Table 4.
Relative Importance of 20 Cannabis Attributes: Hierarchical Bayesian Estimation of Mixed Logit Regressions (Full Sample and Subsamples by Cannabis Use Purposes)
| Attributes | All (N = 817) | By Cannabis Use Purposes Recreational Users (N = 266) | Medical Users (N = 211) | Dual-purpose Users (N = 340) | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Mean | Std. dev. (std. err.) | Share of importance (%) | Mean | Std. dev. (std. err.) | Share of importance (%) | Mean | Std. dev. (std. err.) | Share of importance (%) | Mean | Std. dev. (std. err.) | Share of importance (%) |
|
| Product Characteristics | ||||||||||||
| THC | 0 | 0 | 9.21 | 0 | 0 | 8.49 | 0 | 0 | 6.04 | 0 | 0 | 13.47 |
| CBD | −0.30 (0.07) | 1.28 (0.08) | 7.31 | −0.64 (0.12) | 0.85 (0.14) | 5.09 | 0.81 (0.14) | 1.65 (0.15) | 10.36 | −0.80 (0.11) | 1.13 (0.13) | 7.03 |
| Strain type | 0.29 (0.07) | 1.32 (0.09) | 11.32 | 0.22 (0.13) | 1.21 (0.13) | 9.97 | 0.54 (0.13) | 1.10 (0.15) | 8.78 | 0.04 (0.12) | 1.14 (0.14) | 13.90 |
| Flavor | −1.14 (0.07) | 1.15 (0.08) | 3.54 | −0.98 (0.12) | 1.29 (0.14) | 3.76 | −0.79 (0.13) | −1.03 (0.16) | 3.18 | −1.56 (0.12) | 1.17 (0.12) | 3.53 |
| Product Quality | ||||||||||||
| Quality | 0.57 (0.07) | 1.27 (0.08) | 13.68 | 1.09 (0.14) | 1.57 (0.15) | 16.90 | 0.67 (0.14) | 1.02 (0.18) | 9.52 | 0.22 (0.12) | 1.31 (0.13) | 15.75 |
| Origin | −0.98 (0.07) | −0.98 (0.08) | 4.10 | −0.83 (0.12) | −1.02 (0.14) | 4.32 | −0.39 (0.13) | −1.20 (0.16) | 4.46 | −1.66 (0.11) | 0.94 (0.13) | 3.21 |
| Pesticide | −0.07 (0.07) | 1.32 (0.08) | 8.76 | −0.03 (0.12) | 1.29 (0.14) | 8.33 | 0.68 (0.13) | 1.24 (0.16) | 9.60 | −0.65 (0.11) | 1.50 (0.14) | 8.00 |
| Lab test | −1.22 (0.07) | 0.95 (0.08) | 3.32 | −1.11 (0.12) | 1.03 (0.14) | 3.37 | −0.44 (0.13) | 0.97 (0.15) | 4.28 | −1.88 (0.11) | 1.01 (0.13) | 2.62 |
| Package Characteristics | ||||||||||||
| Label instruction | −2.09 (0.07) | 1.34 (0.08) | 1.46 | −2.05 (0.13) | 1.53 (0.13) | 1.40 | −1.11 (0.14) | 1.05 (0.14) | 2.40 | −2.85 (0.12) | 1.29 (0.13) | 1.02 |
| Label warning | −1.90 (0.07) | 1.17 (0.08) | 1.75 | −1.85 (0.13) | 1.34 (0.17) | 1.68 | −0.83 (0.13) | 1.31 (0.16) | 3.08 | −2.67 (0.12) | 1.04 (0.13) | 1.21 |
| Date of harvest | −1.20 (0.07) | −0.57 (0.10) | 3.37 | −1.14 (0.12) | −0.60 (0.15) | 3.27 | −0.62 (0.13) | 0.47 (0.18) | 3.69 | −1.72 (0.14) | −0.73 (0.13) | 3.03 |
| Package | −2.37 (0.07) | 1.33 (0.08) | 1.11 | −2.36 (0.13) | 1.06 (0.14) | 1.04 | −1.48 (0.13) | 1.25 (0.14) | 1.71 | −3.07 (0.12) | 1.24 (0.11) | 0.82 |
| Price and Free Sample | ||||||||||||
| Price | 0.21 (0.08) | 1.83 (0.09) | 10.77 | 0.30 (0.12) | 1.31 (0.14) | 10.54 | 0.53 (0.14) | 1.65 (0.17) | 8.72 | −0.29 (0.12) | 2.09 (0.14) | 10.75 |
| Free Sample | −1.74 (0.07) | 1.66 (0.08) | 2.04 | −1.47 (0.13) | 1.91 (0.16) | 2.41 | −1.17 (0.14) | 1.59 (0.16) | 2.26 | -2.38 (0.12) | 1.83 (0.14) | 1.61 |
| Store Characteristics | ||||||||||||
| Compliance to law | −0.87 (0.07) | 1.63 (0.08) | 4.52 | −0.67 (0.13) | 1.94 (0.16) | 4.96 | 0.05 (0.13) | 1.80 (0.16) | 6.28 | −1.85 (0.12) | 1.67 (0.13) | 2.68 |
| Staff | −0.86 (0.07) | 1.07 (0.08) | 4.54 | −0.80 (0.12) | 1.12 (0.14) | 4.44 | −0.19 (0.13) | 1.37 (0.16) | 5.23 | −1.43 (0.11) | −0.91 (0.13) | 3.97 |
| Distance | −1.38 (0.07) | 1.37 (0.08) | 2.85 | −1.11 (0.12) | 1.21 (0.14) | 3.37 | −0.77 (0.12) | 0.92 (0.16) | 3.24 | −1.99 (0.12) | 1.48 (0.13) | 2.35 |
| Store hours | −1.94 (0.07) | 1.19 (0.08) | 1.69 | −2.07 (0.125) | 1.32 (0.13) | 1.37 | −1.32 (0.13) | 0.90 (0.15) | 1.98 | −2.45 (0.12) | 0.96 (0.12) | 1.51 |
| Restrictions on Use | ||||||||||||
| Public use | −1.78 (0.07) | 1.51 (0.09) | 1.97 | −1.69 (0.13) | 1.64 (0.15) | 1.97 | −1.25 (0.14) | 1.39 (0.17) | 2.11 | −2.37 (0.12) | 1.82 (0.14) | 1.62 |
| Employer | −1.45 (0.07) | 1.82 (0.08) | 2.68 | −1.13 (0.12) | 2.00 (0.15) | 3.32 | −0.82 (0.14) | 1.74 (0.17) | 3.09 | −2.19 (0.12) | 1.88 (0.14) | 1.93 |
Notes:
The share of importance was calculated with formula , where a was the number of attributes shown per BWS task (in this study it was 4), “mean” was the coefficient mean of each attribute in hierarchical Bayesian estimation of mixed logit regressions, and b is a constant to make the sum of the share of importance to be 100.