation
Total page likes Number of fans currently following page (people who have clicked ‘like’ on the page) Final Ratio of gained page likes to lost
page likes
Ratio of new fans to the page compared to fans who have been lost Average
Average daily gain of page likes Average number of new fans Average
Average daily loss of page likes Average number of lost fans (people who have clicked ‘unlike’ on the page) Average
Total number of page unlikes Total number of fans who have left the page Accrual
Page like sources Trail from where fan first clicked on the page: from within Facebook Accrual Referring sources Trail from where fan first clicked on the page: from outside of Facebook Accrual
Sex of fans Sex of people who had liked the page Final
Sex of fans sharing stories Sex of people who were actively sharing stories from the page (commenting, sharing and so on)
Accrual
Age group of fans Age of fans Final
Most common age group sharing stories
Most common age group of fans actively sharing stories from the page (commenting, sharing, and so on)
Table 5.2. Facebook term glossary: 22 key definitions (continued)
Variable Description Statistic calculation
Country of fans Where fans originated from Final
Total page impressions The total number of impressions (or images) ‘seen’ of any content associated with the TFWCH Facebook page. This represents the number of times any content at all relating to the TFWCH page appeared on somebody’s newsfeed or ticker
Accrual
Mean daily page impressions The same metric as total page impressions but is a daily average rather than a cumulative total
Average Total number of people who had
viewed any content
The total number of people who have seen any content associated with the page (that is, any impressions). Therefore this statistic informs of how many individual, unique people had ‘seen’ something from the TFWCH page – remembering impressions do not
guarantee that the person has definitely viewed the post
Accrual
Mean number of people seeing any content daily
The same metric as total number of people who had viewed content, except that it is the daily average rather than the cumulative total
Average Total page engagements The total number of times people engaged with the TFWCH page by clicking on any of
the content, but without generating a story
Accrual Mean daily page engagements This metric is the same as total page engagements described above, but it represents the
average number of engagements per day rather than a cumulative total
Table 5.2. Facebook term glossary: 22 key definitions (continued)
Variable Description Statistic calculation
Mean number of people engaging daily
The daily average number of people who engaged with the page by clicking on any of the content, but without generating a story. This statistic is similar to mean daily page
engagements just above, but it informs about how many different, unique people have clicked on content to do with the page
Average
Total page stories The total number of stories created about the page. A story is generated when one of the following happens: a person likes a post, shares a post, comments on a post, likes the page itself, mentions the page within their own posts, or tags the page in a photo
Accrual
Mean daily page stories The average daily number of stories (likes, shares, comments, and so on.) created about the TFWCH page. Therefore this metric is the same as total page stories, but presents the daily average number of stories generated, rather than the overall, cumulative total
Average
Mean daily page stories by fans of the page
The average daily number of stories being shared by fans of the page. A fan is someone who has liked the page and is following it
Results
The results of this chapter are descriptive, which in and of itself is a valuable contribution to the literature, as little is known about the characteristics and reach of online emergent groups. Why these results are pertinent to this thesis and therefore included is explored briefly in the Discussion section of this chapter. The results are then considered in Chapter 9 in more detail in light of the four research questions.
Page likes
The following table, Table 5.3, presents statistics relating to the number of fans being attracted to and lost from TFWCH over the nine time periods under investigation.
Table 4.1.3. Locowise insights: TFWCH fan gains, losses, and initial access to the page
Variable Time Period
Week 1 Week 2 Week 3 Week 4 1st Qtr. 2nd Qtr. 3rd Qtr. 4th Qtr. Year 1
Total page likes 20,860 20,564 20,413 20,317 20,089 19,877 20,311 20,243 20,243
Ratio of gained page likes to lost page likes
5.07:1 1.06:1 0.59:1 0.53:1 3.32:1 0.53:1 3.44:1 1.06:1 3.03:1
Average daily gain of page likes
2,212.67 143.83 31.83 14.50 171.93 2.81 7.17 1.16 43.76
Average daily loss of page likes
436.83 135.33 54.00 27.33 51.73 5.32 2.08 1.09 14.45
Total number of page unlikes 632 812 324 164 4,449 479 190 100 5,218
Initial access to page: Mobile device
8,350 413 84 26 9,013 54 148 20 9,235
Initial access to page: Non- mobile device
The results show that the page attracted a large amount of likes in the first week of operation, and that the first week was the busiest period for new likes, with a ratio of gained to lost likes in excess of 5:1. Over the four quarters of the year, the page likes were relatively stable; although the first few weeks and in particular the first week of the page’s life were the most popular in regards to likes from fans. The busiest period for fans unliking the page was during the first quarter of the page’s
operation. In regards to the initial access to the page, mobile devices were the key medium for accessing the page in the first week of the page’s life. This then reversed by the second week, and remained reversed. The results for the full year one period are skewed due to the first week.
