Faith and justice an experimental approach in philosophy

of religion
1. Introduction
One of the arguments against theism says that God, defined as an omnipotent and morally
perfect person who created the world, cannot be reconciled with the presence of evil. Morally
perfect and therefore wholly good person eliminates all the evil that can be eliminated [Mackie,
1955]; evil exists, and thus there can be no person both wholly good and omnipotent. Theists
argue that it is possible that evil is necessary, and God could not create a world without evil or
such a world would be worse than the present one. Because omnipotence is considered as an
ability to do anything which is not logically impossible, the necessity of evil would justify its
existence. It is very important that although theists point this possibility out, they cannot give any
particular reason for all the evil present [Platinga, 1977].
We would like to employ this argument schema for the concept of justice, while originally it is
used with the concept of goodness, and show how the notion of justice used for judging God
differs from the one used for judging human agents. Therefore, if this inquiry is correct, we will
show that God cannot be considered righteous in the common meaning of this term.

2. The problem
According to many religious beliefs God passed on a law in the form of holy scriptures which
forbid, as well as order, some particular behavior. Even if some people have no possibility to
acquaint with those rules, a moral agent is always equipped with a conscience and therefore
capable of choosing between right and wrong and responsible for those choices. In most
religious systems rewards and punishments dependent on agents behavior are promised.
Many people respect Gods rules, and many do not. Among both of those groups there are
individuals who live happily, and ones whose life is miserable. Some of the determinants of
agents happiness are evoked by other agents but many of them have no human cause. Can God
be righteous if living up to principles does not correlate with reward and violent crimes do not
entail severe punishment?
The most common answer the reward or the punishment will be experienced after death is
not as unproblematic as it seems.
(a) First of all, in some cases the repayments are said to occur in this world while the rest will be
realized after agents death. For example, the Bible contains many stories about how God
punished people for their sins still during their lives. The criterion for distinguishing between
cases repaid in this world and those repaid after death cannot be established, at least in the
present state of knowledge.
(b) Maybe it is easier to imagine potential punishments but no simple way of recompensing
someone honest for their sufferings can be revealed. How can God make up for a loss of family
or a growing up without parents? How can one experience happiness after death knowing that
someone very important to her or him suffers for breaking Gods rules?

~1~

(c) No one is put to the same test. Some people do not experience situations when they can break
an important rule, some have to make hard choices every day. Children who died young did not
take any morally significant action for which they could be judged. Moreover, some of the
choices made by an individual are an effect of socialization into a culture which this individual
did not choose.
(d) Even if God passed the law to mankind there are no clear rules of its interpretation. If
interpreted straightforwardly, some articles of law considered as divine look obviously unjust. For
example in Deuteronomy 5:21 it is written: Neither shall you covet your neighbors wife. Neither shall you
desire your neighbors house, or field, or male or female slave []. Does this mean that Decalogue obliges
only males, and that slavery is justified? So if the divine rules cannot be understood in their direct
sense, who is responsible in case of a misinterpretation pursued in a good intention when the
hermeneutical instruments are not sufficient? The interpreter or the author of the unclear text?
There are plenty of other problems connected with the concepts of Gods justice and after-life
repayment. This may be an argument against the existence of a place where all the injustice is
compensated but the strict logical conclusion is weaker maybe it is possible that there exists
a just God who promised reparation for unmerited sufferings but no one can definitely imagine
the way how this divine justice could be realized. Platinga claims that one should not reproach
theists with this ignorance because they are not obliged to know Gods plans. We would like to
check whether Platingas statement can be reconciled with folk intuition about justice.

