Do philosophers of religion rate religious arguments differently from other philosophers? (short answer: yes)
May 23, 2012 — 16:30

Author: Helen De Cruz  Category: Existence of God  Tags: , , , ,   Comments: 5

As promised, here is the second fine-grained analysis of the results of my survey. The analyses have been done by Robert O’Brien, a medical statistician from Miami University. The statistics are fairly technical, and below this short summary you can find the complete statistical analysis.

Here I report how philosophers rate the arguments against theism in my survey. I presented 8 arguments against theism (see here for an overview of the arguments and general info on the survey) and asked participants to rate how strong they found them, ranging from very weak to very strong.

What I was interested in is how religious belief (theism/atheism/agnosticism) affects the assessment of these arguments. Initially, pooling all arguments together, it seemed like PoR had little effect, but when considering each argument separately, it turns out that PoR does influence the assessment of individual arguments. There were also gender effects.


Strikingly, the argument from divine hiddenness is rated a lot stronger by PoRs than non-PoRs (this is controlling for gender and religious belief). The line chart shows the difference well. By contrast, PoRs rated the argument from parsimony weaker than non-PoRs (see graph 2)
graphs_hiddenness.jpg
Graphs_parsimony.jpg
At first, I thought that my survey showed mainly effects of confirmation bias (i.e., theists rate arguments for theism higher, atheists rate arguments against theism higher), but thanks to Robert’s more fine-grained analysis of individual argument I can see now that PoR does make a difference in how a lot of these arguments are assessed (the main predictor is still religious belief though, but controlling for this, PoR makes a difference). If PoRs rate the hiddenness argument stronger than the general philosophical population, does that say anything about hiddenness as a problem for theism? Or, conversely, since PoRs are not as impressed with the argument from parsimony, should the (non-PoR) atheist look for other reasons to support her beliefs (I’m saying this in particular because parsimony came out as a favorite alongside the argument from evil).
Below you can find the full analysis. Robert used a cumulative logit model without proportional odds, except for the argument from lack of evidence, where he used the cumulative logit model with proportional odds. These models are the appropriate ones, because we cannot assume that the psychological distance between the different categories (“very weak” vs “weak” etc) is always constant – so we can’t treat the dependent variables as interval variables (which would make the statistics a lot easier!).

