[FRC Blog] Two Championship Survey Results and Path Forward

[Citrus Dad]

One question we asked was just a simple ‘How do you feel about having two FIRST Championships starting in 2017?’ with an answer of 1 representing ‘Strongly Oppose’, an answer of 5 identified as “Neither Oppose nor Favor” and an answer of 10 identified as “Strongly Favor”. The average answer to this questions among all respondents was 4.45, somewhat below the 5 "Neither Oppose nor Favor" rating.

This statement is misleading. Based on the graphic, the respondents to the survey oppose the championsplit 62.5% to 37.5% (after removing the '"5s"). That's a pretty overwhelming landslide in opposition. I'm not sure if a presidential candidate has ever exceeded that in the popular vote. The decision looks to be deeply, deeply unpopular.

[Siri]

Originally Posted by Citrus Dad

This statement is misleading. Based on the graphic, the respondents to the survey oppose the championsplit 62.5% to 37.5% (after removing the '"5s"). That's a pretty overwhelming landslide in opposition. I'm not sure if a presidential candidate has ever exceeded that in the popular vote. The decision looks to be deeply, deeply unpopular.

That's actually a very interesting comparison. Wikipedia tells me no; the highest popular vote take home was LBJ in '64 with 61.05%. This means that no US president since 1824* has had this level of a popular vote mandate. Granted, even the US has better voter turn out than this. *First available data.

[grstex]

Originally Posted by Citrus Dad

This statement is misleading. Based on the graphic, the respondents to the survey oppose the championsplit 62.5% to 37.5% (after removing the '"5s"). That's a pretty overwhelming landslide in opposition. I'm not sure if a presidential candidate has ever exceeded that in the popular vote. The decision looks to be deeply, deeply unpopular.

26+11+11+7=55%. I don't know where you're getting 62.5%. Even if you split the 12% of Neutral responses, that brings you to 61%. But splitting neutral responses really wouldn't be fair analysis. That's like counting undecided voters as going to one candidate or another, when they've clearly stated they're undecided.

[Knufire]

55/88 = 62.5, neutral voters were removed from the total as well. So 62.5% of respondents who did not answer neutral were in the opposing range, which is exactly what his statement said.

[grstex]

Originally Posted by Knufire

55/88 = 62.5, neutral voters were removed from the total as well.

I see. But that doesn't actually represent the response to the survey. That's only a poll of those who are biased one way or the other. It's not like the neutral opinions don't count.

[Siri]

Originally Posted by Basel A

I don't think you can look at this data and reasonably say "most of FRC is opposed to two Championships," especially when the nonvoters likely don't care/are neutral.

It's not 62% of respondents, but you can absolutely look at the data and say that: 55% of respondents are opposed to two Championships. That's most.

We can speculate all we want about the opinions of the people who didn't vote, but there's nothing to delineate the reasonableness of those speculations.

Originally Posted by grstex

I see. But that doesn't actually represent the response to the survey. That's only a poll of those who are biased one way or the other. It's not like the neutral opinions don't count.

It's a measure of mandate that's intended to elucidate the misleading nature of the "average" purported in the blog. As yet we don't know of any way to properly center the data (the actual average of 1 to 10 is 5.5, whereas neutral is a "5"). Directly calculating the relationship between those who fall on one side or the other of neutral provides another sort of insight into the flaw in the scale.

[grstex]

Originally Posted by Siri

It's not 62% of respondents, but you can absolutely look at the data and say that: 55% of respondents are opposed to two Championships. That's most.

We can speculate all we want about the opinions of the people who didn't vote, but there's nothing to delineate the reasonableness of those speculations.

It's a measure of mandate that's intended to elucidate the misleading nature of the "average" purported in the blog. As yet we don't know of any way to properly center the data (the actual average of 1 to 10 is 5.5, whereas neutral is a "5"). Directly calculating the relationship between those who fall on one side or the other of neutral provides another sort of insight into the flaw in the scale.

But you can't say "62.5% of respondents oppose the split." That's just not true. the "mandate" is that 55% oppose the split. you CAN'T just discard 12% of the responses. That's more misleading than average from the blog.

[Siri]

Originally Posted by grstex

But you can't say "62.5% of respondents oppose the split." That's just not true. the "mandate" is that 55% oppose the split. you CAN'T just discard 12% of the responses. That's more misleading than average from the blog.

I agree that the statement "62.5% of respondents oppose the split" is not true. I suspect everyone does; it's math. As a side note, the statement "the respondents to the survey oppose the championsplit 62.5% to 37.5% (after removing the '"5s")" is completely true.

However, unless you have a mathematical or industry standard to support the conclusion that 62.5% is more misleading than 4.45, I disagree. At the very least, Richard actually told us directly what his calculation was in the midst of a discussion that already took issue with the neutrality of the 5 average. Frank left his misleading calculation to be discovered, which is a huge problem in itself. I don't think that this was intentional by Frank. A very big part of this problem is that this is an intuitive scale on its face, but he should've done his homework before making a highly misleading and unqualified statement that included both the term 'average' and the term 'neither oppose nor favor'.

