In the phase 1 data set, the least amount of measured CBDA content was 11.80% of the total measured cannabinoid content. However, this sample also had unusually high levels of CBD, indicating extensive enzymatic or heat-induced decarboxylation. The highest amount of CBDA was 83.24% of the total cannabinoid content, and the mean CBDA content was 58.59% (SEM 2.84%, n = 30). CBD content is the next most abundant constituent and ranged from a low of 6.09% of the total measured cannabinoids to a high of 75.04% of the total, with the mean being 24.56% (SEM 2.89%, n = 30).

Histograms, Q-Q plots, and scatter plots by both supplier and cultivar were produced for all measured and computed analyte values. The supplemental data addendum contains all of these charts. Visual inspection shows many of the histograms indicate right-tailed skew, with the majority of values clustered on the lower end of the scales.

Visual inspection of each cannabinoid’s content was conducted by applying color scales to the percentage-of-content table on a per sample (row-by-row) basis. The per-sample color-scaling creates a heatmap showing the analytes with the greatest to least percentage of the cannabinoid content, as shown in Table 9. Darker to lighter shares of red indicate the higher percentages, medium percentages are highlighted in shades of yellow, and the lowest percentages of cannabinoid content are in green. Note that some totals do not add up to 100% due to small rounding errors.

Differences in cannabinoid content by cultivar are also observable when samples are placed in a row-oriented heatmap and sorted by cultivar, as shown in Table 10. Differences are also observed when row-oriented heat-mapped data is sorted and categorized by the supplier, as in Table 11. The color-coding in Table 10 and Table 11 both follow the same scheme as described for Table 9, where red indicates more copious amounts of any specific cannabinoid, yellow being moderate levels, and green indicates lower levels of the analyte. Strong similarities were observed between cannabinoid content in the scatter plots by the supplier and by cultivar, but this relationship is not surprising since most manufacturers only supplied a single cultivar.

Each supplier was assigned a unique identifier, and these values were transformed and coded into a series of new dichotomous variables to investigate any supplier-related variance. A series of linear regression tests with 0.05 significance were performed on the analytes having statistically normal distribution using supplier-encoded dichotomous variables as coefficients.

For the phase 1 data set, the regression analysis was limited to CBCA, CBDA, and CBNA. The supplemental data addendum contains the model summary data, ANOVA, and coefficients for the regression of these four tests. A summary of each regression analysis done with the suppliers (n = 30) is provided in Table 1Table 3:

Table 3: Phase 1 regression of analytes and suppliers