61,665 unique samples • 440,545 individual test records • January โ December 2024 • 16 licensed laboratories
Data through: 12/31/2024
This site relies entirely on data published by the Cannabis Control Commission via their Open Data Catalog and is only current through the dates they make available.
Pass rates, failure breakdowns, and distribution shapes โ Overall mean: 24.44% THCA
Total Yeast and Mold (CFU/g) trends across passing and failing batches
Are all labs measuring the same thing? A 5.45% THCA gap between the highest and lowest reporting labs.
| Lab | Mean THCA | % Above Avg | % in Top Quartile | % in Bottom Quartile | Cohen's d | Above +1σ | Below -1σ |
|---|
How to read this: % Above Average = share of a lab's samples above the overall mean (24.44%). If labs tested identical product, this should be ~50% for everyone. Cohen's d = effect size vs overall โ values >0.2 are notable, >0.4 are large. Top/Bottom Quartile = share of samples in the top 25% (>27.8%) or bottom 25% (<21.1%) of all results. A random Lab_B sample beats a random Lab_L sample 81.7% of the time.
| # | THCA % | Lab | Zone |
|---|
Lab_D produces 82% of the top 50 results โ including the highest at 44.63% THCA. The theoretical maximum for raw flower is ~35-38%.
3,395 batches (same source tag) were tested more than once โ what happened?
Points below the line = THCA dropped on retest ยท 49% dropped, 17% stayed, 33% rose
Negative = potency dropped on retest ยท Mean shift: โ0.53%
822 failed samples retested ยท 70.6% passed on retest overall
Different-lab retests pass at 83.5% vs same-lab at 66.1%
Of 822 failed batches retested, 40% saw YM counts drop by 1,000x or more
Pass rate when a different lab receives a previously-failed sample
Confirmed retests โ first test failed, new sample pulled and passed. Lab_B received 4 of the top 10 retests.
| # | 1st THCA | 2nd THCA | Jump | 1st Lab | 2nd Lab | Same Lab? | Days Apart |
|---|
Two different samples from the same source batch, tested at the same lab, both passed โ yet THCA results differ significantly. This reflects batch heterogeneity or measurement variability.
| # | 1st THCA | 2nd THCA | Difference | Lab | Days Apart |
|---|
Benford's Law compliance and statistical anomaly detection
Chi-squared distance from expected Benford distribution ยท Lower = more natural
Benford's Law predicts that in naturally occurring datasets spanning multiple orders of magnitude, the leading digit "1" appears ~30% of the time, "2" appears ~18%, and so on. Deviations can signal data anomalies.
Yeast & Mold (YM) counts are ideal for Benford analysis because they span values from 1 to 100,000+ CFU/g. THCA (%) is not suitable โ its narrow 10โ35% range naturally skews first digits.
Highest deviation:
• Lab_E: ฯยฒ=72.70 โ 17.7% of first digits are "6" (expected: 6.7%)
• Lab_D: ฯยฒ=40.75 โ 15.5% of first digits are "1" (expected: 30.1%)
Closest to natural:
• Lab_A: ฯยฒ=0.35 โ nearly perfect Benford fit
• Lab_M: ฯยฒ=1.37 โ excellent compliance
ฯยฒ critical value at p=0.05 with 8 degrees of freedom is 15.5. Labs above this threshold have statistically significant deviations from Benford's Law.
Showing labs with highest ฯยฒ scores against Benford's expected distribution (dashed line)
Note: The following labs were excluded from charts due to low sample volume (<100): Lab_H (21 samples), Lab_P (40 samples).
Share this dashboard