The Empathy of Mittens

[Mehmentum]

You must be logged in to read this quote.

The proposition is true. Marriage in MA in 2010 is much worse off than it was in 1990.

Obviously gay marriage isn't the cure for what is ailing marriage, especially since the marriage rate in MA continues to decline.

True, but I fail to see how the year the divorce rate in 1990, a full 14 years before Massachusests legalized gay marriage, has any relevance in this discussion.

The divorce rate in MA rose between 1990 and 2000, a decade in which no gay marraiges were performed.  It fell between 2000 and 2010, when gay marraiges were performed for roughly half the decade.

[Alcon]

The confidence intervals indicated that the pattern we did find could be attributed entirely to random chance.

...

The evidence says no such thing. The evidence indicates no change in the general divorce pattern that can be attributed specifically to gay marriage. Stop lying.

...

When the difference is statistically insignificant? Absolutely not. We should treat them exactly the same rather then applying false attribution. This is bad science. 

Yes -- which is true of a statistically significant result too.  It's an arbitrary line.  However, a proposition that fails to reach statistical significance is not a 50/50 proposition.  For instance, if a poll finds Obama +4 with a MoE of +/-3%, it is likelier that Obama is up than he is down.  You seem to be treating the result like we should ignore any evidence that has a >5% probability of being statistical noise. 

The evidence that gay marriage tempers the overall decline in marriage rates is weak, and not statistically significant.  Indeed, I doubt gay marriage does much to temper this rate.  The point wasn't that you keep equivocating statistical insignificance with being unusable data.  That's not true.  See my Obama +4 example above; the fact that it's statistically insignificant does not make it equally likely that Obama or Romney are winning.  Similarly, when you are making a binary decision, the fact that evidence supporting one option doesn't reach statistical significant doesn't mean the decisions are equally good.

(For the record, I don't thin that gay marriage tempers the heterosexual divorce rate.  You simply initially claimed it did the opposite.  Evidently, your intuitions are more reliable to you than validly-conducted results that fail to come out significant at the 95th confidence interval.  Interesting.)

The relative sizes have no bearing on the percentage differences. Are you really a statistician? You certainly don't sound like one. Black/black marriages are more likely to fail than either black/white or white/white. That there are more white/white marriages and they are on average more likely to succeed actually puts more evidence in my favor than yours. Larger samples are less likely to bear extremes.

I'm not a statistician.  You misread.

I was replying to this statement: "That interracial marriages do poorer than marriages that are single-race? If Black/Black does worse, then this premise is not true."  This is not true.  The fact that a small subset of same-race marriages (black/black marriages) have inferior outcomes to same-race marriages overall does not mean that, in aggregate, single-race marriages overall don't do better than interracial marriages overall.

In any case, we can agree that interracial marriages have inferior outcomes to white-white marriages, according to the statistics?  Why should the state recognize interracial marriages involving blacks and whites?  (This is not rhetorical -- there are several answers to this you can give.) 

Not according to Massachusetts. Massachusetts says that it is better that the children do not get adopted at all, than to see Catholic adoption services place children in accordance with their beliefs.

I disagree with that, so I'm not sure why you're demanding my argument has to be consistent with it.

"stable gay couples might have marginally inferior outcomes to stable heterosexual relationships"

Again, MA was willing to shut down adoption agencies and see fewer placements than to see placements outside of what they believed to be the ideal. So clearly, placing children in inferior outcomes is not acceptable to the state.

Is there a typo here?  If not, I'm not following your rationale.

Uh, you do realize that I believe that sex outside of marriage is sinful? If you choose not to marry, you're not having children to raise.

And how does that address my rebuttal?

Of course not, because that would mean you are wrong. Read up on it. The state said comply or close, so they closed. If the state valued children above their agenda then they would have permitted the Catholic adoption agency to continue to operate.

What in the world?  I was implying you oppose gay adoption on theological levels, not as a quid pro quo response to state action you see as offensive.  As far as I can tell, this response has nothing to do with that.  I'm familiar with the case, and I don't agree with it.

I was a physicist, so yes I do understand it. From what I can see, if the numbers support your bias, you'll defend them come hell or high water rather than admit that they are merely the result of random chance.

If you think that failing statistical significance is synonymous with proving results "merely the result of random chance," then you received really crappy Stats training.

