[lit-ideas] Three arguments against quantitative social "science" as science

  • From: Torgeir Fjeld <torgeir_fjeld@xxxxxxxx>
  • To: "lit-ideas@xxxxxxxxxxxxx" <lit-ideas@xxxxxxxxxxxxx>
  • Date: Tue, 2 Sep 2014 10:49:04 +0100

Are the social sciences scientific? Do they deal with causality in ways that 
are sufficiently similar to that of the natural sciences to lay claim to the 
recognition accrued by biology, physics, chemistry, etc? Here's three arguments 
as to why quantitative methods in social "sciences" do not vouch for these 
disciplines`s status as "sciences".

(1) Causality can't be proven -- ala Hume. It's possible to demonstrate that 
the the white billiard ball moved first, the black ball second, and that there 
was an encounter beteen the two. But there is no necessary relation between 
these statements and to say that the encounter was the _cause_ of the black 
ball's movement.

(2) Correlation is not causality. Statistics can demonstrate correlation 
between to variables, say education and longevity. This is not to say, though, 
that a causal relation has been established. The relation between these 
variables are purely convetional -- some variables are considered to be more 
"fundamental" than others. Such variables include income, gender?sex, ethnicity 
etc. However, it is possible to concieve of a relation whereby education has 
some relation to gender -- say, access to and knowledge of sex change 

(3) Hempel's law of symmetry ("nomology") between explanation and prediction 
doesn't apply. Say that I throw a dice 60 times. The result if the experiment 
is that the number 1 came up 10 times, the number 2 came up 10 times, etc. Now, 
from these results there is no way for me to estimate the probability of the 
result of the next throw of the dice: It is still 1:6 for each side of the 
dice. And, importantly, even if I had thriown the dice 60 times with the same 
side coming up, the chances for that sime side coming up the 61st time would 
_still be_ 1:6. The kinds of macro statistics assembled by the socalled 
quantitative social sciences are of the same kind: They purportedly show how 
large data sets behave, but their pridctive force are negligeble.


Med vennlig hilsen / Yours sincerely, 

Torgeir Fjeld 


Other related posts: