Some issues I take with the application of the methodology...
1. You state the following in the OP: "For these purposes, "shown" means displayed footage with an associated title."...yet you later seemed to suggest that you abandoned this when deciding when to count Halo: TMCC as being announced.
2. You are not actually addressing the kind of research question necessary to speak to the general impression you sought to clarify. In other words, if you sought to understand whether the general impression in the community, which suggests Sony has a longer window for announce-->release, then you should not be counting 3rd party titles at all. 3rd party titles do not generally contribute in any way to that impression so why would you include them in trying to address said impression?
3. Starting at E3 2013 would seem to strongly favor one competitor here over the other since it somewhat arbitrarily restricts the analysis only to games announced for current gen. Of course, there are high profile games form last gen that got announced and had to get revived on PS4, like The Last Guardian.
4. Your methods skew the data to favor the platform with more titles recently announced. As those titles get announced their time to release starts at zero and seems like it should drastically drag down averages. You note this issue in your OP but seem to act as if it is no germane to your conclusions somehow. But...it totally distorts the context for how you are interpreting the data if you count it. In other words, the data is now reflecting information that is suddenly relevant to a totally different question than you set out to look at, yet your interpretation did not change.
5. Refusing to share your data makes zero sense. That is borderline indefensible given the stated purpose of your 'research' efforts. That is not how serious research is done. I know, 'vidya games' or whatever but when ya put so much effort into the OP and follow up replies it becomes hard to imagine a rationale for refusing to share the raw data.
6. I know some of these issues are already mentioned and discussed beyond the OP, but the fact you did not seek to address them a priori makes me even more dubious as to your actual application of an appropriate methodology.
7. I checked the 18 titles you said you used in your Xbox One analysis for only first/second party titles, excluding Crackdown 3 since that is not released yet. Using the dates that I found online for when each was announced/released I got pretty significantly different numbers than you did. I can't see what you did in terms of your actual calculations since you refused to post your data, but I got 10.76/12.12 "months" (aka 30 days) for the 18 Xbox One games depending on how we count Ryse.
Further, if I include Dead Rising 4 (why did you leave that out?) I arrive at 10.5/11.78 depending again on how we count Ryse. When I add Zoo Tycoon we get 10.25/11.47.
8. Here, this is what you said was your research question:
I was trying to answer, "If you see a game for the first time in a press conference, what's likely to happen with it? What are the chances it'll be canceled? How long will it probably take to come out, based on the platform, game scale, exclusivity, event, etc.?"
But you did not provide any contextual stats analysis at all, making the notion that you were researching probabilities about future release intervals and/or cancellations and whatnot nonsensical. You don't use small samples and frequentist stats for that kind of analysis, ya look towards Bayesian models instead.
What your data tells you is merely how likely it is that had you made a list of all announced titles thus far and closed your eyes and chosen one at random, how likely it would be that the title you chose was released as of today. That is very different than your stated questions quoted above.