A number of other websites had referred users to the TFWCH Facebook page by providing a link to TFWCH on their own sites. Table 5.4 below lists and briefly describes the most popular different websites that referred people to TFWCH. In Table 5.5, for each time period under investigation, each of these referring sources is then listed and the results are presented. Table 5.5 also includes two other metrics: total number of referrals and total number of referring sources. Total number of referrals refers to the total number of external referrals that had been driven to the page and total number of referring sources refers to the total number of external sources linking to or mentioning the page.
Table 4.1.4. Locowise insights: External websites referring users to TFWCH
Website URL Description Website URL Description
www.themercury.com.au Tasmanian newspaper website www.tfwch.websmistresshq.com.au Temporary TFWCH website
www.google.com.au Search engine – in Australia www.lifeonplanetbaby.com Unknown, page unavailable
www.forum.pixode.net Facebook Forum for Pages application www.fullersbookshop.com.au Tasmanian book shop website
www.google.com Search engine www.sensis.com.au Online buying/selling website
www.news.com.au Online news website www.hightonvet.com.au Vet clinic in Victoria
www.community.ebay.com.au EBay community page www.ask.com Search engine
www.tightlinestasmania.com.au Tasmanian fishing website www.google.es Search engine – in Spain
www.t.co Unknown, page unavailable www.ssl.bing.com Search engine
www.google.co.uk Search engine – in the United Kingdom www.clintonpower.com.au Relationship counselling clinic NSW
www.mumbrella.com.au Media/marketing website www.bookendtrust.com NFP education initiative, TAS
www.au.news.yahoo.com Online news website www.google.com.pk Search engine – in Pakistan
www.bing.com Search engine www.meganaiemma.edublogs.org Personal education blog
www.storify.com Online news/social media service www.google.ca Search engine – in Canada
www.ultra106five.com Tasmanian radio station website www.google.com.my Search engine – in Malaysia
www.australianhorserescue.com Australian horse welfare website www.google.com.br Search engine – in Brazil
www.google.com.ph Search engine – in the Philippines www.google.pt Search engine – in Portugal
www.french.frontierservices.org Unknown, page unavailable www.google.it Search engine – in Italy
www.google.com.sg Search engine – in Singapore www.melanieirons.com Personal website of Admin
www.google.so Search engine – in Somalia www.tassiefireswecanhelp.com TFWCH website
Table 4.1.5. Locowise insights: Top 10 websites referring users to TFWCH
Variable Time period
Week 1 Week 2 Week 3 Week 4 1st Qtr. 2nd Qtr. 3rd Qtr. 4th Qtr. Year 1
Total number of referrals 2,293 594 63 34 3,488 217 279 101 4,085
Total number of referring sources 144 61 28 19 528 120 140 70 858
Referring source www.themercury.com.au 877 337 2 3 1,222 - 7 - 1229 www.google.com.au 528 165 36 16 861 53 130 55 1099 www.forum.pixode.net 246 - - - 246 - - - 246 www.google.com 32 10 2 - 142 19 19 17 197 www.news.com.au 82 - - - 87 - - - 87 www.community.ebay.com.au 63 11 - - 74 - - - 74 www.tightlinestasmania.com.au 70 - - - 71 - - - 71 www.t.co 59 - - - 66 - - - 66 www.google.co.uk - - 1 - 46 4 3 - 53 www.mumbrella.com.au - 11 - - - - www.au.news.yahoo.com 34 - 5 - - - - www.bing.com - 15 5 5 43 3 4 - - www.ssl.bing.com - - - 18 8 - www.tassiefireswecanhelp.com - - - 26 38 7 71 www.webmistresshq.com.au - - - 4 - - - www.tfwch.websmistresshq.com.au - - - 43 - - - www.lifeonplanetbaby.com 28 - - 3 - - - - - www.fullersbookshop.com.au - 6 2 2 - - - - -
Table 5.5.Locowise insights: Top 10 websites referring users to TFWCH (continued)
Variable Time period
Week 1 Week 2 Week 3 Week 4 1st Qtr. 2nd Qtr. 3rd Qtr. 4th Qtr. Year 1
www.sensis.com.au - 4 - - - - www.hightonvet.com.au - 3 - - - - www.ask.com - 3 - - - - www.google.es - - 1 - - - - www.clintonpower.com.au - - 1 - - - - www.bookendtrust.com - - 1 - - - - www.google.com.pk - - - 1 - - - - - www.meganaiemma.edublogs.org - - - 1 - - - - - www.storify.com - - - 1 - - - - - www.ultra106five.com - - - 1 - - - - - www.australianhorserescue.com - - - 1 - - - - - www.google.com.ph - - - 5 - 1 - www.google.ca - - - 4 - - - www.google.com.my - - - 3 - - - www.google.com.br - - - 4 - - www.google.pt - - - 3 - - www.french.frontierservices.org - - - 3 - - www.livinglovinghobartblogspot.com.au - - - 2 - www.google.it - - - 2 - www.google.com.sg - - - 1 - www.melanieirons.com - - - 1 - www.google.so - - - 1 -