Full statistical analysis
Argument from inconsistent revelations
In this sample, atheists are 3.09 times as likely as theists to rate the argument from inconsistent revelations as “weak” vs. “very weak;” are 3.91 times as likely to rate the argument as “neutral” vs. “weak” or “very weak;” are 6.6 times as likely to rate the argument as “strong” vs. “neutral,” “weak” or “very weak;” and are 10.4 times as likely as theists to rate the argument from inconsistent revelations as “very strong” vs. “strong,” “neutral,” “weak” or “very weak” (p-value < 0.0001 for all statements). In this sample, agnostics are 2.22 times as likely as theists to rate the argument from inconsistent revelations as "weak" vs. "very weak" (p-value = 0.002); are 2.2 times as likely to rate the argument as "neutral" vs. "weak" or "very weak" (p-value = 0.0005); are 2.66 times as likely to rate the argument as "strong" vs. "neutral," "weak" or "very weak" (p-value = 0.0007); and are 3.43 times as likely as theists to rate the argument from inconsistent revelations as "very strong" vs. "strong," "neutral," "weak" or "very weak" (p-value = 0.03). In this sample, atheists are 1.78 times as likely as agnostics to rate the argument as "neutral" vs. "weak" or "very weak" (p-value = 0.006); are 2.48 times as likely to rate the argument as "strong" vs. "neutral," "weak" or "very weak" (p-value < 0.0001); and are 3.04 times as likely as agnostics to rate the argument from inconsistent revelations as "very strong" vs. "strong," "neutral," "weak" or "very weak" (p-value = 0.004). There is no statistically significant difference between atheists and agnostics with respect to rating the argument from inconsistent revelations as "weak" vs. "very weak" (p-value = 0.20). In this sample, females are 1.53 times as likely as males to rate the argument as "neutral" vs. "weak" or "very weak" (p-value = 0.02); and are 1.53 times as likely to rate the argument as "strong" vs. "neutral," "weak" or "very weak" (p-value = 0.03). There is no statistically significant difference between females and males with respect to either rating the argument from inconsistent revelations as "weak" vs. "very weak" (p-value = 0.62) or "very strong" vs. "strong," "neutral," "weak" or "very weak" (p-value = 0.29) In this sample, there is no statistically significant difference in how philosophers of religion and participants who are not PoR rated the argument from inconsistent revelations (all p-values > 0.09).
Argument from poor design
In this sample, atheists are 2.64 times as likely as theists to rate the argument from poor design as “weak” vs. “very weak” (p-value < 0.0001); are 4.4 times as likely to rate the argument as "neutral" vs. "weak" or "very weak" (p-value < 0.0001); are 8.43 times as likely to rate the argument as "strong" vs. "neutral," "weak" or "very weak" (p-value < 0.0001); and are 49.3 times as likely as theists to rate the argument from poor design as "very strong" vs. "strong," "neutral," "weak" or "very weak" (p-value = 0.0002). In this sample, agnostics are 1.93 times as likely as theists to rate the argument from poor design as "weak" vs. "very weak" (p-value = 0.01); are 2.4 times as likely to rate the argument as "neutral" vs. "weak" or "very weak" (p-value = 0.0001); are 3.65 times as likely to rate the argument as "strong" vs. "neutral," "weak" or "very weak" (p-value < 0.0001); and are 18.07 times as likely as theists to rate the argument from poor design as "very strong" vs. "strong," "neutral," "weak" or "very weak" (p-value = 0.007). In this sample, atheists are 1.83 times as likely as agnostics to rate the argument from poor design as "neutral" vs. "weak" or "very weak" (p-value = 0.003); are 2.31 times as likely to rate the argument as "strong" vs. "neutral," "weak" or "very weak" (p-value = 0.0003); and are 2.73 times as likely as agnostics to rate the argument from poor design as "very strong" vs. "strong," "neutral," "weak" or "very weak" (p-value = 0.01). There is no statistically significant difference between atheists and agnostics with respect to rating the argument from poor design as "weak" vs. "very weak" (p-value = 0.25). In this sample, males are 1.45 times as likely as females to rate the argument from poor design as "neutral" vs. "weak" or "very weak" (p-value = 0.04). There is no statistically significant difference between males and females with respect to rating the argument from poor design as "weak" vs. "very weak," "strong" vs. "neutral," "weak" or "very weak" or "very strong" vs. "strong," "neutral," "weak" or "very weak" (all three p-values > 0.10)
In this sample, there is no statistically significant difference in how philosophers of religion and participants who are not PoR rated the argument from poor design (all p-values > 0.42).
Argument from evil
In this sample, atheists are 4.01 times as likely as theists to rate the argument from evil as “very strong” vs. “strong,” “neutral,” “weak” or “very weak” (p-value < 0.0001). There is no statistically significant difference between atheists and theists with respect to rating the argument from evil "weak" vs. "very weak," "neutral" vs. "weak" or "very weak," or "strong" vs. "neutral," "weak" or "very weak" (all p-values > 0.11).
In this sample, agnostics are 1.77 times as likely as theists to rate the argument from evil as “very strong” vs. “strong,” “neutral,” “weak” or “very weak” (p-value = 0.02). There is no statistically significant difference between agnostics and theists with respect to rating the argument from evil “weak” vs. “very weak,” “neutral” vs. “weak” or “very weak,” or “strong” vs. “neutral,” “weak” or “very weak” (all p-values > 0.17).
