Quote Originally Posted by Ghishlain View Post
Random stat breakdown (take with a grain of salt as due sample size and limiting respondents to only the official forum community):

Total Votes as of May 23rd, 2016 - 1242 EST (thread approximately going for 17 days - can be skewed as posters can vote for multiple options)
Total Vote - 357 Votes (+51 respondents)
No Parser - 61 Votes (17.09%)
Yes, Group Parser - 180 Votes (50.42%)
*Screams internally* You are still including percentages which are based on the total number of votes and not voters (a sample of the population we are studying). To give you an idea of how badly it allows you to misinterpret the data, I give you a scenario where 61 people voted all 3 as good options and 55 voted for exactly 2 options. A total of 180 voters. Woosh, we get 33.89% no parser, 100% yes (group) parser and 64.44% personal parser. The percentages nearly doubled. How about a scenario where 116 people voted for both parser options, and everyone else picked one option? A total of 241 voters. Now we get 25.31% no parser, 74.69% (group) parser, 48.13% personal parser. "Can be skewed" does not quite cover how utterly wrong your percentages could be. Some people have already announced that they voted for more than 1 option, so we know for a fact it's skewed.

As I said earlier, you should be comparing the options to each other, not to the total number of votes.
180 / 61 = 2.95: The voters want a (group) parser 2.95 times as much as no parser.
180 / 116 = 1.55: The voters want a (group) parser 1.55 times as much as just a personal parser.
116 / 61 = 1.90: The voters want a personal parser 1.90 times as much as no parser.

This method of analysis will not fail regardless of how many options each person voted for. I can do the same calculations with your hypothetical scenario, where everyone only voted once and the total number of votes represents the total number of voters. I can also do the calculations with my own scenarios where the number of voters is much lower. The ratio between the options is always the same, because it is information that does not contain unknown variables.

Your scenario:
50.42% / 17.09% = 2.95
50.42% / 32.49% = 1.55
32.49% / 17.09% = 1.90

My scenarios:
100% / 33.89% = 2.95
100% / 64.44% = 1.55
64.44% / 33.89% = 1.90

74.69% / 25.31% = 2.95
74.69% / 48.13% = 1.55
48.13% / 25.31% = 1.90

Sorry if it seems like I'm nitpicking. I'm a fiend for numbers and this really rustles my jimmies. Just trying to make your random stat breakdowns a little less... random. Misinterpreting data helps no one and there have already been people referencing your posts in their flawed conclusions. (I recall someone summing up that half the population wants public parsers. No, half the population does not necessarily want a group parser. I just showed above that it could be as high as 100%.)