I do. You're just pissed off that you had to admit that the difference was and is statistically insignificant. Hey, it's your work. You want to argue that you screwed up, find by me. Doesn't enhance your credibility when I'm the one defending the accuracy of your work.

Why would I be pissed off that legalizing gay marriage doesn't have a statistically significant correlation with improving (relative or absolute) overall divorce rates?  I never claimed that happened.  I've said several times I doubt that's true.  I have never advocated for this claim, except in saying that a weak positive correlation indicates something is less likely to be true than the inverse.  I've explicitly said how many times I don't forward the hypothesis you're now accusing me of being "pissed off" wasn't confirmed in my analysis?  How many times did I explain that the intention of my analysis was to test your claim that gay marriage should be assumed the causal agent for increasing divorce rates?  Do you not remember this exchange well or something?

[Wisconsin+17]

You must be logged in to read this quote.

It gives us a snapshot of what marriage used to be like. I can go back further, but it will just make my point even more obvious. Marriage is on the decline - gay marriage is simply accelerating the decline.

You must be logged in to read this quote.

With even fewer marriages being performed - shouldn't we expect that fewer of them would end up in divorce?

[Wisconsin+17]

You must be logged in to read this quote.

Christ. No, it's not an 'arbitrary line.' It's directly related to error bars and sample sizes. You clearly do not understand statistics.

You must be logged in to read this quote.

Yes, it is. I know you refuse to see this, but that is because you don't understand statistics. If a poll is within the MOE - we cannot draw any conclusions from the results. A result is only significant, if and only if, it shows a lead beyond the MOE.

You must be logged in to read this quote.

That would be an example of a statistically significant result. If he were +1 with an MOE of +/- 3, would you draw the conclusion that Obama was more or less likely to be ahead?

You must be logged in to read this quote.

I'm treating the result as statistically insignificant, because it fails to show any change beyond our error bars. There is nothing to distinguish what we did find, and random chance. That is what you are not getting.

You must be logged in to read this quote.

Christ.

No result is usable data.

You must be logged in to read this quote.

O+4 with a MOE of +/- 3 is statistically significant. No wonder you are struggling. You don't even understand statistical significance.

You must be logged in to read this quote.

And you continue to lie. I think I'm done here.

I said, very explicitly, that the results surprised me and that I no longer advance the position that gay marriage has a statistically insignificant result on divorce rates.

The difference between you and me, is that I have abandoned my contention in the face of data, while you have not changed your own position an inch.

You must be logged in to read this quote.

I am. Care to continue to call me 'math impaired?'

You must be logged in to read this quote.

The fact that some same race marriages do worse than other same race marriages, indicates that the common factor is that race can and does influence your marriage. Why is this? Because black people are more likely to grow up in broken homes and are less likely to have a positive example of marriage to look up to.

I know you don't like this conclusion, which is why you persist in blasting me. Which is fine. But, it does get tiresome.

You must be logged in to read this quote.

Will you concede that gay marriage has no effect positive or negative on divorce rates in America? Will you finally discard your thesis now that you've conclusively demonstrated it to be false?

You must be logged in to read this quote.

Christ. Why should the state recognize marriage between black men and black women, since they are more likely to fail?

You must be logged in to read this quote.

Because you're insisting that the Catholic church ought to capitulate to your position.

You must be logged in to read this quote.

Really simple. Placement is not the number one goal of adoption agencies. Agenda trumps placement, at least in MA.

You must be logged in to read this quote.

Your rebuttal assumes that since I believe some people shouldn't marry, that I'm condoning children out of wedlock.

You must be logged in to read this quote.

Ok, good. I wasn't sure. Had you said that earlier, would have saved me some time.

You must be logged in to read this quote.

Door is open. Are you willing to finally abandon your contention that gay marriage lowers divorce rates?

You must be logged in to read this quote.

Because you keep bringing this crap up in a thread that has nothing to do with it. Why? I suppose it makes you feel better to fling crap over and over again.

You tested it. You found that the result was no result.

You must be logged in to read this quote.

Your behaviour afterwords confirms, that yes, you will refuse to abandon your hypothesis even in the face of your own evidence to the contrary.

Anyways, door is open. Will you step through it? "Outlook not so good".