In this sample, atheists are 1.72 times as likely as agnostics to rate the argument from evil as “neutral” vs. “weak” or “very weak” (p-value = 0.02); are 1.59 times as likely to rate the argument as “strong” vs. “neutral,” “weak” or “very weak” (p-value = 0.02); and are 2.26 times as likely as agnostics to rate the argument from evil as “very strong” vs. “strong,” “neutral,” “weak” or “very weak” (p-value = 0.0002). There is no statistically significant difference between atheists and agnostics with respect to rating the argument from evil as “weak” vs. “very weak” (p-value = 0.65).
In this sample, males are 1.91 times as likely as females to rate the argument from evil as “weak” vs. “very weak” (p-value = 0.01); are 1.7 times as likely to rate the argument “neutral” vs. “weak” or “very weak” (p-value = 0.006); and are 1.66 times as likely to rate the argument as “strong” vs. “neutral,” “weak” or “very weak” (p-value = 0.004). There is no statistically significant difference between males and females with respect to rating the argument from evil as “very strong” vs. “strong,” “neutral,” “weak” or “very weak” (p-value = 0.45)
In this sample, there is no statistically significant difference in how philosophers of religion and participants who are not PoR rated the argument from evil (all p-values > 0.11).
Argument from divine hiddenness
In this sample, atheists are 1.98 times as likely as theists to rate the argument from divine hiddenness as “neutral” vs. “weak” or “very weak” (p-value = 0.0005); are 2.28 times as likely to rate the argument as “strong” vs. “neutral,” “weak” or “very weak” (p-value < 0.0001); and are 4.9 times as likely as theists to rate the argument from divine hiddenness as "very strong" vs. "strong," "neutral," "weak" or "very weak" (p-value < 0.0001). There is no statistically significant difference between atheists and theists with respect to rating the argument from divine hiddenness as "weak" vs. "very weak" (p-value = 0.40). In this sample, agnostics are 2.75 times as likely as theists to rate the argument from divine hiddenness as "very strong" vs. "strong," "neutral," "weak" or "very weak" (p-value = 0.02). There is no statistically significant difference between agnostics and theists with respect to rating the argument from divine hiddenness as "weak" vs. "very weak," "neutral" vs. "weak" or "very weak," or "strong" vs. "neutral," "weak" or "very weak" (all p-values > 0.59).
In this sample, there is no statistically significant difference between females and males with respect to how they rated the argument from divine hiddenness (all p-values > 0.23).
In this sample, philosophers of religion are 1.54 times as likely as participants who are not PoR to rate the argument from divine hiddenness as “strong” vs. “neutral,” “weak” or “very weak” (p-value = 0.03). There is no statistically significant difference between PoRs and non-PoRs with respect to rating the argument from divine hiddenness as “weak” vs. “very weak,” “neutral” vs. “weak” or “very weak” or “very strong” vs. “strong,” “neutral,” “weak” or “very weak” (all three p-values > 0.14)
Argument from parsimony
In this sample, atheists are 4.44 times as likely as theists to rate the argument from parsimony as “weak” vs. “very weak;” are 8.33 times as likely to rate the argument as “neutral” vs. “weak” or “very weak;” are 14.14 times as likely to rate the argument as “strong” vs. “neutral,” “weak” or “very weak;” and are 28.84 times as likely as theists to rate the argument from parsimony as “very strong” vs. “strong,” “neutral,” “weak” or “very weak” (p-value < 0.0001 for all statements). In this sample, agnostics are 3.99 times as likely as theists to rate the argument from parsimony as "weak" vs. "very weak" (p-value = 0.0001); are 3.27 times as likely to rate the argument as "neutral" vs. "weak" or "very weak" (p-value < 0.0001); are 4.93 times as likely to rate the argument as "strong" vs. "neutral," "weak" or "very weak" (p-value < 0.0001); and are 6.05 times as likely as theists to rate the argument from parsimony as "very strong" vs. "strong," "neutral," "weak" or "very weak" (p-value =0.03). In this sample, atheists are 2.55 times as likely as agnostics to rate the argument from parsimony as "neutral" vs. "weak" or "very weak" (p-value = 0.0001); are 2.87 times as likely to rate the argument as "strong" vs. "neutral," "weak" or "very weak" (p-value < 0.0001); and are 4.76 times as likely as agnostics to rate the argument from parsimony as "very strong" vs. "strong," "neutral," "weak" or "very weak" (p-value = 0.0006). There is no statistically significant difference between atheists and agnostics with respect to rating the argument from parsimony as "weak" vs. "very weak" (p-value = 0.78). In this sample, there is no statistically significant difference between females and males with respect to how they rated the argument from parsimony (all p-values > 0.23).
In this sample, participants who are not philosophers of religion are 1.82 times as likely as participants who are philosophers of religion to rate the argument from parsimony as “strong” vs. “neutral,” “weak” or “very weak” (p-value = 0.02). There is no statistically significant difference between non-PoRs and PoRs with respect to rating the argument from parsimony as “weak” vs. “very weak,” “neutral” vs. “weak” or “very weak” or “very strong” vs. “strong,” “neutral,” “weak” or “very weak” (all three p-values > 0.35)
Pragmatic argument for atheism
