3 ms·

And because means are spectacularly unstable so just one or two feeling that way will show up.

by tel 16y ago

smallblacksun 16y ago
Like the guy who gave $500
tibix 16y ago
In the first Indie Bundle [http://www.wolfire.com/humble http://www.wolfire.com/humble], they didn't show these results until the last days, and the stats were similar. Linux Users weren't trying to show off, it was something that just happened.
- tel 16y ago
  Means are still spectacularly unstable to small numbers of outliers. They're more stable with higher numbers of samples, but so is any given measure of centrality. So, a few Linux users donating $500 could sway an identical distribution to the Mac or Windows users.
jmillikin 16y ago
The current statistics are: total revenue: $301,252.18 number of purchases: 43,819 average purchase: $6.87 average windows: $5.72 average mac: $7.72 average linux: $13.72 Based on their "total revenue by platform" chart, I estimate the per-platform revenues to be: windows: ~54% , or ~$162,676.18 mac: ~23% , or ~$69,288 linux: ~23% , or ~$69,288 An approximate Linux revenue of ~$69,288 and average purchase of $13.72 gives us ~5050 total Linux purchases. Now, it's true that a very large contribution could influence the mean. However, they also include a list of the top 10 contributions: top 10 contributions: $1,000 $700 $666 $650 $600 $500 $500 $337.42 $313.37 $300 We can assume that every one of these was designated a Linux purchase, treat them as outliers, and remove them from the statistics. The adjusted statistics are: outlier revenue: $5,566.79 adjusted total revenue: $295,685.39 adjusted linux revenue: ~$63,721.21 adjusted linux average: $12.61 So even removing the top 10, Linux users are still significantly more generous (on average) than users of the other two major platforms. You could argue that a sufficient number of Linux users donating in the $100-$300 range could skew the results, but I believe such a high number of extremely generous users would prove my point anyway.
- tel 16y ago
  Good argument. I buy it, largely because it wouldn't surprise me in the least if Linux users are actually more generous. I just often dislike data summaries. If you want to show me the data, show it directly.

And because means are spectacularly unstable so just one or two feeling that way will show up.

by tel 16y ago

And because means are spectacularly unstable so just one or two feeling that way will show up.

smallblacksun 16y ago

Like the guy who gave $500

tibix 16y ago

In the first Indie Bundle [http://www.wolfire.com/humble http://www.wolfire.com/humble], they didn't show these results until the last days, and the stats were similar. Linux Users weren't trying to show off, it was something that just happened.

tel 16y ago
Means are still spectacularly unstable to small numbers of outliers. They're more stable with higher numbers of samples, but so is any given measure of centrality. So, a few Linux users donating $500 could sway an identical distribution to the Mac or Windows users.

jmillikin 16y ago

The current statistics are: total revenue: $301,252.18 number of purchases: 43,819 average purchase: $6.87 average windows: $5.72 average mac: $7.72 average linux: $13.72 Based on their "total revenue by platform" chart, I estimate the per-platform revenues to be: windows: ~54% , or ~$162,676.18 mac: ~23% , or ~$69,288 linux: ~23% , or ~$69,288 An approximate Linux revenue of ~$69,288 and average purchase of $13.72 gives us ~5050 total Linux purchases. Now, it's true that a very large contribution could influence the mean. However, they also include a list of the top 10 contributions: top 10 contributions: $1,000 $700 $666 $650 $600 $500 $500 $337.42 $313.37 $300 We can assume that every one of these was designated a Linux purchase, treat them as outliers, and remove them from the statistics. The adjusted statistics are: outlier revenue: $5,566.79 adjusted total revenue: $295,685.39 adjusted linux revenue: ~$63,721.21 adjusted linux average: $12.61 So even removing the top 10, Linux users are still significantly more generous (on average) than users of the other two major platforms. You could argue that a sufficient number of Linux users donating in the $100-$300 range could skew the results, but I believe such a high number of extremely generous users would prove my point anyway.

tel 16y ago
Good argument. I buy it, largely because it wouldn't surprise me in the least if Linux users are actually more generous. I just often dislike data summaries. If you want to show me the data, show it directly.