3. The questionnaires and the interviewees

We have constructed a questionnaire in two forms: MG and GM. The MG questionnaire
contains two descriptions and two questions related to each of those descriptions.
Description 1. Suppose that Ms. Elizabeth has four children. She has established some
rules, for example: brush your teeth after eating, neither swear nor provoke a fight, do
your homework. Nevertheless, Ms. Elizabeth rewards and punishes her children utterly
randomly: a child who violated a proscription or did not fill a warrant might be punished
as well as rewarded. A child who obeys all the rules might be rewarded but also punished.
Often Ms. Elizabeth does not pursue any action.
According to modern (and former) theories of psychology and pedagogy, the behavior of
Ms. Elizabeth cannot be accounted for by any meaningful pedagogical aim. (Although it
does not mean that Ms. Elizabeth does not raise her children in accordance with some
hidden plan).
Question 1: is (according to your moral judgment or the definition of justice you
consider right) Ms. Elizabeth just with reference to her children?
(a) Yes

(b) No

(c) I cannot determine

Question 2: in case of answer (c). I cannot determine, because:

Description 2. Suppose that God exists, is capable of influencing the world, and passed
a law to the people (for example in the form of myths or holy scriptures), which they are
ordered to obey. But there is no correlation between obeying this law and personal
~2~

well-being. The life of people who respect the law may be fortunate and worthy as well as
miserable and full of disappointments. The life of individuals who do not live up to the
rules varies from happy to very unhappy. One can be unhappy for many reasons
because of actions of others, because of ones own wrong choices but there are also many
causes for which no one is responsible epidemics, famines, quakes, floods. The same
situation occurs in case of happiness. Some can achieve happiness by their own effort but
the fact that someone is born in a rich peaceful country rather than in a poor one is
completely arbitrary.
According to the current state of knowledge this situation cannot be justified (although it
does not mean that God does not have a hidden plan which is realized in this way).
Question 3: is (according to your moral judgment or the definition of justice you
perceive right) God from description 2 just with reference to mankind?
(a) Yes

(b) No

(c) I cannot determine

Question 4: in case of answer (c). I cannot determine, because:

The only difference between this questionnaire and the GM is the order of the descriptions
(which is revealed by the abbreviations: MG stands for Mother/God and GM denotes
God/Mother). Both questionnaires contained also the standard questions about age, sex and
education level. The sixth question concerned faith.
6. I think that:
(a) God does not exist and I am (completely or almost) certain,
(b) God does not exist but sometimes I consider the opposite,
(c) I have thought whether God exists and I have found as many pros as cons (or I have
not found any),
(d) God does exist but sometimes I consider the opposite,
(e) God does exist and I am (completely or almost) certain,
(f) it has never bothered me and I have not thought about it.
236 questionnaires have been gathered. Among them there are 115 (48.7%) of MG-type and 121
(51.3%) of GM-type. The distribution of answers is presented in table 1 and figure 1.
(f) 3

Table 1: the distribution of answers to the question

concerning faith

(a) 63

(b) 29
(e) 68
(c) 40
(d) 33

answer:
(a)
(b)
(c)
(d)
(e)
(f)

group:
convinced atheists
doubtful atheists
uncertain
doubtful believers
convinced believers
uninterested

Figure 1: the distribution of answers to

the question concerning faith

~3~

number (share):
63 (26.7%)
29 (12.3%)
40 (16.9%)
33 (14.0%)
68 (28.8%)
3 (1.3%)

aggregate:
atheists:
92 (39.0%)
believers:
101 (42.8%)

The following conventions were accepted. Atheist are identified by answers (a) and (b), and
believers by answers (d) and (e). For both groups an interviewee is called convinced if he or she is
completely or almost certain about the statement given. Otherwise, he or she is called doubtful.
Together it makes six groups: convinced atheists (answer a), doubtful atheists (b), the uncertain
(c), doubtful believers (d), convinced believers (e), and the uninterested (f), and two aggregates
(atheists and believers).
Numbers were assigned to the answers to the first and the third question: yes 2, no 0, I cannot
determine 1, further used in testing statistical hypotheses. There are arguments for using in this
case the nominal scale as well as for the ordinal one. On the one hand, one should not guess
what the interviewee thought marking the answer I cannot determine and therefore whether it can
be treated as a position between yes and no or rather as something quite different. On the other
hand, if the statements in question were voted, exchanging the answer I cannot determine for a pair
of answers yes and no would maintain the judgment. Therefore, one can consider that the
assignment discussed reflects a natural order of the answers. In this inquiry in the statistical tests
the nominal scale is used, as well as the ordinal one, and also samples both with and without the
problematic answer are investigated.
Three separate tests were chosen to verify the hypotheses: the Student's t-test (which is based on
the ordinal scale), the test of equal proportions, and the Fisher's exact test (both are based on the
nominal scale). In the first test the means of the sample are compared, in the second test the
proportions of answers yes are compared, and in the third one a matrix for answers yes and no is
used. The Fisher's exact test is used only for samples with answers I cannot determine excluded. The
alternative hypothesis for the Students t-test is always H1: max > min where max is the greater
one from the means of the samples compared and min is the lesser one. The alternative
hypothesis for the test of equal proportions is that the proportion of answers yes is greater in the
sample with a greater number of this answer. The Fishers exact test is always one-sided and its
alternative hypothesis is consistent with the hypothesis H1 from the Students t-test.

