Inverse Rule of Gaming: Part III–Early Results

Okay, the initial results are in!  I sampled 100 games at random on Boardgamegeek, recording the average rating (of all games that I found with at least 5 ratings) and assigned a salicousness score from 0 (no females depicted) to 5 (practically softcore pornography) based on the cover art/photo (nb: if there was no cover art/photo, I omitted that game).

I found games from 2017, games from pretty much all decades since the 1940s, some classic games (like monopoly), and even some games from my childhood that I had forgotten about (I’m talking about you, Chopper Strike!).

Here are the results:

Is there a correlation between game rating and salacious rating?



Game rating: score on BGG 1 (lowest) to 10 (highest)

Salacious rating: score from 0 (lowest) to 5 (highest)


Hypothesis: the higher the Salacious rating, the lower the Game rating.  


RESULTS (from n=100, simple random sampling)



The trend line is clearly a negative relationship: then greater the Salcious rating, the lower the BGG rating.  However, there are a number of issues.

  1. There are not many cases where Salcious > 0.
  2. There is not a single case where Salacious > 2.
  3. There is a single case with both a low BGG and Salacious = 1 that might be driving the estimate.

What to do?  Well, for my next analysis I am going to use cluster sampling.  I already have enough cases where Salacious = 0.  I will need to look at batches of possible cases and sample only those with Salacious > 0.  This will give me enough cases in the sub-groups to run a more robust analysis.


Early Verdict:

The small sample (n=100) seems to imply that the hypothesis has some credibility.  However, a larger sample with more cases of salacious marketing on covers needs to be done.

Next Time: The Inver Rile of Gaming: Part IV–The Final Results


Inverse Rule of Gaming: Part II – The Methodology

Welcome back!  Today in Part II of my examination of the Inverse Rule of Gaming, I outline my research methods.  Wait…you don’t remember what the inverse rule of gaming is?  Well, I am here to help!

Inverse Rule of Gaming: The more female flesh and/or salacious images used to market a board game/table-top game/RPG/war game/etc., the more likely the game is poor.

If you need more information, check out Inverse Rule of Gaming: Part I — The Theory.

Research Methods

The first thing is to operationalize my variables.

Independent variable: Salaciousness — the degree that sex as represented by female flesh, sexual poses, sexual innuendo, etc, is depicted in the cover art of the product.  This is an objective measure and your faithful narrator, me that is, is going to code box covers.

Here is the ordinal scale that I am going to use:

0 – No female representation at all

1- Female(s) depicted, but in normal/appropriate clothing

2 – Female(s) depicted with exposed flesh/nudity

3 – Female(s) depicted with/without nudity and in an alluring/suggestive pose

4 – Female(s) depicted in a pose that connotes a sexual posture or a great deal of flesh exposed

5 – Female(s) depicted in a pose that connotes pornography or sexual acts

Clockwise from top left: Indy Car Unplugged=0, One Deck Dungeon=1 (females, but all clothed appropriate for combat), Warlord: Sage of the Storm=2 (notice the breasts sticking out and unneeded skin showing), Android: Infiltration=3 (basically a nude robot), Tales of the Arabian Nights=4 (a lot of flesh and a sexual posture), Oral Sex! The Game=5 (duh!).

Sampling Method

I will employ simple random sampling for my poll.  How do I do this?  Here is the method:

1- go to

2 – Hover the cursor over “Browse”

3 – Click on “Random game”

4 – Obtain the “average rating” and determine the “salaciousness” of the art.

I intend to sample 100 games for my “early results” just to see if any association is present.  I hope to sample 1000 games for my complete results.

Data Analysis Method

Given that the independent variable is ordinal and the dependent variable is interval and likely normally distributed (or a simple transformation can make it approximate a normal distribution), a One-Way Analysis of Variance (ANOVA) would be the best associative method to use.  For those unfamiliar with the method, check out the Wikipedia entry  here.

Okay, that’s it for now until Part III – Early Results.

Make Mine Marvel!