Exploring survey methods for measuring consumption quantities of cannabis flower and concentrate products

Abstract

Background:

Researchers need accurate measurements of cannabis consumption quantities to assess risks and benefits. Survey methods for measuring cannabis flower and concentrate quantities remain underdeveloped.

Objective:

We examined “grams” and “hits” units for measuring flower and concentrate quantities, and calculating milligrams of THC (mgTHC).

Methods:

Online survey participants (n=2,381) reported preferred unit (hits or grams), past-week hits and grams for each product, and product %THC. Quantile regression compared mgTHC between unit-preference subgroups. Hits-based mgTHC calculations assumed a universal grams-per-hit ratio (GPHR). To examine individualized GPHRs, we tested a “two-item approach,” which divided total grams by total hits, and “one-item approach,” which divided 0.5 grams by responses to the question: “How many total hits would it take you to finish 1/2g of your [product] by [administration method]?”

Results:

Participants were primarily daily consumers (77%), 50% female sex, mean age 39.0 (SD 16.4), 85% White, 49% employed full-time. Compared to those who preferred the hits unit, those who preferred the grams unit reported consuming more hits and grams, higher %THC products, and consequently, larger median mgTHC (flower-hits mgTHC: 32 vs. 91 (95%CI: 52–67); flower-grams mgTHC: 27 vs. 113 (95%CI: 73–95); concentrate-hits mgTHC: 29 vs. 59 (95%CI: 15–43); concentrate-grams mgTHC: 61 vs. 129 (95%CI: 43–94)). “Two-item” and “one-item” approach GPHRs were similar and frequently 50% larger or smaller than the universal GPHR.

Conclusion:

Allowing respondents to choose “hits” or “grams” when reporting cannabis quantities does not compromise mgTHC estimates. A low-burden, one-item approach yields individualized “hit sizes” that may improve mgTHC estimates.

1. INTRODUCTION

Cannabis products and use patterns continue to diversify in parallel with evolving cannabis laws and norms in the US (1,2). Characterizing these complex patterns requires new, multidimensional survey methods, yet researchers often rely on simple frequency-of-use measures (3–5). Although consumption frequency is an indispensable measure, consumption quantity is also crucial for studying the multidimensional biopsychosocial effects of substance use on individuals and populations (6–11). However, methods for assessing cannabis quantities remain underdeveloped (12–18).

Cannabis measurement approaches need to accurately estimate milligrams of Δ9- tetrahydrocannabinol (mgTHC) because Δ9-THC underlies many of the positive and negative biopsychosocial consequences of cannabis use (19–21). However, estimating mgTHC with surveys is more difficult for certain products than others. For example, edible product labels inform consumers directly about milligrams of Δ9-THC. In contrast, labels for flower (i.e., “bud”) and concentrate products typically indicate product weight (e.g., 1 gram) and potency (i.e., %THC). Consequently, those who smoke or vape these popular products (12,22,23) cannot reliably report their total mgTHC ingestion. Estimating mgTHC consumed via flower and concentrate products requires novel survey and quantification methods.

Researchers have explored different quantity units for assessing cannabis flower and concentrate use such as grams (24–27), “hits”, “puffs”, or “tokes” (i.e., inhalations)(24,28), joints, pipes, or bowls (29,30), and hours of intoxication (31–33). An increasingly common method is the “unit-preference” survey design, which allows participants to choose a unit to report their quantities (12,34–36). For example, one participant might choose to report quantity in hits, while another might choose to report in grams. This survey design presumably reduces participant burden and increases response accuracy by accommodating heterogeneous use patterns, products, and methods of administration.

In a prior study, we employed the unit-preference survey design and calculated mgTHC consumption using participants’ reported number of hits or grams with formulas incorporating product potency (37,38). We found that those who chose to report in grams had significantly larger mgTHC estimates than those who chose hits. However, it was not clear why this difference occurred. One possibility was that those who chose the grams unit actually consumed more mgTHC than those who chose hits; alternatively, some form of measurement error could have contributed to the discrepancy (e.g., different psychometric properties of the hits and grams units, differences in the formulas used to convert hits or grams into mgTHC). Additional analyses indicated that those who chose the grams unit were more likely to engage in behaviors indicative of heavier consumption (e.g., daily use, morning use, higher-potency products), which supported, but did not confirm the former explanation, i.e., that they actually had greater consumption.

