Thanks, osso.
stlstrike3 wrote:JPB wrote:
Reading... what, exactly, are you trying to discern?
Whether or not education and/or intelligence has anything to do with someone is more/less likely to be religious.
Haven't read any further yet, but some general observations on the study design and your question.
1) Correlation does not infer causation. The result that those who self-identify as religious correlate to significantly lower QSAT scores might, or might not have a cause-effect relationship.
2) Correlation coefficients are biased when a subset of the overall population range is used. The methods and summary statistics indicate that the study subjects are 17 - 24 years old, with above average intelligence when measured as a function of both their SAT scores and their IQ scores. The correlations generated would not necessarily mirror those of the general population when looking at wider ranges of age and intelligence.
3) I'm concerned that he "computed Pearson correlation coefficients on numerous pairs of continuous variables" on what is largely ordinal data. The footer for Table 2 does not appear in my copy of the paper, but I'm assuming that * infers statistical significance (perhaps the * is missing from the title of the table and "all = p<0.05" should read "* = p<0.05" in the title). The number of comparisons made would require control of the overall Type I error. There are a total of 45 comparisons, each with a significance standard of 0.05. Without actual p-values we can't tell whether they would still be significant if controlling for overall error. I should note that with the exception of "How often do you read other religious literature" the significant results are all on continuous variables.
4)The significant difference between religiosity and QSAT scores was primarily due to the religious vs atheist comparison. The atheist group represents only 8% of the study participants, or 6 individuals, compared to the religious group which represents 48% of the subjects, or 37 individuals. Although the numbers probably closely reflect the percentage of college students in these groups, I would prefer to see a more balanced study design.
That's it for now. I haven't spent much time on the "Discussion" but will do so later.
I'm also intrigued by something in the other abstract (did you buy it yet
)
Quote:In the United States, religious attendance rises sharply with education across individuals, but religious attendance declines sharply with education across denominations.
This puzzle is explained if education both increases the returns to social connection and reduces the extent of religious belief. The positive effect of education on sociability explains the positive education-religion relationship. The negative effect of education on religious belief causes more educated individuals to sort into less fervent religions, which explains the negative relationship between education and religion across denominations. Cross-country differences in the impact of education on religious belief can explain the large cross-country variation in the education-religion connection. These cross-country differences in the education-belief relationship can be explained by political factors (such as communism) which lead some countries to use state-controlled education to discredit religion.
source
I wonder if the study tries to answer the "if" question posed. That's really what I think you're trying to get your hands around.... does education increase the returns to social connection but reduce the extent of religious belief in a linear relationship?