4. Results
As one might have expected, the judgments about God highly differ between believers and
atheists (figure 2 and table 2). It is however interesting that there is a very clear connection between
the strength of the belief that God exists and the judgment whether God from description 2 is just.
As well as that almost a half of believers, and more than 40% of convinced believers consider
God unjust.
Table 2. is God just? The differences of the answers between the groups examined
groups compared:
1 (1)
2 (2)
T
P
F
n1 + n2
believers / atheists
1.188 (0.902) 0.424 (0.683) <0.0001 <0.0001
101 + 92
believers / atheists *
1.223 (0.980) 0.274 (0.692) <0.0001 <0.0001 <0.0001
85 + 73
Explanations: i (i) denote the mean and the standard deviation, the i-index the first or the second group from
the pair compared. The integers in the sum n1 + n2 stand for the number of answers in the first group and the
second one. The symbol * means that the answers I cannot determine were excluded. The columns T, P, and F contain,
respectively, p-values from the Student's t-test, the test of equal proportions, and the Fisher's exact test.

~4~

100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%

100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
yes

believers

I cannot
determine

yes

I cannot
no
determine
convinced believers
doubtful believers
uncertain
doubtful atheist

no

atheists

Figure 2. is God just? The distribution of the answers

The more important result is that Elizabeth is judged differently by different groups. Although
the interviewees generally regard her behavior as unjust, she is more likely to be justified by
believers than atheists (figure 3 and table 3). Does this mean that at least some believers use
a different concept of justice than atheists? If so, where does the difference lie?
Table 3. is Elizabeth just? The differences of the answers between the groups examined
groups compared:
1 (1)
believers / atheists
0.327 (0.650)
believers / atheists *
0.227 (0.638)
Explanations: the same as for table 2.

2 (2)
0.141 (0.408)
0.048 (0.308)

100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%

P
0.0273
0.0232

F
0.0207

n1 + n2
101 + 92
88 + 83

100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
yes

believers

T
0.0088
0.0100

I cannot
determine

no

yes

atheists

convinced believers
uncertain
convinced atheists

I cannot
no
determine
doubtful believers
doubtful atheist

Figure 3. is Elizabeth just? The distribution of the answers

The answer is rather simple. The tests conducted only for the MG-questionnaires show no
difference between atheists and believers (table 4). That means the mother is defended by
believers mainly when the question referring to her occurs after the question about God. This
conclusion is also confirmed by the fact that there is a statistically significant difference between
believers answers concerning the mother from the MG- and from the GM-questionnaire (figure 4
and table 5). Such a difference for atheists was not observed.
~5~

Table 4. is Elizabeth just? The differences of the answers between the groups examined. MG-questionnaire
group:
1 (1)
believers / atheists
0.217 (0,467)
believers / atheists *
0.053 (0.324)
Explanations: the same as for table 2.

2 (2)
0.149 (0.416)
0.047 (0.308)

T
0.2289
0.4719

P
0.500
0.500

F
0.7275

n1 + n2
46 + 47
38 + 42

Moreover, the order of the questions affects also the answers about Gods justice. The effect is
weaker than in the case of the mother, but still quite significant (figure 5 and table 5).
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%

100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
yes

GM

MG

I cannot
determine

no

yes
GM

Figure 4. is Elizabeth just? The distribution of the

answers depending on the questions order.
Believers

MG

I cannot
determine

no

Figure 5. is God just? The distribution of the

answers depending on the questions order.
Believers

Table 5. The impact of the questions order. Believers

group:
Mother: MG / GM
Mother: MG / GM *
God: MG / GM
God: MG / GM *
Explanations: the same as for table 2.