[Alcon]

Christ. No, it's not an 'arbitrary line.' It's directly related to error bars and sample sizes. You clearly do not understand statistics.

You're telling me 95% isn't an essentially arbitrary measure for statistical significance?  The fact that concepts are based on it doesn't mean it fundamentally isn't arbitrary.

That would be an example of a statistically significant result. If he were +1 with an MOE of +/- 3, would you draw the conclusion that Obama was more or less likely to be ahead?

...

O+4 with a MOE of +/- 3 is statistically significant. No wonder you are struggling. You don't even understand statistical significance.

Pretty sure you'e wrong.  Here is a theoretical poll where Obama is up 4:

Obama 47%
Romney 43%

The range of within-MoE values for Obama is +/-3% -- i.e., between 44% and 50%.  Romney's MoE is 40% to 46%.  Anything from Romney +2 to Obama +10 would be within MoE.  MoE isn't calculated on percentage margin.

But you're avoiding the point: say it's an Obama +1 poll.  Does that indicate it's more likely that Obama is leading -- yes or no?  Your argument says no; reality says yes.

And you continue to lie. I think I'm done here.

I said, very explicitly, that the results surprised me and that I no longer advance the position that gay marriage has a statistically insignificant result on divorce rates.

The difference between you and me, is that I have abandoned my contention in the face of data, while you have not changed your own position an inch.

Lie about what?  None of what you just "corrected" me on is inconsistent with what I said.  You did, as I've said before, concede that gay marriage does not have a statistically significant deleterious effect on divorce rates.  What argument did I forward that my data invalidated (outside of the contested interpretation of statistics here)?

I am. Care to continue to call me 'math impaired?'

Yes, until you argument makes sense.

The fact that some same race marriages do worse than other same race marriages, indicates that the common factor is that race can and does influence your marriage. Why is this? Because black people are more likely to grow up in broken homes and are less likely to have a positive example of marriage to look up to.

I know you don't like this conclusion, which is why you persist in blasting me. Which is fine. But, it does get tiresome.

...

Christ. Why should the state recognize marriage between black men and black women, since they are more likely to fail?

YES.  That's the point of a reductio ad absurdum.  You're finally getting it.

By your logic -- shouldn't the government prohibit interracial marriages so as to encourage more "optimal" (white-white) marriages?  Is that a problem for you because it increases the number of black-black marriages?  Or is there some other reason why this is objectionable?

Also, I like how you totally blew off my rebuttal to your ridiculous claim that same-race marriages couldn't be better-off in aggregate because a subsample of them (black-black marriages) weren't.

Will you concede that gay marriage has no effect positive or negative on divorce rates in America? Will you finally discard your thesis now that you've conclusively demonstrated it to be false?

Please quote me when I've ever advanced this thesis.  I've said explicitly several times that this is not my thesis.

Because you're insisting that the Catholic church ought to capitulate to your position.

I don't mind if the Catholic Church doesn't marry gays.

Really simple. Placement is not the number one goal of adoption agencies. Agenda trumps placement, at least in MA.

Have I suddenly turned into the state of Massachusetts?

Ok, good. I wasn't sure. Had you said that earlier, would have saved me some time.

Wait, so I'm not the state of Massachusetts...I just have to preemptively list things I disagree with the state of Massachusetts on, in case you might mention it?

Your rebuttal assumes that since I believe some people shouldn't marry, that I'm condoning children out of wedlock.

So, you wouldn't want to incentivize inferior-outcome interracial and/or black-black marriages, if that didn't result in more children out of wedlock?  If it weren't for the sex-out-of-marriage issue, you'd oppose government incentivization of those relationships?  Or would you support interracial marriage to "elevate" black-only marriages even if they have inferior outcomes to white-only marriages?

(I'm not trying to ask shock questions -- I'm trying to see how your position would ethically analyze these situations, and then explain why I think it's problematic, if I think it is.)

[Alcon]

Also, dude, please find me one instance in which I ever claimed I thought it was likely that legalizing gay marriage decreases heterosexual divorce rates.  In the process, I'm sure you'll find me saying the opposite a bunch of times.

