Eugine_Nier comments on 2012 Survey Results

[Eugine_Nier]

I wonder whether consequentialism endorsement and possibly some of the probability questions correlate with the two family background questions.

[gwern]

Two? I see FamilyReligion but I dunno what your other one is. But to test family & MoralViews:

R> lw <- read.csv("2012.csv")
R> lwr <- subset(lw, as.character(FamilyReligion) != " ")
R> lwr <- subset(lwr, as.character(MoralViews) != " " & as.character(MoralViews) != "Other / no answer")
R> levels(lwr$FamilyReligion); levels(lwr$MoralViews)
[1] " "                         "Agnostic"                  "Atheist and not spiritual"
[4] "Atheist but spiritual"     "Committed theist"          "Deist/Pantheist/etc"
[7] "Lukewarm theist"           "Mixed / Other"
[1] " "                                     "Accept / lean toward consequentialism"
[3] "Accept / lean toward deontology"       "Accept / lean toward virtue ethics"
[5] "Other / no answer"
R>
R> cor.test(as.integer(lwr$FamilyReligion), as.integer(lwr$MoralViews))

    Pearson’s product-moment correlation

data:  as.integer(lwr$FamilyReligion) and as.integer(lwr$MoralViews)
t = −0.6631, df = 858, p-value = 0.5075
alternative hypothesis: true correlation is not equal to 0
95 percent confidence interval:
 −0.08935  0.04429
sample estimates:
     cor
−0.02263

I wondered if maybe the levels were screwing things up, even though they’re in a logical order which should show any correlation if it exists, so I binned all the results into just binary ‘atheist’ and ‘theist’ (as it were), and looked at a chi-squared:

R> fr <- sapply(as.integer(lwr$FamilyReligion), function(x) if(x>4) {1} else {0})
R> mv ← sapply(as.integer(lwr$MoralViews), function(x) if(x>2) {1} else {0})
R> ct ← chisq.test(fr,mv); ct

    Pearson’s Chi-squared test with Yates’ continuity correction

data:  fr and mv
X-squared = 2e-04, df = 1, p-value = 0.9894
R> ct$expected; ct$observed
   mv
fr      0      1
  0 200.6  58.43
  1 465.4 135.57
   mv
fr    0   1
  0 200  59
  1 466 135

I am a little surprised. Maybe I messed up somehow.

[Eugine_Nier]

Two? I see FamilyReligion but I dunno what your other one is.

The one about which religion.

[gwern]

That’s FamilyReligion then… I don’t see why there’d be two such questions about family religion as you seem to think.

[Eugine_Nier]

I meant RELIGIOUS BACKGROUND.

[gwern]

That field has 41 levels, oy gevalt (I particularly like the religious background “Mother: Jewish; Fat”). Someone else can figure out that analysis!

[A1987dM]

“Mother: Jewish; Fat”

;-D

(Yvain should use larger text fields the next time.)

[gwern]

The lesson I have drawn from the survey is that free-response text fields are the devil and no one is to be trusted with them.