1 (1)
0.418 (0.762)
0.360 (0.776)
1.327 (0.883)
1.375 (0.937)

2 (2)
0.217 (0.462)
0.053 (0.324)
1.022 (0.906)
1.027 (1.013)

T
0.0540
0.0069
0.0456
0.0545

P
0.0205
0.0280
0.0472
0.0796

F
0.0234
0.0797

n1 + n2
55 + 46
50 + 38
55 + 46
48 + 37

Nevertheless, the only ones who judge God and Elizabeth alike (and according to only two out
of the three tests) are the convinced atheists (table 6). It would be a fatal mistake to assert that
believers are wrong because they judge similar cases differently. The descriptions were designed
as a pair of congruent instances but maintaining that they are indeed congruent would be taking
ones individual intuitions for a universal one. On the contrary, the three facts indicate that the
interviewees themselves find them similar, at least to some extent. Firstly, the mother is defended
more and God less frequently if the answer is given as the second one. Secondly, the distributions
of answers to the question concerning God and the one concerning Elizabeth are similar among
convinced atheist. Thirdly, the Pearson's Correlation Coefficient between the first and the second
answer in the whole sample is 0.36.
Among those who could not determine whether God is just many stated that it depended on the
existence of the after-death repayment God is just if and only if a reward or punishment awaits
everyone. Unfortunately, the description referring to God did not contain any of the objections
mentioned above. It is very likely that the results would be even more explicit if those objections
were enclosed.

~6~

Table 6. The difference between the judgments of God and Elizabeth among the groups examined
group:
believers: God / Elizabeth
believers: God / Elizabeth *
atheists: God / Elizabeth
atheists: God / Elizabeth *
convinced atheists: God / Elizabeth
convinced atheists: God / Elizabeth *
Explanations: the same as for table 2.

1 (1)
1.188 (0.902)
1.223 (0.980)
0.424 (0.683)
0.274 (0.692)
0.349 (0.600)
0.163 (0.553)

2 (2)
0.327 (0.650)
0.227 (0.638)
0.141 (0.408)
0.482 (0.308)
0.143 (0.396)
0.036 (0.267)

T
<0.0001
<0.0001
0.0004
0.0058
0.0123
0.0730

P
<0.0001
<0.0001
0.0183
0.0096
0.1807
0.1419

F
<0.0001
0.0086
0.1426

n1 + n2
101 + 101
85 + 88
92 + 92
73 + 83
63 + 63
49 + 56

5. Interpretation. Two models of judging

The first result, rather unexpected, is that almost one in three believers from the GM-sample (and
whose answers were thus uninfluenced) perceives God from description 2 as unjust. For those
believers, the conditions described in description 2 are sufficient to refuse God the attribute of
fairness. Therefore, Platingas schema of theodicy (it is possible that God possesses some positive
attribute, although no one has the slightest idea how this attribute could be realized) in case of
justice does not persuade them.
We have assumed which is an important but rather weak stipulation that interviewees find the
part of description 2 referring to human situation true. This assumption allows us to claim that the
God of one in three believers is not just and therefore does not fall under the concept of the
omnipotent and morally perfect person who created the world.
Philosophically more fruitful is the case of those believers whose judgment is influenced by the
order of the questions. As mentioned above, the interviewees who have read the description of
Elizabeths behavior tend to judge God as more unjust than those who have not. The
interviewees who have read the God case tend to judge Elizabeth as more just than those to
whom her case was presented as the first one. This allows us to state that those two descriptions
are usually found similar. Under this assumption, we introduce two models of reasoning which
can explain the results quite well (although they are not implied by the data directly).
In ordinary cases the following reasoning occurs. When an agent pursues some morally
significant actions, those who judge him or her seek for a consistent pattern or explanation of the
behavior observed. Let us, for the sake of the argument, assume, as in the cases discussed here,
that the judges are to determine whether the agent is just. If only one sufficiently matched pattern
is found, the agent is judged according to it. It does not matter that it is possible that some other
better pattern exists the judgment remains until such a better pattern is found. The incomplete
knowledge of agents motives does not matter either since they are deducted from agents
actions.
The judges can also find two (or more) sufficiently plausible explanations, in accordance with one
of which the agent is just whereas according to the other he or she is unjust. Under those
circumstances the judgment cannot be unequivocally determined. This may explain some part of
the answers I cannot determine, especially when the interviewees made their judgments conditional
on the existence of the after-death repayment. In this case the interviewees found two sufficiently
probable explanations: God is just if such a repayment exists and unjust if the repayment does
not exist.

