Right. I'm not confident, either. Hence, ellipses and a question mark. The problem for this is that the context is unclear. I have no idea what other studies of this type have been done and what sorts of results they're ending up with. And maybe because I've been exposed to much more "pure science" type of research, the gut reaction is that the number seem "too low."

But that reflex might be wrong.

To clarify my comment about skepticism a bit more:

The bolded is really the emphasis for me. The first half is pretty much a response I would have for studies of this type, regardless of the outcome, at least until further/similar studies have been performed with similar results. But with regards to the bolded, I think I've stated several times that I have a lot of misgivings towards science reporting because of the tendency for sensationalism. (For example, the title of the article linked by OP.)

Just FYI, they did take in data for maternal education as an indicator of socioeconomic status.

It's an interesting point to raise because the original dictator sticker game study (http://www.emiliocuilty.com/uploads/...5/children.pdf) indicates that socioeconomic status has an influence on the outcome of the result. Given that religiosity is (negatively) correlated with economics, it does open the question of the role that it played here.

Given the list of countries (Canada, China, Jordan, Turkey, USA, and South Africa), I would have liked to see more data on the distribution by geography. For example, only 18 people (out of 1170) reported to be Buddhist but 510 reported to be Muslim. That's not exactly conveying a sense of a good multi-cultural distribution, and suggests that Jordan/Turkey are possibly overrepresented in the overall sample. It could well be that geographical cultural factors are dominating and not religious cultural factors.

That was also my complaint. "Complaint" is possibly not the right word. I think that the study is fine but should be interpreted fairly cautiously because there are obvious limitations. That's not unusual though.

Both socio-economic status and country of origin are included in the regression analysis and thus controlled for.

The variance they can explain would not be attributed towards religiosity of the household.

This is what I wrote before:

Did I just miss it? I didn't see regression on those variables for that specific question. But I also thought (and perhaps I'm just bad at understanding stats) that the explanation of the regression analysis for generosity found that SES was about as significant a predictor as religiosity, although country of origin was less so. And the stats don't really let you see how the variance is distributed, which is why I was curious to see results by country.

ANCOVA combines ANOVA and regression and is typically used to control for effects of other variables ("covariates").

The entire thing is pretty densely written, so it is easy to miss stuff.

I would have liked to see more discussion of it. There's a bit of black-box-ness going on (which I acknowledge is not uncommon) because there are lots of ways to try to control for various variables.

The most common way of doing this sort of thing for numerical data is to just have a whole bunch of variables in a linear combination. So for this study it would be something like

Altruism = A * Age + R * Religiosity + S * SocioeconomicStatus

And then you run the linear regression on the data set and get the coefficients A, R, and S. And, at least in this case, they would have "controlled for socioeconomic status" because you have an equation where you can hold that variable fixed and look at how age and religiosity influence the shape of the graph.

But that isn't what they seem to have done, because they have a graph of just alturism, age, and religiosity, and there's not information about the specific socioeconomic level they're looking at. But they list N = 1170 on that graph. So where is that data? Did they do some sort of projection of the data onto the three dimensional Altruism/Age/Religiosity subspace, possibly by just ignoring the SocioeconomicStatus variable? Because that would be a very bad (and potentially misleading) way to present the data.

Here, we have that they are dummy coding the categorical data. The charitable reading of this is that they've dummy coded both the religious identification and the country of origin. (I assume they did, because to code country of origin data without dummy coding would be really bad.)

But did they do any analysis of interaction effects between the two classes of categorical data? I didn't see anything in the paper about that. It seems to me that country of origin and religious identification are going to be variables that are related, and there might be something else there.

I'm not saying that they necessarily made mistakes in the analysis. Maybe they did it but didn't report that aspect of it. I do know that a lot of people use statistical tools in the wrong way, and that the more complicated the data they're working with the more ways there are of screwing it up, and the more careful they need to be.

And so that's why when I see N=18 Buddhism and N=510 Muslim, it gives me pause. I see a high potential for overrepresentation in one country, and it may have influenced the regression, especially because it's not entirely clear what their controlling mechanism was. There are just a whole lot of pitfalls here.

-----

Here's a link about the interaction of dummy coded variables. It may be too mathy for most, so I've just snipped it up. The data set has three variables: job prestige (numberical), marital status (categorical), and gender (categorical). And then it proceeds to do an analysis

http://www.theanalysisfactor.com/int...ar-regression/

In other words, even though the original analysis showed no gender in job prestige, there is one. But you would have to parse the data by marital status in order to see it, and you can't see it if the two different categorical variables were left independent of each other.

I had indeed overlooked that they say they took those factors into account with the ANCOVA. What I noticed is that they don't report anything like the β/p values for the other factors (SES, origin, age...) in that case. I'm not very familiar with ANCOVA so maybe there just isn't an equivalent.

I also am not sure what exactly it means to say that the ANCOVA revealed a "main effect" of religious identification, if it means something about the comparison between religiosity as a factor and the other variables. I read it the first time as just meaning that religiosity was significant, but maybe it means more than that?