In this sample, atheists are 3.12 times as likely as theists to rate the pragmatic argument for atheism as “weak” vs. “very weak;” are 3.41 times as likely to rate the argument as “neutral” vs. “weak” or “very weak;” are 4.04 times as likely to rate the argument as “strong” vs. “neutral,” “weak” or “very weak” (p-values < 0.0001 for previous three statements); and are 6.45 times as likely as theists to rate the pragmatic argument for atheism as "very strong" vs. "strong," "neutral," "weak" or "very weak" (p-value = 0.005). In this sample, agnostics are 2.44 times as likely as theists to rate the pragmatic argument for atheism as "weak" vs. "very weak" (p-value = 0.0005); and are 2.03 times as likely to rate the argument as "neutral" vs. "weak" or "very weak" (p-value = 0.006). There is no statistically significant difference between atheists and agnostics with respect to rating the pragmatic argument for atheism as "strong" vs. "neutral," "weak" or "very weak" or "very strong" vs. "strong," "neutral," "weak" or "very weak." In this sample, atheists are 1.68 times as likely as agnostics to rate the pragmatic argument for atheism as "neutral" vs. "weak" or "very weak" (p-value = 0.03); are 2.12 times as likely to rate the argument as "strong" vs. "neutral," "weak" or "very weak" (p-value =0.01); and are 2.57 times as likely as agnostics to rate the pragmatic argument for atheism as "very strong" vs. "strong," "neutral," "weak" or "very weak" (p-value = 0.046). There is no statistically significant difference between atheists and agnostics with respect to rating the pragmatic argument for atheism as "weak" vs. "very weak" (p-value = 0.35). In this sample, females are 1.75 times as likely as males to rate the pragmatic argument for atheism as "weak" vs. "very weak" (p-value = 0.01); are 2.27 times as likely to rate the argument as "neutral" vs. "weak" or "very weak" (p-value < 0.0001); are 1.85 times as likely to rate the argument as "strong" vs. "neutral," "weak" or "very weak" (p-value = 0.01); and are 2.22 times as likely as males to rate the pragmatic argument for atheism as "very strong" vs. "strong," "neutral," "weak" or "very weak" (p-value = 0.03). In this sample, participants who are not philosophers of religion are 2.26 times as likely as participants who are philosophers of religion to rate the pragmatic argument for atheism as "strong" vs. "neutral," "weak" or "very weak" (p-value = 0.02). There is no statistically significant difference between non-PoRs and PoRs with respect to rating the pragmatic argument for atheism as "weak" vs. "very weak," "neutral" vs. "weak" or "very weak" or "very strong" vs. "strong," "neutral," "weak" or "very weak" (all three p-values > 0.08)
Argument from incoherence
In this sample, atheists are 4.27 times as likely as theists to rate the argument from incoherence as “weak” vs. “very weak;” are 8.51 times as likely to rate the argument “neutral” vs. “weak” or “very weak;” and are 20.44 times as likely to rate the argument as “strong” vs. “neutral,” “weak” or “very weak” (all three p-values < 0.0001). There is no statistically significant difference between atheists and theists with respect to rating the argument from incoherence as "very strong" vs. "strong," "neutral," "weak" or "very weak" (p-value = 0.88) In this sample, agnostics are 3.5 times as likely as theists to rate the argument from incoherence as "weak" vs. "very weak;" are 4.16 times as likely to rate the argument "neutral" vs. "weak" or "very weak;" and are 8.96 times as likely to rate the argument as "strong" vs. "neutral," "weak" or "very weak" (all three p-values < 0.0001). There is no statistically significant difference between agnostics and theists with respect to rating the argument from incoherence as "very strong" vs. "strong," "neutral," "weak" or "very weak" (p-value = 0.89) In this sample, females are 1.81 times as likely as males to rate the argument from incoherence as "neutral" vs. "weak" or "very weak" (p-value = 0.004); and are 1.71 times as likely as males to rate the argument from incoherence as "very strong" vs. "strong," "neutral," "weak" or "very weak" (p-value = 0.04). There is no statistically significant difference between females and males with respect to rating the argument from incoherence as "weak" vs. "very weak" or "strong" vs. "neutral," "weak" or "very weak" (p-values > 0.07).
In this sample, atheists are 2.04 times as likely as agnostics to rate the argument from incoherence as “neutral” vs. “weak” or “very weak” (p-value = 0.002); are 2.28 times as likely to rate the argument as “strong” vs. “neutral,” “weak” or “very weak” (p-value = 0.0003); and are 3.44 times as likely as agnostics to rate the argument from incoherence as “very strong” vs. “strong,” “neutral,” “weak” or “very weak” (p-value = 0.0002). There is no statistically significant difference between atheists and agnostics with respect to rating the argument from incoherence as “weak” vs. “very weak” (p-value = 0.50).
In this sample, participants who are not philosophers of religion are 1.58 times as likely as philosophers of religion to rate the argument from incoherence as “neutral” vs. “weak” or “very weak” (p-value = 0.03); and are 1.76 times as likely to rate the argument from incoherence as “strong” vs. “neutral,” “weak” or “very weak” (p-value = 0.03). There is no statistically significant difference between non-PoRs and PoRs with respect to rating the argument from incoherence as “weak” vs. “very weak,” or “very strong” vs. “strong,” “neutral,” “weak” or “very weak” (all three p-values > 0.17)
Argument from lack of evidence
In this sample, atheists are 10.22 times as likely as theists and 3.71 times as likely as agnostics to rate the argument from lack of evidence more favorably (p-value < 0.0001 for both statements). In this sample, agnostics are 2.75 times as likely as theists to rate the argument from lack of evidence more favorably (p-value < 0.0001). In this sample, there is no statistically significant difference between the sexes in how they rated the argument from lack of evidence (p-value = 0.60). In this sample, participants who are not philosophers of religion are 1.64 times as likely to rate the argument from lack of evidence more favorably (p-value = 0.002)

