Significance of sanity, user group, and other posting characteristics on saloon pecking order rank for 60 salooners: a cross-sectional study.
ABSTRACT: Forum popularity is ill-defined, and often based on highly subjective criteria. This study's aim is to determine the most important correlates of saloon popularity ranking, using generally accepted criteria for measuring poster performance.
Methods: 60 Salooners were nominated for inclusion in the study based on their recognizability among daily contributors during a two-week period leading up to the study. The Salooners were placed roughly into 3 tiers and then assigned relative sanity and pecking order values by 1 observer (also participating in the study). Minor adjustments were made to the values at 4 different time points in the study, in response to complaints from study subjects about their initial assignments. The names were then added to a white board in their approximate X, Y positions on a hypothetical graph of sanity vs. rank.
Data Analysis: The most recent iteration of the white board was opened in Photoshop with a 1-pixel grid overlaid to help find the centroid of each username. Usernames were converted to all lowercase to avoid serif bias when calculating username centroids. The X, Y value of the centroid of each username were found using the Area Info tool in Photoshop. A separate reading was made using the 10-pixel box method. The values were averaged and entered into an Excel spreadsheet for each of the 60 participants (by one of the participants).
In addition, the data examiner scanned each user's actual profile on HG for the following: User group, tier, numbers used in username (a surrogate marker of personal maturity), special characters in username (a surrogate marker for length of membership), history of smurfs (maturity) for that user (banned or extant), and history of bans for any reason for that user (a touchstone variable for relative sanity). These were tabulated in the spreadsheet.
To detect correlations between the user variables, the data was plotted in scatterplot form, and correlation coefficients were obtained for each variable pair. To determine any correlation between relative sanity and rank, a two-tailed T test was performed with p0.05 held significant. To test for user group influence, mean relative sanity for each user group was compared.
RESULTS
Overall, there was no correlation between relative sanity and Saloon pecking order rank.
Correlation coefficients were highest between Tier and Pecking Rank (r=0.92) and Tier and Usergroup (r=0.60). These may be influenced by the nature of assigning users to tiers or usergroups based on the same criteria as Pecking Rank. The next highest group of correlates were only indirectly biased: Relative Sanity and Career Bans (r=-0.57) were highly inversely correlated; Relative Sanity and Smurfs (r=-0.32) less strongly. Smurfs and Bans (r=0.39) were also strongly correlated.
Sanity was highest in HG staff, followed by visitors. Regular Salooners and GDers had similar average sanity, with slightly higher variation in GDers. The Saloon Core Group scored poorly and was not statistically significantly different from the Legends group.
DISCUSSION
The question of popularity and relative ranks in a community is an old and thorny one. Soft criteria such as maturity, length of membership, and relative sanity are often invoked to explain differences in popularity. This study sought to validate relative sanity as a reasonable marker of worth, by looking at sanity over a wide range of subgroups, including forumers who have been banned. While relative sanity is indeed a marker of forum responsibility and the roughest limits of the CoC, this study showed that it does not correlate with pecking order within a group of otherwise CoC-compliant users.
What other factors could be important in popularity? The study did not look at popularity, but with pecking order, which implies a power structure more than a social cachet. As we have seen, core group status (a proxy for seniority) and tier (a surrogate marker of frequency and entertainment quality of recent posts) correlate very strongly with pecking order. Tier is in fact measuring the same quantity and quality as pecking order and should not be considered an independent variable; that is, it's a confounder. Lastly, it can be argued that good guys do not finish last: HG Staff tended to have higher pecking order rank than legends or GDers (p=insignificant).
The study has a number of methodologic flaws and leaves many questions unanswered. Firstly, the rankings, sanity values, and tiers were largely the arbitrary effort of one person who also happened to be participating in the study; the data crunching was done by a non-blinded observer who is also a study subject. The potential for bias is hueg, hueg liek Xbox.
In addition, the data analyst was too lazy to look up each of the 60 Saloon study subjects' profiles to extract such undoubtedly interesting and relevant information such as Registration Date, Total Posts, Posts Per Day, and Last Activity. These are important factors in determining whether someone is remembered and placed in a Pecking Order graph. A currently active, vocal, and voluminous poster is more likely to be remembered than one who is AFK for a year.
Lastly, assumptions about the value of bans and smurfs is in question. A puzzling disparity exists for the Legends: these are posters who have been banned, or who have skirted the rules and earned infamy. Their distribution on the Sanity scale is predictably low, however their popularity remains sharply divided between the highest and lowest values for Rank or Tier. This indicates that although Sanity is predictably low, it is no handicap to Pecking Order rank, and indeed their placement high on the pecking order rank list seems to invite speculation about how much a power structure depends on people who are entirely unable or unwilling to respond to the needs of a community; either through being banned, or being insane.
More research is needed to find the best determinants of Pecking Order and further differentiate it from popularity; and a multiuser, double-blinded ranking system should be used to gather less biased data on Saloon posters. Lastly, the duration of Saloon posting should be compared with relative sanity over time in select posters, to determine if greater exposure to the Saloon is deleterious to mental health.
EDIT: ACKNOWLEDGEMENTS: We would like to recognize Mountain Dew/Pepsico Inc., and Haldiram Foods Nagpur Inc., for their considerable support during the study period.
ERRATA: Under Methods, it should be noted that KS (Salooner) and Beagle (Legend) were added to the study to round out the number of participants to 60.
Crunkatog on ESO
Bart331 balance suggestion: aztec: remove civ
Voltiguer: Ender, Sioux in 1.04 will be a top civ, no matter how many layers of Sioux goggles you put on
schildpad on Elephants: ...their mansabdar unit sucks so hard it looks like a black hole
Crunkatog on Steam.[This message has been edited by As_Saffah (edited 10-30-2009 @ 01:50 PM).]