Originally posted by ALightinDarkness
No, a simple correlation does not signify any sort of relationship. The number of sales by Apple Computer, Inc. per month and the number of domestic
dogs in the United States per month has also gone up. Put it in SPSS and run a correlation, you'll find a surprisingly high value. Surely, this means
we have causation!
Many things correlate and theres no causation going on.
Yes, of course. However, is there a logical reason why we might expect sales by apple computers to be related to the number of dogs? If not, I would
suggest it was most likely spurious finding.
However, you are taking the 'correlation does not imply causation' mantra incorrectly. It really means correlation is not sufficient to claim cause.
It is often suggestive
of one, it is certainly necessary
for one, no? A tentative and suggestive first step.
Is there a logical reason to suggest that intelligence and belief/non-belief might be related. Why, yes, I think there is. I'm sure the introduction
to the article will attempt to justify such a claim. Should be an interesting read.
Cum hoc ego propter hoc applies is you want to logically claim the correlation necessarily means a causes b is true. I'm not, I'm saying it is
suggestive. A hint. A tentative indication
And there is absolutely now't wrong with that.
I think that extraordinary research claims require extraordinary research methods. If your going to make a grandiose claim and don't use a
research technique that is most appropriate for what your doing, then it should be no big surprise you get rightly criticized for it and have the
validity of your study be questioned.
I'm not sure it is so 'extraordinary'. The research technique of bivariate correlation is actually fine, there are just more thorough and expansive
I've been a graduate student for a few years, and worked on research all that time. I come across high R squares all the time, especially when
people are making grandiose claims. It is not the norm for your average paper, but it is expected when your making huge claims like this.
They are not making 'huge' claims. They are saying that belief/non-belief can predict intelligence (or more correctly, IQ). We can quibble over the
IQ = intelligence point, but it is actually fairly well accepted in their field.
Spurious correlations are just matters of fact when dealing with this sort of research. Your eagerness to accept it and make it seem more than
it is is very troubling to me, especially for someone who is throwing subliminal insults like this.
Heh, I not that willing to accept it. My first post in this thread was pointing out some relevant weaknesses and issues.
I really just want to read it in detail, then I'll make my mind up what I really think about it.
As a independent researcher (I think publishing my own papers off my own original data makes me independent)
Hmmm, not in my mind if you have a PI, but whatever. Not really that relevant, wasn't meant to be another touchy subject.
Just meant to illustrate that you sometimes have to use what you can get your grubby mitts on.
I use what I have - and I know my limitations. I've went down possible research paths many times only to have to back up and reframe because I
was going down a path that would lead to claims I could not back up. Must be all that humility and ethics I have getting in the way.
Of course! No research is perfect. We use what we have available, in the time available, with the funds available, and do the best we can in those
They can back up their claims though. They did a study, somehow measured belief/non-belief and IQ, performed appropriate stats, and found a strong and
significant relationship. They claim that one variable appears to predict the other.
That's a fair inference in normal science.
But it is. And I've run across those scores many times and had them be not statistically significant at that level.
I take your comments of being so sure of the peer reviewed process as pretty poor form. Getting into some journals is a matter of politicking,
knowing the editor, and making sure you throw in cites that are from the reviewer list. The "peer review" process has little meaning, except for a
certain tier of journals and higher.
Since your displaying quite a level of subliminal arrogance here (or at least that is what it appears to be), I'm not sure how to respond.
I'll just say I work with a expert in quantitative methodology in the social sciences on a daily basis. I will trust her PhD training from a top
ranked program and 35 years of research experience in this area - which go completely against your claims.
Well, what do you want me to say? I know that is the general situation. It's certainly the one I've worked to since whenever. Moderate correlations
are viewed around .3-.4 and stronger ones above .5. You appear to think this needs a conference or something, just general stuff.
Sorry for knowing that.
ABE: Here, here's another academic who knows that:
Two main factors make Dr. Cartman more or less willing to conclude that there is a non-zero correlation between SAT and GPA in the entire Freshman
student body. One factor affecting his confidence is the size of the correlation in his sample. In his sample’s data, Dr. Cartman found a
correlation of r = .40, which represents a positive correlation of moderate size.
But what if Dr. Cartman had found that the correlation in sample was very strong, say r = .80? A correlation of r = .80 is very far from zero –
it expresses a very strong association between two variables.
Pass it around, it's a useful rule of thumb.
Yet again: correlation does not equal causation. Just because its probably not a random correlation in terms of statistical significance does
not mean there is any causation going on. It is most likely spurious.
And I never said that correlation does equal causation. See earlier.
Both. IQ is not the best measure of intelligence, and it could be combined with other variables if there was no agenda going on. Also,
measuring belief is complex. Measuring belief as a dichotomous variable not only has lots of validity problems but I would bet there was no logit
regression to make up for it if it was measured that way. Given the simplistic methodology I would be surprised if they used any sort of index to
measure belief appropriately by asking different questions about belief in god, religious practices, etc.
OK. Until I bother getting the study (I'll U2U you it if you want it), we can only really guess. I have an idea what they did (knowing previous work
from these dudes), and it wouldn't be the best approach, IMHO.
This has become circular. Your going to claim I'm dismissing it, and I'm going to claim your making more out of it than it really is. We
agree measuring IQ is probably not the best way to do it, and the methods are probably too simple, and some others things. I would leave it at that
until something new comes up.
I do agree somewhat. I suppose the problem is just how do we measure intelligence? People in that field are generally happy to use IQ as some sort of
index, it does have predictive validity. So it is probably the best we do have. But like you, I'm not so sure it equivalent to Intelligence (TM).
I'm not actually making that much out of it, IMO. At most, I've said, if methodologically robust, it could well be suggestive of some form of causal
But I'm not exactly running around calling theists dopes or anything.
[edit on 13-6-2008 by melatonin]