potency_audit.ipynb Methods appendix

Laboratory potency comparison

In [1]: 
import pandas as pd, numpy as np
from canndata import baseline, figures
In [2]: 
df = pd.read_parquet("state_results_2025.parquet")
flower = df[(df.category == "flower") & (df.analyte == "total_thc")]
flower.shape
Out[2]: (48217, 14)
In [3]: 
by_lab = flower.groupby("lab").value.mean()
by_lab.round(1).sort_values(ascending=False).head()
Out[3]: 
lab
D    31.2
A    25.6
F    24.1
B    23.4
In [4]: 
over30 = flower.assign(hi = flower.value > 30)
share = over30.groupby("lab").hi.mean()
ratio = share["D"] / baseline.national_share(30)
f"{share['D']:.0%} above 30% THC, {ratio:.1f}x national"
Out[4]: '45% above 30% THC, 5.6x national'
Findings · Potency Full report · p. 7
Figure 1
Average reported Total THC by laboratory
Source: state COA submissions, Apr–Sep 2025. n = 48,217 results.
Figure 2
Laboratory average compared with the national benchmark
Executive summary

One laboratory's THC levels are far above the norm

Prepared for the State Regulatory Authority · 2025 review

The laboratory's average Total THC for flower is higher than every other laboratory in the state and well above the national average. At that laboratory, 45% of flower samples are reported above 30% Total THC, 5.6× the national rate. This pattern warrants a closer look.

45%
flower samples above 30% THC
5.6×
the national rate
Potency Laboratory comparison National benchmark

The analysis ranks laboratory-level results against the rest of the state and a national baseline, providing a clear basis for prioritizing further review and independent verification.

1