Even if I found a weak, statistically insignificant correlation, I'd never assume causality.  Why would legalizing gay marriage cause a decrease in heterosexual divorce, especially in the short-term?  That hypothesis seems ridiculous to me.  Which is why I never argued it.

Also, for anyone out there still reading this: if I'm missing the obvious and Ben is somehow right here, let me know...

[Politico]

In states that have done it, banning gay marriage is perfectly correlated with the sun rising in the east therefore the sun rising in the east causes gay marriage bans

Folks, correlation does not necessarily imply causality in either direction. I like Ben and Alcon, so I am not getting into that dick measuring contest...

[Wisconsin+17]

You must be logged in to read this quote.

Statistical significance is no more arbitrary than, say, newtonian mechanics. Probability is probability. Is there a chance that the results we obtained could have arisen through random chance? Yes. Is there a chance that results that are statistically significant could have arisen through random chance? Also yes, but extremely unlikely.

You must be logged in to read this quote.

That's O+2.

You must be logged in to read this quote.

That's what you're struggling with. Again, back to our coin flip example. I flip six coins, 4 heads and 2 tails. What's the probability that the next flip will be tails?

You must be logged in to read this quote.

Thank you. Why then are you asserting I said otherwise? I conceded this point when you first asked about it here.

You must be logged in to read this quote.

That you were unwilling to concede that your results indicate no inverse correlation whats

You must be logged in to read this quote.

I think we have a winner here.

I believe marriage is a sacrament. I've never said that the reason that gay marriage should not be permitted is because they are more likely to fail. I have always stated that I believe the state has a desire to perpetuate itself - something gay marriage does not do.

You must be logged in to read this quote.

You clearly have reading comprehension issues. I blew it off because it's already been addressed. Go reread what I said.

White/white marriages fare better than white/black and black/black. White/Black marriages fare better than black/black - ergo single race marriages are not automatically superior. Why? Already addressed.

You must be logged in to read this quote.

Your first post in this thread. Now, can we move on please?

You must be logged in to read this quote.

Reread what you just asked me.

70 percent of black children are born out of wedlock. Are you asserting that black/black marriages are responsible for 70 percent of black children being born out of wedlock? Really, alcon, really?

You've lost the thread quite some time ago...

You must be logged in to read this quote.

One, I support marriage between one man and one woman. I believe this to be a part of the common law, no different than say, habeaus corpus. I would not want to change this for the same reason that you probably don't want habeaus corpus changed.

There are alternatives to english common law. One of these alternatives is Sharia. Insofar as the US is governed on the principles of english common law, the US should retain the entirety of the common law.

You must be logged in to read this quote.

You're asking all the wrong questions. Look, just stop, ok?

You're working out of a worthless framework right now that is not doing the job.

The questions you should be asking are:

"what do you believe the role of the government should be wrt marriage?"
"what do you believe is the purpose of marriage within society"
"what do you believe are the reasons why people should get married?"
"how do you believe people who are married should conduct yourself?"

See where I'm going? Just ditch the crappy argument and we can make progress!

[Mehmentum]

It gives us a snapshot of what marriage used to be like. I can go back further, but it will just make my point even more obvious. Marriage is on the decline - gay marriage is simply accelerating the decline.

Your own post said that marraige in MA was 7.9 in 1990, 5.8 in 2000, 5.5 in 2010.  The decline between 1990 and 2000 was 2.1 or by about 27%. The decline between 2000 and 2010 was .3, or about 5%.

The decline is actually decellarating. 

You must be logged in to read this quote.

I'm not quite sure what your arguing here, sorry but either you were a little vauge or I'm just being overly cautious in interpretting what your saying.
Regardless, I figure:
a) You could be arguing that the divorce rate is decreasing more slowly than the marriage rage.
b) You could be arguing that the people who do get married nowadays are more likely to be in a heathier relationship with each other.

B actually makes sense and I suspect it could be ture, it certainly deserves some looking into, but I fail to see how its a bad thing.  I would actually consider that a good development for our society.

A on the other hand is simply not true.  Your marraige divorce ratios in your other post show that while between 1990 and 2000, there were more divorces per marriage.  Between 2000 and 2010 there were fewer divorces per marriage. 

[Wisconsin+17]

You must be logged in to read this quote.

It's still deteriorating, which is not good.
 

You must be logged in to read this quote.

