I would like to thank everyone who has completed my survey on natural theological arguments. This survey's aim was to get a rough idea on how philosophers today evaluate various natural theological arguments in terms of their strength/plausibility. My study was motivated by the observation that philosophers frequently voice intuitions about the general plausibility of natural theological arguments, e.g., "since Darwin, the argument from design has lost its appeal", or "the hiddenness argument is a strong contender to the problem of evil as an argument against the existence of God." However, actual data on philosophers' assessments of these arguments was, to my knowledge, unavailable. I'm very pleased with the large sample (802 respondents!). The data will be used in a monograph I am currently writing on the cognitive basis of natural theology.
Descriptive statistics about the sample
- Respondents (N = 802) were recruited through a philosophy mailing list and several philosophy blogs
- Average age: 36.5 years (SD = 11.8 years)
- Gender: 75.8 % were men and 24.2 % were women. This is a gender imbalance, but it is not out of line with other philosophy surveys, and may reflect the general gender imbalance of philosophy.
- Religious self-identification: 40.5 % theists, 40.4 % atheists, 19.1 % agnostic or undecided (I'll refer to this group as agnostic for short, realizing that not all agnostics see themselves as undecided).
- Target group: 85.8 % of respondents self-identified as philosophers; the remaining 14.2% did not (the real percentage may be higher, as some respondents said they had some training in philosophy at the undergraduate or graduate level, but moved on to major in other fields).
- AOS: The most mentioned philosophical specialization was philosophy of religion (33.8 %). The other most mentioned areas of specialization were, in descending order, metaphysics (27.8 %), ethics (26.8 %), epistemology (25.8 %), history of philosophy (22.2 %) philosophy of mind (19.2 %) - The total is more than 100 % because respondents could indicate multiple AOS
- Academic position: graduate students (33.3 %), faculty including tenure track (32.9), non-tenure track with PhD (15.8%), undergraduates (8 %), non-academics (10 %).
I used a linear model to examine which factors (age, religious self-identification, gender, AOS, philosopher or not, academic position) might influence the overall assessment of natural theological arguments. As we will see below, the only statistically significant independent variable for the assessment of the natural theological arguments turned out to be religious belief. Whether or not one is a theist, atheist or agnostic influences to a very significant extent how one evaluates the arguments (more on this below). Other seemingly significant independent variables like philosophical specialization were no longer significant when controlling for religious self-identification.
How strong do philosophers rate natural philosophical arguments?
Respondents were asked to rate how strong they found a series of natural theological arguments, on a likert scale of 1 (very weak) to 5 (very strong). They also had the option of indicating they were not sufficiently familiar with the argument in question. Arguments were organized in two groups (arguments for and arguments against the existence of God) of 8 items each. The results are summarized in figures 1 and 2, grouped in arguments for and arguments against.
Overall, the strongest natural theological argument was the argument from evil (mean score: 3.55). The second strongest rated argument was also an argument against God, namely the argument from parsimony (mean score: 3.01), followed by a positive argument, the cosmological argument (mean score: 2.984), and the argument from lack of evidence (mean score: 2.855). Least popular overall were the argument from miracles and the argument from beauty.
Update: Following Mohan Matthen's observation at NewApps, I thought it would be a good idea to include a figure where you can see the mean score of each argument, viz. the mean score with theists, atheists and agnostics lumped together. This picture indicates that arguments against the existence of God (the second half of the bars) are overall rated somewhat highly than arguments for the existence of God. The boxplots on Figs 3 and 4 show what is going on here: atheists rate arguments for lower than theists rate arguments against.
Fig. 0: Overall mean scores of arguments for the existence of God (first 8 bars) and arguments against the existence of God (next 8 bars) - lumping together theists, atheists and agnostics
Fig. 1: mean scores for arguments for the existence of God
Fig. 2: mean scores for arguments against the existence of God
Do theists, atheists and agnostics significantly differ in how they assess arguments for and arguments against the existence of God? To assess this, I computed a total overall score for arguments in favor and one for arguments against. This overall score was not normally distributed, so I used non-parametric analyses (Kruskal Wallis test, with post hoc Mann Whitney U) to assess the differences. However, parametric tests (ANOVA and Scheffé) yielded very similar results and an identical significance level. In the light of this, and also given that ANOVA is reasonably robust against deviations to normality, I will report the parametric results here.
Arguments for the existence of God
Theists, agnostics and atheists differed significantly in their overall assessment of arguments for the existence of God. F (2, 684) = 457.9, p < .001.
Posthoc tests (Scheffé) reveal all these differences to be significant. Theists were more positive in their evaluation of arguments for the existence of God than atheists. Not only is there a significant difference between theists and atheists, but agnostics occupy their unique position, in between both. All tests are at a p level of <.001.
