dave said:
What I did was tried to find the average and go from there. I looked at the tens of thousands of photos in the database and looked at when the photos were taken. Using that information I was able to make a report of the average time of people taking photos of each kind of plant, and we went from there.
I understand about the Southern Hemisphere - and the notion that a worldwide audience is tough to cater to. But it would seem to me that the overarching value here (when selecting dates) would be to parse the data and make decisions with an eye to being inclusive rather than exclusive. Tho I suspect (and hope) that your methodology ("go from there") is a little more robust than how it reads in your reply.
But forgive me for saying so, but I don't understand the math of taking an average in this case (and using the average as the time). Assuming that "go from there" means using the average bloom time for the week of the discussion, that effectively (and necessarily) leaves out half of the available pool. That's the nature of average. Half of the datapoints are excluded when you draw that line.
Further, say that you have a sample size of X individual entries in a dataset...contributed by a variety of people. If each photo counts as a datapoint toward that average, and if a single individual contributed even one more photo than his/her cohorts, the average bloom time would necessarily be skewed. Which is fine if you're catering to a single individual who enjoys posting pictures, but makes no sense if you're thinking about it on a per-garden basis (i.e. reaching all those who participated in some way).
Your site and you'll do what you do. But that explanation of the math used to support the dates is arguably more exclusive than inclusive.
If this explanation doesn't make sense, I can create a few graphs with sample datasets in support. But if you think of your dataset as a smoothish bell curve of some sort (and if it's not smoothish, the problem that I raised in my 2nd point is in play), each additional day on the x-axis equals more inclusiveness. So the first question for me would be, "what percentage of the contributors do I want to include in the discussion" rather than even
think about the average. That number isn't relevant, in my opinion, in decision support here if the goal (which I'm admittedly mapping on to you) is to get as many people as reasonably possible contributing to the discussion.
YMMV, of course.
