I’d strongly urge the OP or anyone interested in this topic to dig into the SSC 2017 survey’s EA population and investigate mental health statistics… I imagine the likelihood of oversampling people with mental health concerns would be lower (though not nonexistent) there.
## Download the file from https://slatestarcodex.com/2017/03/17/ssc-survey-2017-results/ , then move into the directory the file is in with:
## setwd(“directory”)
C ← read.csv(file=”Survey_CSV.csv”, header=TRUE, sep=”,”, stringsAsFactors=FALSE)
c(69:77) → mental_conditions # “Schizophrenia”, Autism”, “Depression”, “Anxiety”, “OCD”, “Eatingdisorder” “Bipolar”, “Alcoholism”, “Drugaddiction”.
Has_mental_condition_diagnosis=rep(0,dim(C)[1])
for(mental_condition in mental_conditions){
ifelse(C[,mental_condition]==”I have a formal diagnosis of this condition”, 1, Has_mental_condition_diagnosis) → Has_mental_condition_diagnosis
}
ifelse(C$EAID==”Yes”, 1,0) → Is_EA
summary(lm(Has_mental_condition_diagnosis ~ Is_EA))
Has_mental_condition_diagnosis_or_intuited = rep(0,dim(C)[1])
Diagnosis_or_intuited = c(“I have a formal diagnosis of this condition”,”I think I might have this condition, although I have never been formally diagnosed” )
for(mental_condition in mental_conditions){
ifelse(C[,mental_condition]%in%Diagnosis_or_intuited, 1, Has_mental_condition_diagnosis_or_intuited) → Has_mental_condition_diagnosis_or_intuited
}
summary(lm(Has_mental_condition_diagnosis_or_intuited ~ Is_EA))
ifelse(C$SSRIs == “I have never taken these drugs”, 0, 1) → Has_taken_SSRIs
summary(lm(Has_taken_SSRIs ~ Is_EA))
Also of note is that the SlateStarCodex 2017 survey offered an EA ID question in addition to a mental health inventory (David links to the 2019 SSC Survey which sadly does not have the EA ID question). We benchmarked the 2017 EA Survey to the SSC 2017 survey and found the EA populations to be mostly similar.
I’d strongly urge the OP or anyone interested in this topic to dig into the SSC 2017 survey’s EA population and investigate mental health statistics… I imagine the likelihood of oversampling people with mental health concerns would be lower (though not nonexistent) there.
Using the 2017 SSC, I’ve looked at:
Has been diagnosed with a mental illness ~ Is EA
Weak relationship, is not significant.
Has been diagnosed with a mental illness or intuits having one ~ Is EA
Weak relationship, 0.05 < p < 0.10
Has taken SSRIs ~ Is EA.
Weak negative relationship (??).
Code used to produce that (in R)