A second issue affecting the unit-preference design is that the grams unit can be converted directly into mgTHC, but the hits unit cannot. Hits-based mgTHC formulas require a “grams-per-hit” ratio (GPHR) (i.e., “hit size”) to convert hits to grams (e.g., 0.05 grams of flower consumed per hit). To date, many studies have assumed a universal (i.e., same for all participants) GPHR to convert hits into grams (12,39–41). In our prior study, we also used universal GPHR values of 0.06 for flower products and 0.012 for concentrate products based on results from human lab studies (42–44). Importantly, however, this universal approach obscures individual differences in GPHRs that are apparent from hit sizes observed in the smoking topography literature (44–47), which impacts the accuracy of individual mgTHC estimates.

1.1. Current Study

This study sought to understand why mgTHC estimates differ between those who prefer different quantity units (prefer hits or prefer grams) and determine the feasibility of obtaining individualized (i.e., participant-specific) GPHRs for improving hits-based mgTHC estimates. We conducted an online survey that required all participants to indicate their preferred quantity reporting unit (hits or grams) but estimate their consumption with both units, regardless of their preference. Aim 1 of the study was to determine whether those who prefer the grams unit report using more grams and more hits than those who prefer the hits unit. If those who prefer grams report greater consumption regardless of the measurement unit, it would help explain why grams-based mgTHC estimates were significantly larger than hits-based mgTHC estimates in our prior study. This, in turn, would support the unit-preference survey design for estimating mgTHC by suggesting that differences between grams-based and hits-based mgTHC estimates are at least partly due to genuine behavioral differences and not entirely due to measurement error.

Aim 2 was to further explore methods for improving estimates of GPHRs. Specifically, we sought to (1) document how the universal GPHR assumption masks the individual differences needed to improve hits-based mgTHC estimates; and (2) compare two approaches for estimating individualized GPHRs: a “two-item approach” and a “one-item approach”.

2. METHODS

2.1. Sample Recruitment

Adults aged ≥ 18 years living in the U.S. were recruited via Facebook/Instagram with cannabis-related advertisements and cannabis-related keyword-targeting from June 1^st, 2022, to July 9^th, 2022 (48). Those who clicked the advertisement were directed to the survey consent page. Those who reported age ≥ 18 and consented to participate were permitted to complete the survey. The Dartmouth Committee for Protection of Human Subjects approved all study procedures. No compensation was provided for participation.

A total of n=3,658 clicked the advertisement link, of whom n=4 were ineligible (age <18). Survey data quality checks included reCAPTCHA verification (n=6 responses excluded), metadata to detect multiple submissions (n=22 responses excluded), items only visible to bots (n=0 responses excluded). Those reporting incompatible past 30-day and past 7-day cannabis use frequencies (n=21) were excluded (e.g., a participant who reports using on all 30 days within the past 30 days but also reports using on fewer than 7 days within the past 7 days). The primary analyses also excluded n=792 participants who reported at least one inconsistent pair of hits and grams quantities (e.g., reported one hit in the morning but zero grams in the morning for the same day). Additionally, n=319 did not consume cannabis in the past week, and n=113 did not indicate whether they consumed in the past week. Thus the analytical sample for this study included n=2,381 participants who reported using a cannabis product at least once in the past seven days.

2.2. Survey Design

The survey included 59 items designed using prior literature (12,38,48,49). Items assessed sociodemographics, frequency of use (past 30-day and past 7-day), and past 7-day use of multiple product types and methods of administration (smoking flower, vaping concentrates, vaping flower, dabbing concentrates, edibles, tinctures, and capsules, all yes/no items) and consumption quantity. Details on additional survey items are provided below.

2.3. Relevant Survey Items and Calculations

2.3.1. Quantity unit preference

Participants were first asked whether they prefer reporting their consumption as “number of hits / puffs / tokes per day” or as “number of grams per day”, separately for each product used (flower, concentrate).

2.3.2. Consumption quantity (Figure 1)

Figure 1.

Survey items used to assess grams and hits quantities for cannabis flower and concentrates.

For each product used, participants were asked whether they used the same total amount on each day they used in the last seven days. Those who responded “yes” reported their typical number of hits and grams on using days; those who answered “no” reported the number of hits and grams consumed on their most recent using day. Using the four time-of-day quadrants (morning, afternoon, evening, night), participants reported their typical or most recent quantities twice: once using the “hits” unit, and once using the “grams” unit (Figure 1). To control for item order effects, participants were randomized (0.5 probability) to one of two conditions: (a) report quantities in hits then grams, or (b) report quantities in grams then hits.

Flower and concentrate quantities were reported as number of hits in each of the four time-of-day quadrants of the day (response options: 0 to 10 hits in increments of 1 hit, 11–15 hits, 16–20 hits, 21–25 hits, 26–30 hits, > 30 hits) and number of grams (response options: 1/16g, 1/8g, 1/4g, 1/2g, 3/4g, 1g, 1 1/4g, 1 1/2g, 1 3/4g, 2g, > 2g) (Figure 1). The responses to these items were used to calculate mgTHC (Table 1). Of note, we recoded interval hits response options using the middle value. For example, 11–15 hits was coded and analyzed as 13 hits.

Table 1.

Formulas used to calculate milligrams of THC (mgTHC; corresponds to Figure 3)

Quantity Unit	Calculation of mgTHC per using day
Hits: Flower	Total hits × Potency × 1000 × MAEC × 0.06*
Hits: Concentrate	Total hits × Potency × 1000 × MAEC × 0.012*
Grams	Total grams × Potency × 1000 × MAEC

^*These values are universal grams-per-hit ratios derived from Lynch et al. 2021; Varlet et al. 2016;

MAEC = “Method of administration efficiency constant”; See Budney et al. 2022 for details.

2.3.3. Product Potency

Participants estimated the %THC of their flower products, pre-filled concentrate vape cartridges, and other concentrate (oils, wax, shatter, crumble) products. Flower %THC response options ranged from 0 to 30. Concentrate %THC response options ranged from 40 to 100.

2.3.4. Calculating mgTHC using universal GPHRs (Table 1)

mgTHC was calculated using the formulas for hits and grams developed in our previous study (37) (Table 1). To replicate our prior study procedures and results, all hits-based mgTHC estimates were calculated using the universal GPHRs of 0.06 for flower products and 0.012 for concentrate products. Of note, the mgTHC formulas also include a method of administration efficiency constant (i.e., “MAEC”) to account for loss of THC due to the method of administration, such as the continuous burning of a joint (50). Additional details about the MAEC and other aspects of these formulas are provided in Budney et al. (37).

2.3.5. Individualized GPHRs (Table 2)

Table 2.

Survey items and calculations used to estimate grams-per-hit ratios (GPHR)

Approach	Example Survey Items	GPHR Calculation*
Two-Item Approach	Item 1 (Figure 1A): “On the days that you used buds in the last 7 days, approximately how many grams of bud did you typically use during each time of day?” Item 2 (Figure 1C): “On the days that you used buds in the last 7 days, approximately how many hits of bud did you typically use during each time of day?”	Total grams per day ÷ Total hits per day
One-Item approach	Item: “How many total hits would it take you to finish 1/2g of your bud by smoking?”	0.5 grams ÷ # hits to finish 0.5 grams

^*Note: Respondents reported the number of grams they typically consume within each of the four time-of-day quadrants (morning, afternoon, evening, night). The total number of grams consumed per day is the sum of the responses for these four times of day. For example, reporting use of 0.5 grams in the morning, afternoon, evening, and night = total of 2 grams per day. The same procedure is used to calculate total hits per day.

The universal GPHRs (0.06 and 0.012) assumed in Table 1 mask between-subject variability of individual GPHRs, which reduces the accuracy of mgTHC calculations. To explore alternatives to the universal GPHR assumption, we examined the feasibility of obtaining individualized (i.e., participant-specific) GPHRs using a higher-burden “two-item approach” and a lower-burden “one-item approach”. Summaries of GPHR calculations for these approaches are provided in Table 2.

2.3.5.1. “Two-item approach” GPHRs (Figure 1, Table 2)

We calculated a unique GPHR for each participant by dividing total grams per using day (i.e., items in Figures 1A or 1B) by total hits per using day (i.e., items in Figures 1C or 1D). Because this approach requires participants to provide multiple, repetitive, fine-grained reports of the same behavior using different units, it presumably yields more accurate GPHRs but also imposes a high response burden on participants.

2.3.5.2. “One-item approach” GPHRs (Table 2)

We also tested a “one-item approach” by asking participants, “How many total hits would it take you to finish 1/2g of your [product type] by [method of administration]?” For example, participants were asked “How many total hits would it take you to finish 1/2g of your bud by smoking?” Responses were provided using a visual analog scale (range: 1 to 100 hits in 1-hit increments). This item assessed flower and concentrate products separately and displayed an image of ½ gram of flower or concentrate next to a U.S. quarter to aid participants’ estimation. We used this item to estimate individualized GPHRs by dividing 0.5 grams by each participant’s response to the item (i.e., 0.5 grams divided by the number of hits the participant believes is required to consume 0.5 grams).

