Chapter 7 — Key Takeaways

A one-page field card. The full argument is in index.md.

The core claims

  • DNA is the one forensic method that earned its courtroom authority honestly — built and validated by outside scientists before it was trusted, not after innocent people were convicted on it. That sequence is the whole difference from the pattern disciplines.
  • DNA typing reads variation, not identity. It targets short tandem repeats (STRs) in non-coding DNA — deliberately blind to your health, traits, and ancestry. It answers exactly one question: whose cells are these?
  • A profile is a list of alleles at many loci. One locus narrows (class-like); ~20 loci together approach individualization. Two profiles "match" if they share alleles at every locus.
  • Mismatch refutes; match supports. A single clean, reproducible mismatch excludes with near-certainty. Agreement at every locus only makes the suspect consistent with the source — its strength is then a probability (the RMP).
  • The random match probability (RMP) is the chance a random unrelated person would coincidentally share the profile. State it as a fraction and in words ("1 in 1.2 billion — far more than the U.S. population").
  • DNA identifies a source, never a time, a mechanism, or a guilt. "His DNA is on the object" and "he committed the act" are different statements; the distance between them is the transfer problem (§7.6).

The method, in one breath

Extract the DNA → quantify it → PCR-amplify the target STR loci a billion-fold → sort the fragments by size with capillary electrophoresis → read the peaks on the electropherogram → assemble the DNA profile → compare directly, or search CODIS (a lead, never a verdict).

Each step has a failure mode: extraction (inhibitors, contamination); PCR (exponential amplification of any template, including contaminants); CE/interpretation (judgment at the edges — noise vs. real minor peaks).

The two traps to refute on sight

  • The prosecutor's fallacy. "RMP of 1 in a billion" ≠ "1 in a billion chance the defendant is innocent." The RMP is conditioned on the person being a random non-source; it does not invert into a probability of guilt. (Full treatment: Chapter 9.)
  • Presence = guilt. A correct match misread as proof of presence/guilt (the Lukis Anderson case) can convict an innocent person more persuasively than junk science, because the number sounds like certainty.

What the RMP does NOT cover

  • Close relatives — a sibling's match probability is far higher than the headline figure.
  • Laboratory error — swaps, contamination, mislabels. This is a real, human rate that does not shrink by typing more loci, and is realistically larger than a one-in-a-billion coincidence.

Method-validity verdict (NAS 2009 / PCAST 2016)

Single-source nuclear DNA: STRONG — the top of the validity spectrum, and its anchor. Understood biological basis, quantified error structure, objective core measurement. But validity is highest for clean, single-source samples; it degrades as samples become trace, degraded, or mixed (Chapter 8), where interpretation — and error — return. Position on the spectrum is a ceiling on reliability, not a guarantee: contamination, a mislabeled tube, or a misread mixture can still produce a wrong result.

What you can honestly say on the stand

"This evidence profile matches the defendant's reference profile at all twenty loci. A profile like this is found in approximately one in [N] unrelated individuals in the relevant population. That tells you about the source of the DNA. It does not tell you when the DNA was deposited, how it got there, or that the person who left it committed any crime — and it does not exclude a close relative or account for the possibility of laboratory error."

What you may NOT say: "The DNA proves he did it." · "There's a one-in-a-billion chance he's innocent." · "The database identified the perpetrator." (The database identified a candidate source, to be confirmed by direct re-typing.)

Key terms

DNA · STR · locus · allele · PCR · electropherogram · DNA profile · CODIS · random match probability (RMP)

Cold Case — status after Chapter 7

A partial DNA profile was developed from the gas-can handle and searched against CODIS. A profile exists; its source, timing, and meaning are all undetermined. No one is included; no one is excluded; the accidental-fire assumption stands untouched by this evidence alone. Flagged for Chapter 8: the sample may be a heat-degraded mixture, and a touch deposit on a much-handled object is weak associative evidence even if the match is strong.