As can be seen in Table 5.5, a number of websites had links to the TFWCH Facebook page: 858 in the first year, and together these were responsible for 4,085 people arriving at the page. The largest referring website was The Mercury
newspaper homepage, the state’s leading local newspaper and one of the mainstream media sources most closely linked to the events.46
The Australian version of the Google search engine was the second largest referrer, suggesting a number of fans had searched for the Facebook page through Google, having perhaps heard about it elsewhere first, or perhaps they were simply conducting a general search for information on Google and were led to this page. The third most influential referrer was only operational for the first week of the fires. This referrer was an unidentifiable discussion group built on Facebook forums, but this forum can no longer be accessed, and seems to have closed down sometime in the first week of the fires.
The number of referrals drops over time, presumably as information and stories about the bushfires dropped out of the news cycle and out of the content on other external websites. There was a jump in referrals in the 3rd quarter of the year, which is attributed to the airing of the ABC Australian Story on the topic of the Administrator of the TFWCH page and the TFWCH story. Interestingly there were still 70 referrals to the page as late as during the fourth quarter of the year.
Page users
The next set of results examines the sex and age breakdown of the fans and most engaged users of the page. Table 5.6 below presents the results in frequencies.
46The Mercury had posted the TFWCH Facebook page as their banner for a number of days in the first week of
the emergency, as well as posting a number of stories about TFWCH, with a link to the page embedded. It is unknown what stimulated the decision to promote the page. It is argued that this gave the online emergent group some credibility, legitimacy and promotion.
Differentiating between people who are simply fans of a page (people who like a page) and people who share stories is important because fans can be following a page, but be relatively inactive, while those who actively generate stories are more engaged.
As described earlier,47 for the sex and age statistics, the results represent neither an average nor an accumulated score. They represent how many fans were of each sex and each age on the final day of each time period under investigation. Thus Table 5.7 and Table 5.8 below present the average sex and age breakdowns of the fans of the page. In order to do this, three time periods (the final day of Week 1, the final day of the first quarter, and the final day of the year) are averaged in
percentages.
47 See this chapter, Data Analysis
Table 4.1.6. Locowise insights: TFWCH fan base by sex and age across all time periods (frequencies)
Variable Time period
Week 1 Week 2 Week 3 Week 4 1st Qtr. 2nd Qtr. 3rd Qtr. 4th Qtr. Year 1
Frequency
The number of male fans 4,083 4,048 4,027 4,012 4,000 3,999 4,295 4,210 4,210
The number of female fans 16,562 16,301 16,173 16,092 15,888 15,684 15,918 15,897 15,897 The number of males sharing stories 11,020 2,683 884 345 16,096 386 443 62 16,987 The number of females sharing stories 53,731 17,788 5,893 2,189 87,253 2,540 1,813 266 91,872
Male fans 13 – 17yo 416 408 409 402 398 371 350 324 324
Female fans 13 – 17yo 947 939 927 921 883 822 790 724 724
All fans 13 – 17yo 1,363 1,347 1,336 1,323 1,281 1,193 1,140 1,048 1,048
Male fans 18 – 24yo 891 878 874 872 871 873 915 915 915
Female fans 18 – 24yo 2,963 2,898 2,869 2,851 2,759 2,683 2,649 2,581 2,581
All fans 18 – 24yo 3,854 3,776 3,743 3,723 3,630 3,556 3,564 3,496 3,496
Male fan 25 – 34yo 1,237 1,217 1,199 1,191 1,182 1,186 1,212 1,211 1,211
Female fans 25 – 34yo 5,398 5,271 5,209 5,171 5,125 5,039 5,038 5,045 5,045
Table 5.6. Locowise insights: TFWCH fan base by sex and age across all time periods (frequencies)(continued)