Except for the fact that a marriage is more likely to break up in 2010 than in 1990, even with the higher divorce rate. That's the problem.

You must be logged in to read this quote.

Christ. It shows just the opposite. There were FEWER divorces per marriage in 1990 than in 2010.

[Mehmentum]

You must be logged in to read this quote.

You must be logged in to read this quote.

Let me put it in the simplest terms I can.  The divorce/marraige situation was getting worse between 1990 and 2000.  The situation has gotten better since 2000.

How about a hypothetical example.

Suppose somebody's weight in 2008 was 160 pounds.
This increased to 240 pounds in 2010.
By 2012, they've started working out and now wiegh 180 pounds.

Would you say that they are currently gaining or losing weight?  Would you project them to increase or decrease their weight if they continue the workout program? 

[Wisconsin+17]

You must be logged in to read this quote.

I would say the fad diet is a failure.

[Mehmentum]

And I'd say you are either trolling me or deluding yourself.  Good day.

[Alcon]

Statistical significance is no more arbitrary than, say, newtonian mechanics. Probability is probability. Is there a chance that the results we obtained could have arisen through random chance? Yes. Is there a chance that results that are statistically significant could have arisen through random chance? Also yes, but extremely unlikely.

I wouldn't call a 5% chance "extremely unlikely."  My point is that you're acting like a 4% probability of statistical chance should be accepted as truth, but a 6% chance shouldn't.  Moreover, you're equivocating anything between 5% and 95% to 50%.

That's O+2.

No, it's not.  Where would you see that poll expressed as Obama +2?  Certainly not on this site, FiveThirtyEight, RCP, any web site I've ever heard of.

That's what you're struggling with. Again, back to our coin flip example. I flip six coins, 4 heads and 2 tails. What's the probability that the next flip will be tails?

50% with a fair coin, but that's definitional, not empircal.  Here's a better example: you have a gumball machine with red and blue gumballs.  The gumball machine comes 50/50 by default, but you're buying it used, so it's possible someone has extracted red and blue gumballs -- basically, there's a good reason it would be 50/50, and a potential reason why it might not be.  (Like with the gay marriage example.)

You randomly draw six gumballs.  Four are red, and two are blue.  Statistical significance is failed.  Is it equally reasonable to predict the next gumball is red as it is to predict it's blue?

Thank you. Why then are you asserting I said otherwise? I conceded this point when you first asked about it here.

I haven't asserted you've said otherwise since you clarified that, as far as I know.  When did I?

That you were unwilling to concede that your results indicate no inverse correlation whats

Statistically insignificant correlation is not zero correlation.  Also, was there more to that sentence?

I think we have a winner here.

I believe marriage is a sacrament. I've never said that the reason that gay marriage should not be permitted is because they are more likely to fail. I have always stated that I believe the state has a desire to perpetuate itself - something gay marriage does not do.

You clearly have reading comprehension issues. I blew it off because it's already been addressed. Go reread what I said.

White/white marriages fare better than white/black and black/black. White/Black marriages fare better than black/black - ergo single race marriages are not automatically superior. Why? Already addressed.

Are you serious?  We've been speaking in aggregates and averages this entire time.  Why would you possibly think that we suddenly shifted to talking about whether all black-black marriages were inferior to all white-white marriages?  That's absurd.  Obviously, there are straight marriages that are inferior to most gay marriages in terms of child-rearing...tons of them.

You are, again, missing the critical question of the analogy.  Interracial marriages tend to have outcomes worse than white/white marriages, but marginally better than black/black marriages -- in other words, incentivizing interracial marriages can be expected to reduce the overall societal outcomes.  Didn't you say that was a reason against incentivizing gay marriage?  Didn't you object to gay adoption on the basis that it was inferior in outcome, even if it was better than being not adopted?  I know you have separate religious arguments, but I'm trying to thresh out your secular argument.

Incentivizing interracial marriage, by your argument (I think), would mean encouraging inferior outcomes, because the outcomes of interracial marriages are closer to black/black marriages than white/white marriages; therefore, if a white and black marry each other instead of someone of their own race, the aggregate utility is lower.  Think of it as a black/black marriage having a utility of .4, interracial marriage as having a utility of .6, and a white/white marriage having a utility of .9.  Incentivizing interracial marriages accordingly incentivizes inferior outcomes for children, both in raw terms (like black/black marriages would) and overall (unlike incentivizing black/black marriages.)