Comments:
  • Joshua Rasmussen

    Neat work.
    Here I report how philosophers rate the arguments against theism in my survey
    Since there are many versions of each argument with many, many sub-arguments, one wonders whether the participants rated the same argument–or even argument type (since the “type” will be defined by its instances). I’d be surprised if more than a handful of participants rated the same argument or argument type. For this reason, I’m not sure we can infer from the data that different philosophers rate “religious” arguments differently.
    (It would be interesting to see the results of a survey where specific arguments are spelled out.)

    May 24, 2012 — 6:27
  • What might they contend about [ Google:] Lamberth’s naturalistic arguments about God; in particular Google:] the atelic/teleonomic argument, the one from pareidolia, the reduced animism one, the Flew-Lamberth the presumption of naturalism, the one of rationalism, the one of skepticism, the ignostic-Ockham, Lamberth’s non-genetic argument and the divine protection argument and also McCormick’s why God cannot think, the argument from physical mind, Fr.Meslier’s the problem of Heaven, Reichenbach’s argument from Existence and Angeles’s infinite regress argument?
    His arguments make explicit what others state implicitly.

    May 24, 2012 — 22:55
  • Dominic

    Might it also have to do with the fact that the argument from divine hiddenness has been a bit new on the scene (well, not *new* of course, but much more prominently and explicitly discussed in recent years than earlier) and therefore better known to “insiders” while other arguments are widely discussed since a much longer time?

    May 25, 2012 — 14:26
  • Necessary Concreta Survey

    Continuing in the tradition of Helen De Cruz’s intriguing survey, I present an interactive survey to collect additional data. This survey asks you questions and determines if your answers logically entail that there is one or more necessarily existing …

    August 6, 2012 — 11:11
  • The Psychology and Cultural Evolution of Religion Predicts Bias in Philosophy of Religion

    “What is the first business of philosophy? To part with self-conceit. For it is impossible for anyone to begin to learn what he thinks that he already knows.” –Epictetus, Discourses, Book 2, Ch. 17.
    At HECC and CERC we study religion–its cultural ev…

    March 8, 2013 — 13:41
  • Leave a Reply

    Your email address will not be published. Required fields are marked *