2.4. Analyses

2.4.1. Aim 1 analyses

The first set of analyses compared consumption patterns of the two unit-preference subgroups (prefer hits vs. prefer grams). We used quantile regression to compare the two unit-preference subgroups on median values of seven consumption outcomes: total grams of (1) flower and (2) concentrate; total hits of (3) flower and (4) concentrate (Figure 2, Table 3); potencies (%THC) of (5) flower, (6) prefilled vaporizer concentrates, and (7) other concentrate (e.g., wax, shatter, crumble)(Table 3). Finally, we compared the two unit-preference subgroups on median values of (1) hits-based mgTHC estimates, and (2) grams-based mgTHC estimates (Figure 3, Table 3). Using participants’ preferred unit as a predictor in these models allowed us to determine if actual differences in reported consumption behaviors between unit-preference subgroups contribute to the discrepancies between hits-based and grams-based mgTHC estimates observed in our prior study. All regression models included robust standard error corrections. Because the analyses aimed to clarify survey item functioning, rather than develop a complex prediction model, we opted for greater interpretability by using simple models that only adjusted for randomized item sequence.

Figure 2.

Consumption quantity differences between unit preference subgroups. Each of the four graphs compares the distribution of quantities reported by those who prefer the hits unit to the distribution of quantities reported by those who prefer the grams unit. A hollow circle indicates a participant’s reported quantity, and a solid black horizontal line indicates the median reported quantity. Comparing the two median values within each of the four graphs demonstrates that those who prefer the grams unit reported larger quantities – regardless of the unit (hits or grams) used to report those quantities. * indicates statistically significant (p<0.05) difference between median values. Green graphs (left side) display reported flower quantities; orange graphs (right side) display reported concentrate quantities. Graphs in the top row display reported number of hits; graphs in the bottom row display reported number of grams.

Table 3.

Consumption pattern differences between unit-preference subgroups

	Subgroup: Prefer Grams Unit	Subgroup: Prefer Hits Unit	Difference between subgroups
	Median (IQR)	Median (IQR)	Adjusted Quantile Regression Q2 $\widehat{\mathbf{\beta }}$, (95% CI)
Quantity (corresponds to Figure 2)
# Hits: Flower	23 (29)	9 (11)	14, (12.3, 15.7)*
# Hits: Concentrate	12 (15.5)	6 (9)	4, (1.8, 6.2)*
# Grams: Flower	1.5 (2.3)	0.5 (0.9)	1.1, (0.9, 1.2)*
# Grams: Concentrate	0.31 (0.56)	0.19 (0.25)	0.13, (0.03, 0.22)*
Potency (%THC)
Flower	22% (5)	20% (6)	2, (1.6, 2.4)*
Concentrate cartidge	85% (18)	78% (22)	6, (1.9, 10.1)*
Open concentrate	80% (12)	80% (16)	1, (−2.7, 4.7)
mgTHC (corresponds to Figure 3)
mgTHC via flower hits	91 (128)	32 (44)	59.2, (51.8, 66.6)*
mgTHC via concentrate hits	59 (78)	29 (44)	29.0, (15.4, 42.6)*
mgTHC via flower grams	113 (185)	27 (62)	83.9, (72.6, 95.2)*
mgTHC via concentrate grams	129 (228)	61 (98)	68.1, (42.5, 93.6)*

Notes: mgTHC = “milligrams of THC”; Quantity and mgTHC values refer to total amount consumed on a single using day (e.g., total number of hits on using days); Model is only adjusted for item presentation order randomization group (i.e., report hits then grams vs. report grams then hits)

^*p<0.05

Figure 3.

Milligrams THC (mgTHC) calculated using hits or grams responses for different unit preference subgroups. Each of the four graphs compares the mgTHC distribution of those who prefer the hits unit to the mgTHC distribution of those who prefer the grams unit. Graphs on the left display mgTHC estimates calculated using participants’ reported number of hits; graphs on the right display mgTHC estimates calculated using participants’ reported number of grams. A hollow circle indicates a participant’s reported quantity, and a solid black horizontal line indicates the median mgTHC estimate. Comparing the two median values within each of the four graphs demonstrates that mgTHC estimates among those who prefer the grams unit are larger than mgTHC estimates among those who prefer the hits unit – regardless of the unit (hits or grams) used to calculate mgTHC estimates. * indicates statistically significant (p<0.05) difference between median values. Green graphs (top row) display mgTHC estimates based on reported flower quantities and potencies; orange graphs (bottom row) display mgTHC estimates based on reported concentrate quantities and potencies.