~7~

We called this model bottom-to-top firstly one judges the actions and then, according to these
judgments, one judges the agent. Elizabeth was judged this way because in the light of her
behavior she was usually found unjust although her motives were hidden.
Judging God, when defending, proceeds in an opposite direction. Firstly, God is presumed just.
Then, Gods actions (which in description 2 seem rather unjust) are assumed to fit some pattern
according to which they would turn out to be just. This pattern does not have to be known. It is
only necessary that such a pattern is possible. The only way to refute this reasoning in the eyes of
those who use it is to prove that such a pattern is impossible, that its existence is logically
incoherent. No highly probable but not unquestionable patterns matter as a counter-argument.
Thus, in case of the incomplete knowledge of agents motives refutation is not possible. We
called this model top-to-bottom first the agent, then the actions.
At this point one can assert that God cannot be just if the term just is used in its common
everyday meaning as a bottom-to-top concept. But the effect of changing the order of the
questions proves something more. Those whose answers depend on the questions order realize
that they use two concepts of justice instead of one. This bothers them. If they did not mind this,
such an effect would not occur. Therefore, instead of reasoning in two separate ways
bottom-to-top for Elizabeth and top-to-bottom for God they employ the mechanism from the
first question to answer the second one. They do not want to relinquish the cohesion of their
concept of justice for the sake of following their initial intuitions. This is how the difference in
GM- and MG-distribution of answers occurs.
It would be a very interesting result if it turned out that those believers who defend God reason
generally rather in the top-to-bottom way whereas others use the bottom-to-top mechanism in
their judgments. Then there would be two distinct groups with different concepts of justice and
therefore also different languages and different ways of judging. Within the groups their members
would be able to conduct moral reasoning while one group would find the proofs of the other
unpersuasive and incoherent.
The consistency of the distribution of answers regarding Elizabeth, observed in the MG-sample,
suggests quite a dissimilar situation. We suppose that all regular judgments fall under the
bottom-to-top model. The reverse mechanism is employed only when one is intent on defending
a presumed position. Thus the motivation for using this mechanism seems to be emotional rather
than rational. It is possible, however, that this explanation is not valid or at least not sufficient.
Although we tried to make description 2 as concrete as we could, the results might indicate that the
top-to-bottom mechanism might occur in abstract cases (if Gods case may be found concrete)
whereas the bottom-to-top in concrete ones (such as Elizabeths). The importance of
abstractness and concreteness in moral intuitions is, indeed, discussed in experimental philosophy
[Sinnott-Armstrong, 2008].

6. Conclusion
The general conclusion which we propose here is that according to the concept of justice, used
both by atheists and believers when judging regular cases, God cannot be just. The schema of
argument defending God proposed by Platinga in [Platinga, 1977] is incoherent with this concept
of justice (by concept we also mean the rules of its application) and therefore we find his
theodicy unsuccessful. Moreover, the believers who reason in accordance with Platingas schema
~8~

often realize that it implies a different concept of fairness. Because they tend to use one instead
of two separate concepts, they cope with this problematic situation by changing their initial
judgments.

7. References
1. Knobe J., Nichols S., An Experimental Philosophy Manifesto, [in:] Knobe J., Nichols S.,
Experimental Philosophy, Oxford 2008.
2. Mackie J. L., Evil and Omnipotence, (1955) Mind, New Series, 254.
3. Platinga A., God, freedom and Evil, Grand Rapids, Michigan: Eerdmans Publishing Co., 1977.
4. Sinnott-Armstrong W, Abstract + Concrete = Paradox, [in:] Knobe J., Nichols S., Experimental
Philosophy, Oxford 2008.

~9~