The Study Data
Before beginning, the survey data need to be pruned back in order to make the graphical representations manageable.
First, 21 game periods are identified in the survey. For this work, the Top 6 game periods (highlighted Periods in graphic below) will be kept for later analysis. These Top 6 include three fantasy/sci-fi periods and three historical periods.
|Top 6 First Choice of Game Period|
|Top 4 First Choice of Game Type|
|Top 3 First Choices of Figure Size|
Multiple Correspondence Analysis
Multiple Correspondence Analysis (MCA), briefly, is a statistical technique from which any underlying structure in the categorical response survey data may be detected. The results are then presented in graphical form in two dimensional space. Don't throw in the towel quite yet. Interpretations are understandable and intuitive without knowing the underlying technique.
MCA of Top Game Periods by Game Type and Figure Size
Pushing all of these data through the algorithms produces the graphic shown in Figure 1.
Dimension 1 Tendencies
In Dimension 1, only four of the attributes are identified as loading in this space. Those attributes are 28mm Heroics, 15-18mm, Ancients, and Napoleonics (see Figure 2). Notice two distinctions in the Dimension 1 space. Ancients, Napoleonics, and 15-18mm are grouped in close proximity in the right half of the graphic. 28mm Heroics groups by itself in the left half of the graph and far from the Ancients, Napoleonics, and 15-18mm group. These results suggest that 28mm Heroics is very dissimilar to 15-18mm and that 15-18mm gaming tends toward Ancients and Napoleonics game periods. Also notice that non-historical game periods group on the left and historical periods on the right frame.
Dimension 2 Tendencies
In Dimension 2, all remaining attributes are identified in this space and encircled to ease identification (see Figure 3). Notice that Warhammer40k finds itself in the top half of the graph along with Campaign and Big Battle game types. All other attributes group into the bottom half of the graph. What inferences can be made in the Dimension 2 space?
First, notice the distance between Warhammer40k in the upper half and SCI_FI in the lower half. This result suggests that respondents choosing Warhammer40k are very different from those choosing Sci-Fi. Notice also that Warhammer40k tends toward Campaign and Big Battle gaming over Skirmish gaming. Finally, Fantasy, Sci-Fi, and WWII game periods tend toward 25-28mm figure size and Skirmish game type.
With tendencies identified in the first two dimensions, does a breakdown into quadrants make sense? Are more tendencies discoverable? See Figure 4.
As for the WH40k and Campaign relationship, notice that 'Campaign' is situated near '0.0' along the Dimension 1 axis (x-axis). Being closer to the axis signifies a less strong relationship. so, I suggest the relationship between WH40k and 'Campaign', while present, is not a strong one. 'Campaign' simple tends toward WH40k with respect to the other game types.
Now, to me, the analysis is remarkable in that data from the nearly 11,000 surveys can be distilled down to produce these results and conclusions. This exercises demonstrates the possibilities of data analysis and the insights attainable given some effort.
What, in this MCA analysis, stands out of interest to the reader? Are there surprises or are these results accepted, common knowledge?
Next time, I plan to toss age group into the mix and see if any clear and insightful tendencies surface. Will age muddy the insights gained or contribute to a more meaningful interpretation of relationships between game period, game type, figure size, and age group?