The Gamebuino concept: a survey about coding experience and gaming habits

Extract

[...] We proceed by steps - we use the same specification whereby purchase of a Gamebuino at a price 79 euros is controlled for by variables extracted from the survey, and gradually added to the original specification. In order to select the appropriate variables for our regression we report the correlation matrix between 10 variables, and point out those with statistically significant results. Table 12: Correlation matrix results - Purchase Gamebuino at 79 euros Gender 0,141 Price range -0,168 (standard error) Age 0,256 Actual coding experience - Studies -0,038 Try New Game - Code knowledge 0,168 Gaming Platform Past experience coding 0,183 Self-building game (note: standard error for correlation levels are in the grey rows. [...]

[...] The first cluster is to the left, where the Gamebuino may interest prospective players with a good to solid background in coding and coding experience. These are likely to be interested in the challenge of building their own game. The second cluster also shares the interest in self-building a game, but it is less reliant on past coding experience, and more on gaming experience itself, or in this case a gender-based game instead. These two profiles suggest that there are two main customer profiles that Gamebuino may be interested in marketing their products to. [...]

[...] To that effect, we start by looking at potential imbalances in the gender composition of the surveyed sample. The figures are reported in the table below, using the ANOVA technique. Table ANOVA regression results Variable Age Education Coding Experience Build Game Male 0,207* -1,628* Intercept 18,465*** Sample Size R RMSE RSS Fisher Log-Likelihood -197,882 -172,024 -75,628 -292,290 -172,584 (note: p-value is reported in star. Legend: * p [...]

[...] Table Coding experience and coding platforms. Prior coding experience No Yes Total Calculator Softwares Others No Prior Experience Total By contrast, the sample appears to have some working knowledge of coding on calculator, something that is mainly reported by high school students, who make up 60% of respondents for this particular platform. On the other hand, a substantially large percentage of those would have declared they have coded before have done so on software - it can be assumed that the main reason why individuals who have no experience in coding declared they knew it were mainly thinking of their high school experience of coding on calculators. [...]

[...] We can see that there are substantial differences, both statistically and qualitatively across the board. In this case, coefficients should be only interpreted with respect to their statistical significance, and not their sign or size. That is because each category variable has been allocated an arbitrary number, - though it does not affect the statistical result itself, its interpretation as a number is meaningless. We can see that coding knowledge is not statistically different across groups with respect to the Business school reference group, an interesting result given that the sample includes Engineering students. [...]