Fig. 3: mean total scores for arguments for the existence of God for theists, atheists and agnostics
Arguments against the existence of God
Theists, agnostics and atheists differed significantly in their overall assessment of arguments for the existence of God. F(2, 632) = 102,98, p < .001.
Posthoc (Scheffé) tests again suggest that these differences are all significant. This time, atheists are significantly more positive in their assessment of arguments against the existence of God, whereas theists were least compelled. As with the arguments in favor, agnostics occupy their position in between, being more positive than theists, but less than atheists.
In how far do atheists and theists differ in their assessment of individual arguments?
The largest disagreement is about the cosmological argument: theists find this a very strong argument (mean = 3.92), atheists assess it as significantly weaker (mean = 2.48). For arguments against God, the largest disagreement is on the argument from lack of evidence. Whereas atheists clearly hold that there is a lack of evidence (mean = 3.77), theists disagree (mean = 2.17).
There is most agreement about the argument from evil, which is assessed as strong by both theists (mean = 3.49) and atheists (mean = 3.73); also, both theists and atheists agree on the strength of the hiddenness argument (theists mean = 2.69; atheist mean = 3.12)
Although I found no statistically significant differences in how people assess the arguments according to gender, there were other interesting gender differences.
Males are disproportionately represented in philosophy of religion: of the philosophers of religion in the sample, 90.4% self-identified as male, and 9.6% self-identified as female. This disparity is higher than one would expect given the gender imbalance in the sample, Chi2(1, N = 802) = 47.5, p < .001.
Women in the sample were more likely to be atheists and agnostics Chi2(2, N = 802) = 45.4, p < .001.
- Among the female respondents, 45.9 % were atheists, 32 % agnostic and 22.2 % theists.
- Among the male respondents, 46.4% were theists, 38.7 % atheists and 15 % agnostics. [there was a typo before here - thanks to the one who pointed this out to me]
This may be due to sampling. Speculatively, since the data collection does not allow me to test this hypothesis, it seems to me that this is due to a high percentage of theists who are overwhelmingly male at Prosblogion, and a high percentage of atheists at feminist philosophers who are female. But that's just pure speculation.
Philosophy of religion and theism
This is an update, included as a result of a question by Alexander Pruss. Like in the PhilPaper survey, I noticed a strong correlation between theism and philosophy of religion. Of those who listed philosophy of religion as an AOS, 73.1% were theists, 17% were atheists and 10% were agnostics.
Of those who did not include PoR as an AOS, 23.9% were theists, 52.4% were atheists and 23.7% were agnostic/undecided.
This result is highly significant, Chi2(2, N = 802) = 180,34, p < .001.
The results indicate that theists, atheists and agnostics significantly differ in their assessment of the strength of natural theological arguments. Does this mean that one's religious beliefs already strongly influence one's assessment of arguments, as Jennifer Faust suggested? Faust holds that religious arguments are question-begging in a doxastic sense, i.e., whether or not one accepts the premises of an argument depends to a large extent in one's antecedent beliefs in the conclusion (which is either 'God exists' or 'God doesn't exist). So, for instance, an atheist when confronted with a particular version of the cosmological argument will intuit there is something fishy with one of the premises (she may question, for instance, whether or not whatever begins to exist has a cause for its existence). An alternative explanation is that the respondents were influenced by particular natural theological arguments in the shaping of their religious beliefs. This seems a bit less plausible to me, however. It is also conceivable that people change their minds on the basis of arguments. The argument from evil, rated as strong by all groups, may lead a theist to doubt his beliefs and to adopt an agnostic or atheistic stance. This short survey does not allow me to decide which of these explanations is more likely.
If you've read this far, you might also want to read some methodological notes on the stimuli: many respondents argued in the optional comments form that pragmatic arguments are not strictly speaking arguments for/against the existence of God, but rather for/against belief in God. Since there are both a pragmatic arguments for theism and one for atheism, this is not problematic for the statistics presented below. Another frequently voiced concern was that the arguments are coarse grained. Many respondents proposed that they were unimpressed with the Anselmian ontological argument, but thought the modal ontological argument was strong. Similarly, the fine tuning argument was judged to be a lot stronger than its classical Paleyan predecessor. My main reason for not differentiating was to keep the survey accessible, also to non-specialists (I'm confident not many non-philosophers of religion would know about the modal ontological argument). Finally, the arguments were presented as arguments for/against God, but I did not specify which concept of God (pantheism, classical theism, deism, etc.). This was done for simplicity's sake. Allowing participants to choose between concepts of God would have made the statistics harder to interpret and less robust, and would also made the survey less accessible for non-specialists. I should note there is a survey on alternative concepts of God.