I don't want to ignore your other points yet, but I'm not quite sure how they fit into our back-and-forth on this, so I'm going to temporarily put them on reserve.

70 percent of black children are born out of wedlock. Are you asserting that black/black marriages are responsible for 70 percent of black children being born out of wedlock? Really, alcon, really?

Uh, what?

Your first post in this thread. Now, can we move on please?

No, we can't, because I didn't say I agreed with the causality argument.  I've said I don't many times.  Find me the quote where I said I agreed with it.

[Wisconsin+17]

You must be logged in to read this quote.

Yawn, you're only arguing this in order to salvage your position. Rather than actually change your position in the face of contrary data, you're arguing over what "is" really means.

Is it statistically significant? No. Is statistical significancy arbitrary? No.  End of story.

You must be logged in to read this quote.

Is it O+2? Yes. Now, you're arguing from authority.

You must be logged in to read this quote.

In order to prove that you've got a coin that is not fair - you have to do a statistically significant number of trials and show that the deviation from true is statistically significant.

What you've done here, is take a statistically significant number of trials, and found a result that was statistically insignificant. Ergo - we can conclude that the coin is fair.

You must be logged in to read this quote.

Not unless I can empty the tank and count the contents. Just the 4 and 2 is insufficient information.

You must be logged in to read this quote.

Earlier up in the thread, which is why I had to clarify it YET AGAIN. I'm tired of having to clarify my position.

You must be logged in to read this quote.

Yes, it is zero correlation.

I'm not going to answer any more posts from you until you concede this point.

[Wisconsin+17]

Also, didn't I already say:

You must be logged in to read this quote.

If you want to talk to yourself go right ahead. If you want to actually listen, feel free to ask me any of these questions.

You must be logged in to read this quote.

Nope. Never did. You've been persuing this non sequitor the entire thread.

You must be logged in to read this quote.

Yes. You're the one equating the two arguments. Adoption is all about outcomes - what is in the best interests of the child. Marriage is about the common law.

[Alcon]

Yawn, you're only arguing this in order to salvage your position. Rather than actually change your position in the face of contrary data, you're arguing over what "is" really means.

Is it statistically significant? No. Is statistical significancy arbitrary? No.  End of story.

No, I'm arguing over how statistics work.  And you're not responding to my argument.  Seriously, you're the worst.

Is it O+2? Yes. Now, you're arguing from authority.

Obama +2 over what?  I don't even understand how that usage would make sense.  Moreover, no one uses it.  This isn't an appeal to authority.  This is an appeal to definition.  You presumed I meant an unspecified, arcane definition for no particular reason.  This is especially ironic considering that you just accused me of arguing over the meaning of "is."

In order to prove that you've got a coin that is not fair - you have to do a statistically significant number of trials and show that the deviation from true is statistically significant.

...

Not unless I can empty the tank and count the contents. Just the 4 and 2 is insufficient information.

Yes, so if you get a 4-2 flip on a coin, you can't be 95% or more confident that it's not a fair coin.  You cannot be 95% or more confident that the next flip is likelier to be heads than tails.  However, you can be between 50% and 95% confident that the next flip will be heads.  As you flip more times, if the 2:1 heads patterns sustains, the confidence gradually will shift up from <95% to >95%.  However, in the interim -- because statistical significance is a continuum -- you will have a point where you are about 60% confident, about 70% confident, and so on.  You are erring by treating this as a 50/50 proposition until you hit 95%. 

In simple terms:  The question is not whether you reject the null hypothesis of the coin being 50/50.  That is not required to hold that the next flip is empirically more likely to be heads than tails.  Think of it as a binomial distribution.  Say you flip a coin 50 times.  Thirty flips come back heads.  Your confidence interval is .451 to .736.  In other words, it's not statistically significant.  Does that mean a prediction that the next flip will be heads is equally reasonable to a prediction that it will bet tails?  No; the confidence interval (a bell curve, I think) is much more toward the "likelier heads" side than the "likelier tails" side -- in fact, 50/50 is barely within the confidence interval.  Therefore, even though you can't reject the null hypothesis that the coin is 50/50, it's more reasonable to predict heads than tails.  In other words, although the observed probability that the coin is unfair may be well under 50%, the observed probability that the next flip will be "heads" may not be equal to the observed probability that it be tails.