Let's say you are looking at a link between bad roads and car accidents. You decide to explore 3 effects.

1. Is there a difference in severity of the crash based on how bad the road is?
2. Is there a difference in severity of the crash based on how fast the person drives?
3. Is there an interaction between how bad the road is and how fast the person drives that influences the severity of the crash?

The first two are main effects, their hypothetical influence on the severity of the crash is theirs alone. The last is an interaction effect, something that makes the independent variables statistically influence each-other is effecting the independent variable.

The example also shows the (hypothetical) value of such analyses, by (potentially) giving us more relevant findings to the problem we are trying to solve or the question we are posing.

As always it is important to remember that we are working within our model. A statistical model can work perfectly, but be completely useless for saying anything about the world. You could make a model that said that inside every conservative's head lived a leprechaun which made them vote Republican, and you would find overwhelming evidence for it if you only applied inferential statistics. Incidentally, that example also shows why very high correlations can be indicative of a problem with your model in social science.

I think I understand. Thanks for the explanation.

Hehe, you seem to have a good grasp on these things.

They did do some multivariate analyses where they claim to have controlled for these things.

Though, the problem with that is even deeper. Analyses like that are mostly meant for applications in hard science or economics, and make a lot of assumptions about the data-set that has to be in place.. making the instruments (surveys) used in social science often be more about ensuring proper distributions than about real world applications.

All that said though, it is very good that they used actual behaviors as the independent variables for the main hypotheses. And though an r ranging from .17 to .2 might seem low, it's pretty decent when it comes to suggesting links to morality. And since it is children, it is not an unfair assumption that the morality stems from the religion and not vice versa.

I think my main criticism would be the use of sensationalist terms like "meaner" and "less kind". They aren't completely out of place, but the everyday meaning of the words are (to me at least) a lot harsher than the study suggests, they give the impression of lack of morality, which is not really supported at all.

If they had simply stuck to questioning if lack of religion equaled suspect morality, which is the claimed basis for the study, I think the overall impression would have been better (to my eyes).

That seems fair to me.

The "Dictator Game" used is considered to be an economic study which purports to prove that everything which a person does economically is not self serving. It has been transferred to the question of whether children are more or less altruistic

http://www.ehbonline.org/article/S10...fulltext?cc=y=

Jean Decety:

https://en.wikipedia.org/wiki/Jean_Decety

An interesting thing about Professor Decety is that he states that "empathy" can and does lead to "immoral" behavior, not only the "moral".

The considerations I have perceived concerning "empathy" which is akin to "compassion" is that one "lives within" the being of another and this can only be accomplished by an act "self sacrifice" to which to which we are all capable.

In order to "empathize" or display "compassion" our very being must be "given up" in relation to another and to wit we "experience" the woes or tribulations of our beloved. In order to do this a "warmth of being" immersed in "Love" is brought forth by the compassionate.

Our needs, wants, desires, proclivities, etc...are sacrificed on the alter of "love" in the act of the compassionate or empathetic being. He speaks to us and we are the recipients of the pathos of another.

We are all capable of doing as such but not so clear as to what we are doing. A similar activity is the experiencing of a fine piece of music in which we become immersed in a Beethoven Sonata or a Pachelbel Canon. It is here where life enters into beauty the picture of Love; beyond the logic to which we apply.

https://www.youtube.com/watch?v=JvNQLJ1_HQ0

This is Religion, God and Theology and so I introduce you to a man who brought forth the magnificence of compassion in the healing of his fellow beings, Francis of Assisi, a youthful warrior who turned about and brought compassion to life.

To echo others, a low R value isn't surprising really. We want to know whether the effect of some independent variable on a dependent variable is a) real and b) important. The p value tells you if we have reason to believe its real, and the R value tells you whether it is important. As an example, if we want to know how different IVs affect scores on exams, we might find that diet, hours of sleep the night before, number of hours studied, socio-economic status, etc etc all have real effects, but some will explain more or less of the variance. If I were to guess, in this example number of hours studied would probably have a fairly decent R value, with the others having low R values.

That is well summarized, but it is also prudent to remember that these tests on the data are merely checking that the statistical model is holding up, not that the model accurately reflects the world.

Like I said in an earlier example, we could make a model where leprechauns lived inside conservative people's heads and influenced them to vote Republican, and statistical analysis would find overwhelming evidence for this model - but the assumptions behind the model are bogus.

I'm not saying that I doubt the accuracy of the proposed model in this paper, I think stating a causal relationship from religiosity to morals is a good model, I'm just pointing out basic limitations of tests of inference in statistics.

Agree with all this.

This article will only affirm 1. atheists who believe that christianity is what you claim it is, and 2. Christians that the world is of the devil and fallen human nature and anything those things write about christianity that contradicts the bible is false.

This just adds evidence that the two worldviews are diametrically opposed and there will never be a consensus. (interestingly, the bible says this about itself)