The correct 'intuitive' truth that we're looking for--i.e. what the average looks like when centered about neutral--is somewhere between Richard's calculation and Frank's average. There's no way to access it. Do you have a better method of getting closer? This is an iterative issue; Karthik took one approach, I tried another averaging technique.

[Mike Schreiber]

As someone who has participated in the planning and execution of customer surveys and clinics before, I'm just going to leave this here for future reference.

Likert Scale

[Citrus Dad]

Originally Posted by grstex

But you can't say "62.5% of respondents oppose the split." That's just not true. the "mandate" is that 55% oppose the split. you CAN'T just discard 12% of the responses. That's more misleading than average from the blog.

I approached this issue as a presidential election. You can't vote for "neither of the above" or "both of the above." You have to choose. The "5s" refused to choose. In an election, those folks don't vote--it's a very common assumption by pollsters making projections for election results.

Similarly, we don't apply an intensity of like or dislike to presidential candidates. It's either "A" or "B". There's some indication in 2012 that Romney supporters were more intense in their positions, but there were fewer of them. Ultimately, I believe we should really care about which side people fall on.

One other polling note: while this is a voluntary poll so it could be biased, pollsters find that usually the opinions of respondents generally reflect the views of non respondents.

I used a set of common polling assumptions to provide a clearer view of how community preferences fall out. I see others have provided other metrics that arrive at the same conclusion--that opposition is running 2 to 1 against.

[grstex]

Originally Posted by Citrus Dad

I approached this issue as a presidential election. You can't vote for "neither of the above" or "both of the above." You have to choose. The "5s" refused to choose. In an election, those folks don't vote--it's a very common assumption by pollsters making projections for election results.

Similarly, we don't apply an intensity of like or dislike to presidential candidates. It's either "A" or "B". There's some indication in 2012 that Romney supporters were more intense in their positions, but there were fewer of them. Ultimately, I believe we should really care about which side people fall on.

One other polling note: while this is a voluntary poll so it could be biased, pollsters find that usually the opinions of respondents generally reflect the views of non respondents.

I used a set of common polling assumptions to provide a clearer view of how community preferences fall out. I see others have provided other metrics that arrive at the same conclusion--that opposition is running 2 to 1 against.

There is such a thing as a 3rd party candidate. Ross Perot captured almost 19% of the popular vote in 1992, and over 8% in 1996. in 1928 Robert La Follette even won a state. (of course, you could hold the Kang and Kodos perspective of politics)

But more importantly, in this survey you DID NOT HAVE TO CHOOSE. You WERE GIVEN A NEUTRAL OPTION. 12% chose that option. FIRST could have structured the survey as a simple for or against, but they didn't. As others in this thread have already stated, the people who chose neutral did so for a reason. Their response counts too.

[Basel A]

Originally Posted by Citrus Dad

I approached this issue as a presidential election. You can't vote for "neither of the above" or "both of the above." You have to choose. The "5s" refused to choose. In an election, those folks don't vote--it's a very common assumption by pollsters making projections for election results.

This is absolutely nothing like a presidential election. That is a choice between two people, whereas this is a statement of approval/disapproval of a policy. A good example of what that should look like is here. Note how the approval and disapproval numbers don't add up to 100%. That's because some people are neutral. Here's another potentially enlightening link: http://en.wikipedia.org/wiki/Self-selection_bias

[mklinker]

Originally Posted by Citrus Dad

This statement is misleading. Based on the graphic, the respondents to the survey oppose the championsplit 62.5% to 37.5% (after removing the '"5s"). That's a pretty overwhelming landslide in opposition. I'm not sure if a presidential candidate has ever exceeded that in the popular vote. The decision looks to be deeply, deeply unpopular.

This statement is misleading! There is no valid interpretation of the data that shows 62% of respondents opposing the championship split.

[efoote868]

Originally Posted by Citrus Dad

This statement is misleading. Based on the graphic, the respondents to the survey oppose the championsplit 62.5% to 37.5% (after removing the '"5s"). That's a pretty overwhelming landslide in opposition. I'm not sure if a presidential candidate has ever exceeded that in the popular vote. The decision looks to be deeply, deeply unpopular.

Voluntary survey. Not a census. Not a study. Statistically speaking, the survey means very little to me.

Now if they had picked individuals in FRC at random, with a very high response rate, I might be more inclined to give weight to the results. Otherwise the survey is suffering from a tremendous response bias (how many people that don't have a strong opinion on a subject are going to volunteer their time to do a survey?)

[s_forbes]

Originally Posted by efoote868

Voluntary survey. Not a census. Not a study. Statistically speaking, the survey means very little to me.

Now if they had picked individuals in FRC at random, with a very high response rate, I might be more inclined to give weight to the results. Otherwise the survey is suffering from a tremendous response bias (how many people that don't have a strong opinion on a subject are going to volunteer their time to do a survey?)

This is my opinion towards this as well, and I think it's interesting that so many people are drawing the conclusions they want from the survey results.

The response rate to the survey was only 10%. I'm one of those neutrals in the remaining 90% that didn't respond.