Seriously, print this explanation out, take it to a Stats prof, and tell him to email me if I'm in error.  It can be a simple Yes/No.  I can provide my email address.

Earlier up in the thread, which is why I had to clarify it YET AGAIN. I'm tired of having to clarify my position.

Haha, I asked you when I said something and you respond "earlier."  Thanks. 

Yes, it [statistically insignificant] is zero correlation.

Alternatively, email this sentence to a statistics professor and see what he thinks.  I really wonder about all of those times I ran statistical significance tests on an observed correlation; I never noticed that the correlation reset to zero after I ran the statistical significance test.  I wonder what the hell "weak correlations" were, since those didn't always reach statistical significance.  So much of my life is being turned upside down!

I'm not going to answer any more posts from you until you concede this point.

lol

[Brittain33]

Alcon is doing the Lord's work here but I kind of miss when this thread was all about Romney's lack of empathy.

[Wisconsin+17]

You must be logged in to read this quote.

O+2 R-2. Think vector notation. Romney needs to gain 2 points to tie Obama, since a gain for Romney will be the same as a loss for Obama.

You must be logged in to read this quote.

And yet, you won't accept an appeal to definition when it concerns statistical significance.

You must be logged in to read this quote.

So you're willing to be 20:1 odds that the next flip is heads?

You must be logged in to read this quote.

This is why there's no correlation. It is, as I stated before, a mathematical equivalency. Not arbitrary whatsoever and is fundamentally connected with sample size.

You must be logged in to read this quote.

Weak correlation is when you have multiple studies, some with statistical significance, and some without.

Glad I could help you pass your stats class!

[Alcon]

O+2 R-2. Think vector notation. Romney needs to gain 2 points to tie Obama, since a gain for Romney will be the same as a loss for Obama.

Wrong.  Did you not notice how Obama and Romney don't add up to 100%?  A gain for Romney could be a loss for Undecided/Other.  I clearly gave an example where this was possible.  Again, I've never seen anyone present polling results in the format you're claiming you assumed I meant.

And yet, you won't accept an appeal to definition when it concerns statistical significance.

Wrong.  I've never objected to defining statistical significance.  I objected to the equivocation you are making between statistically insignificant

This is why there's no correlation. It is, as I stated before, a mathematical equivalency. Not arbitrary whatsoever and is fundamentally connected with sample size.

wrong.  there's no correlation because it's a binomial distribution? what?  binomial distributions don't have to be 50/50, they just have to entail two options.

So you're willing to be 20:1 odds that the next flip is heads?

Wrong.  How bad at this are you?  Why would I be willing to bet that the probability of the coin flip being heads is .952?  You've bizarrely substituted the standard statistical significance confidence level for the observed probability (and messed that up too; it's 19:1.)  I don't even.

Let's put aside the coin example, because an unfair coin seems counterintuitive.  Imagine that they're ballots from an unspecified voting precinct from a tied election like 2000.  Let's pretend the only candidates were Bush and Gore.  If I pulled 20 random ballots from that precinct, and 14 were for Bush and 6 were for Gore, my confidence interval would be 48.09% to 91.91% Bush.  As you can see, 50% Bush is within the confidence interval there.  You cannot (with statistical significance) reject the null hypothesis that the precinct is 50/50.  However, the vast majority of the confidence interval is to the "Bush" side.  Therefore, if we picked another random ballot, it is more reasonable to expect it to be a Bush vote than a Gore vote, even though we can't reject the null hypothesis that Bush and Gore were even in the precinct.  Get it now?

Weak correlation is when you have multiple studies, some with statistical significance, and some without.

wrong. "weak correlation" can exist within a study.  query "weak correlation" (in the texts, not the titles) in JSTOR or Google Scholar or whatever and you'll see plenty of examples.

Glad I could help you pass your stats class!

[Alcon]

Alcon is doing the Lord's work here but I kind of miss when this thread was all about Romney's lack of empathy.

Sorry.  If it's any consolation, this exchange is helping to show that stone-hearted antipathy is a pretty reasonable response to the world.