September 18, 2025

Science and Checking Reasons

Outline

  • Recap
  • Scientific Evidence and Diagnostic Frames
  • Scientific Evidence and Prescriptive Claims
  • Outline of Remainder of the Course

Recap

Why do we care about truth?

One main argument is:

Exercise of power involves someone else motivating us with reasons to do or think something else.

Reasons

The reasons given to use greater enforcement powers involve claims:

  • “most political violence is perpetrated by the left”
  • “rhetoric from left-leaning organzations led to the assassination”
  • “political violence is illegitimate”
  • “US government should use investigations to target organizations producing the rhetoric that leads to violence”

Reasons

Some of these reasons are empirical claims; some are normative claims

  • diagnostic frames: there is a “problem” to be fixed.
  • prescriptive claims: there is a “solution” we should use.

Science and Reasons

Scientific Evidence can only be used with empirical claims:

  • “most political violence is perpetrated by the left”
  • “rhetoric from left-leaning organzations led to the assassination”
  • Is this still worthwhile?

Probing Claims

When we are given reasons to do something (e.g. accept the use of police powers to target political rivals)

  • claims diagnosing a problem
  • claims prescribing a solution

involve value judgments. But they also involve empirical claims.

  • We want to look at whether the evidence for the empirical claims was capable of showing those claims to be wrong
  • We want evidence that has thoroughly checked whether the claim is wrong.

Severity and Good Reasons

If the evidence we are given to accept power over us would always support the claims, even if the claims are false, we are not free. Bad evidence \(\to\) Bad reasons

Science and Diagnostic Frames

Science and Diagnostic Frames

Claims that there is a “problem” involve a value judgment and descriptive claims

  • “political violence is bad” (a value judgment)
  • “political left is perpetrating political violence”

Science and Diagnostic Frames

Evaluating descriptive claims are relevant to “diagnostic frames”:

  • How much political violence is there, really? (Compared to other violence)
  • How much political violence is committed by the left? (As opposed to other groups)
  • When people tell us there is a problem, there is a value judgement about what is good or bad. We can use scientific evidence to check whether this “problem” is real

Severity and Descriptive Claims:

Where can evidence for descriptive claims go wrong?

  • How do we define what we are trying to observe? (What is “political violence”? Who is a “leftist”?) Can that definition be used objectively?
  • What will we observe that correspond to these definitions? (What could we observe that counts as “political violence”?)
  • How/What procedure will we use to actually observe these things?
  • What could go wrong at each stage that would lead us to incorrectly conclude the claim is right (wrong)?

Science and Prescriptive Claims

An attempt to motivate people to behave differently. An attempt to exercise power. Access to media outlets enables power.

An Example:

(1) US is not experiencing (comparatively) high levels of immigration

  • Lebanon (4.4 million people) has had more than 1 million refugees in less than 10 years (>25%)
  • American (325 million people) has 44 million immigrants (13.7 percent)
  • Rate of immigration to US has slowed over the past 10 years

(2) High immigration rates do not lead to political instability.

  • Canada and Australia have populations that are 20 and 28 percent foreign-born, but no major political problems

An Example:

(3) New immigrants are employed at high rates.

  • Unemployment among immigrants is lower than native-born Americans
  • not a burden on state provision of social services

(4) Lower immigration slows economic growth

  • Fertility rates among native-born Americans are dropping
  • Because of that, future workforce will be smaller, productivity will be less, growth will slow

(5) America should admit 1 million more immigrants per year

  • So growth rates can remain high

“Actually, the Numbers [Don’t] Show That”

\(\checkmark\) if science could test:

  1. US is not experiencing high levels of immigration \(\checkmark\)

  2. High immigration rates do not lead to political instability \(\checkmark\)

  3. New immigrants are employed at high rates \(\checkmark\)

  4. Lower immigration slows economic growth \(\checkmark\)

  5. America should admit 1 million more immigrants per year

Even if 1-4 are true, what must we to assume to conclude that (5) is true?

Prescriptive Claims

prescriptive claims:


are normative claims that assert what kinds of actions should be taken

  • hint: like a doctor or pharmacist, it prescribes a course of action.
  • overlap with justifications/reasons given by power.

The basis for a prescriptive claim includes

  • evidence supporting an empirical claim about the consequences of some action (causal claim)
  • an assumption that some value judgment is correct.

Revisit our Starting Example

(5) America should admit 1 million more immigrants per year

This is a prescriptive claim:

For it to be true…

  1. What value judgments must we assume to be true?

  2. What empirical claims must be true?

“Actually, the Numbers [Don’t] Show That”


“America should admit 1 million more immigrants per year”

Even if 1-4 are true: we need to assume that economic growth is desirable to conclude that (5) is true.

Revisit our Starting Example

(5) America should admit 1 million more immigrants per year


scientific evidence cannot “prove” this claim

even if we evidence that is very capable of finding any flaws in claim that increasing immigration increases economic growth… (strong severity)

people who value cultural/ethnic homogeneity more than economic growth can’t be persuaded

Another Example

Another Example

You and your friends win a large sum of money in a lottery

You and your friends agree: you want to do the most good by donating the money.


You consider some options…

Another Example

Which should you donate to?

  • Option 1: Make-a-Wish (more Batkid, pls)
  • Option 2: Mosquito Nets
  • Option 3: Direct transfer of cash to impoverished people

Can science solve our problem?

Peter Singer and effective altruists say yes!

  • we can evaluate which of these does the most good!

“Saving a child’s life has to be better than fulfilling a child’s wish to be Batkid.”

Can science solve our problem?

Empirical Evidence

  • Malaria kills ~500k per year
  • Half of global population possibly exposed
  • Mosquito nets reduce likelihood of exposure
  • For each 100 to 1000 nets, 1 death prevented
  • Cost of mosquito nets is low
  • Cash transfers are expensive, effects on mortality unclear
  • Batchildren encourage vigilante justice

What should you do?

Malaria nets!

Can science solve our problem?

But wait, your friend says: experiments show that directly giving cash

  • benefits the health, education, and life choices of children
  • empowers women to be financially independent, escape abuse
  • improve mental health

What should you do?

Can science solve our problem?

If you value minimizing suffering, but your friend values maximizing individual freedom…

then science cannot help us, because the disagreement is rooted in value judgements

Another Example

“We should donate money for mosquito nets” is a prescriptive claim.


  • Scientific evidence for empirical claim that “Mosquito nets (A) prevent malaria (B)”
    • NOT enough to conclude that does not imply “we should do (A)”: it depends on how we value B
  • if we also accept value judgement that it reducing suffering is more important than maximizing freedom (or coolness)
    • THEN we can accept the prescriptive claim

Need to accept the causal (empirical) claim that \(A \to B\) AND a value judgment that \(B\) is good.

Another Example

Science is still be helpful!

  • If we assume less mortality is good (B) (a value judgment)
  • What if science shows: mosquito nets don’t prevent malaria deaths. (empirical evidence)
  • “A does/does not cause B” is informative!

Severity and Causal Claims:

Where can evidence for causal claims go wrong?

  • What is causality?
  • When and why might the relationship (correlation) between factors not be the result of causation
  • What comparisons can we make to provide evidence of causation?
  • What do these comparisons assume to count as evidence of causation?
  • Is it possible that the evidence we are using could show a relationship that is not causal?

Conclusion

Conclusion

Learning to recognize what evidence is capable of proving claims wrong:

  • how and why could evidence find support for a claim even if the claim is wrong
  • what assumptions permit evidence to able to find claims to be wrong (when they are wrong)