2.4.2. Aim 2 analyses

The second set of analyses compared GPHRs derived using the “two-item”, “one-item”, and “universal” approaches. We used descriptive statistics to examine the extent to which the universal GPHR assumption obfuscates the variability of individualized GPHRs for flower and concentrate products. To determine if the one-item and two-item approaches yield similar GPHRs, we used a one-sample T-test to determine whether the difference between the two ratios for each study participant (i.e., within-subject) differed from zero (H₀: μ=0, H_a: μ≠0) (Figure 4). Results demonstrating that both approaches yield similar estimates would justify using the low-burden one-item approach in future studies.

Figure 4.

Within-subject difference of grams-per-hit ratio (GPHR) estimates: two-item approach estimate minus one-item approach estimate. Two GPHR estimates were calculated for each participant using different methods. Graphs in Figure 4 display the distribution of values obtained by subtracting each particiant’s two GPHR estimates. The distributions are primarily centered around zero – suggesting that the two methods yield similar GPHR estimates.

3. RESULTS

3.1. Demographics, cannabis use patterns, and quantity unit preferences.

The mean age of the analysis sample was 39.0 (SD, 16.4) years, and 50% reported female sex. The sample was primarily White (85%), 89% identified as non-Hispanic, 49% were employed full-time, and 38% had completed at least a bachelor’s degree. The average age of cannabis use onset was 16.7 (SD 5.6) years. Most participants (77%) reported using daily in the past week and 28 or more days in the past month (69%). Approximately 85% reported their past week as “typical” of their cannabis consumption. Additionally, using approximately the same amount of product on each using day was reported by 65% and 56% of those who used flower and concentrates, respectively. Regarding “unit preference”, 45% and 80% preferred the hits unit for reporting flower and concentrate consumption, respectively. The average reported %THC potency was 21.3% (SD 5.1%) for flower, 75.6% (SD 15.7%) for concentrate cartridges, and 77.0% (SD 13.7%) for other concentrate products.

3.2. Aim 1 results: unit-preference subgroup comparisons.

3.2.1. Consumption quantity differences between unit-preference subgroups (Table 3, Figure 2).

Regardless of the unit used to assess quantities, those who preferred the grams unit reported significantly larger median quantities (number of hits or number of grams) than those who preferred the hits unit for flower products (23 hits vs. 9 hits; 1.5 grams vs. 0.5 grams, respectively) and concentrate products (12 hits vs. 6 hits; 0.31 grams vs. 0.19 grams, respectively).

3.2.2. Potency differences between unit-preference subgroups (Table 3).

Participants who preferred the grams unit reported significantly greater median potencies compared to those who preferred the hits unit for cannabis flower products (22% vs. 20%) and concentrate vaporizer cartridges (85% vs. 78%) (Table 3).

3.2.3. Differences in mgTHC between unit-preference subgroups (Table 3, Figure 3).

Those who preferred the grams unit had significantly greater median mgTHC than those who preferred the hits unit, regardless of the unit (hits or grams) or product (flower or concentrate) that was used to calculate mgTHC (Table 3, Figure 3).

3.3. Aim 2 results: Individualized GPHRs.

3.3.1. Variability of individualized GPHRs.

GPHRs varied substantially among participants. The median (IQR) of participant-specific GPHRs were 0.065 (0.092) and 0.05 (0.075) for flower products and 0.029 (0.047) and 0.01 (0.017) for concentrate products, using the two-item approach and one-item approach, respectively. To contextualize this variability, consider that regardless of product type (flower or concentrate) or estimation approach (one-item or two-item approach), at least 45% of participant-specific GPHRs were more than 50% larger or 50% smaller than the universal ratios (0.06 or 0.012).

3.3.2. Individualized GPHRs: one-item approach vs. two-item approach (Figure 4).

Subtracting the one-item approach GPHR from the two-item approach GPHR for each participant produced the distributions presented in Figure 4. T-tests indicated that the two-item approach produced slightly larger GPHRs than the one-item approach for flower (Mean: 0.021; 95% CI: 0.014, 0.027) and for concentrates (Mean: 0.013; 95% CI: 0.008, 0.018). Although these differences were statistically significant, the distributions in Figure 4 suggest a notable within-subject consistency between the two-item and one-item approaches.

4. DISCUSSION

This study used an online convenience sample of primarily daily cannabis consumers to examine “grams” and “hits” as units for measuring cannabis quantities and estimating milligrams of THC (mgTHC). Findings indicated that those who preferred to report quantities with the grams unit were heavier consumers than those who preferred the hits unit, which translated into larger mgTHC estimates among those who preferred the grams unit. Additionally, results suggest that the standard assumption of a universal grams-per-hit ratio (GPHR) masks substantial individual differences in GPHRs and likely reduces the accuracy of mgTHC estimates. However, results also suggest that it is possible to obtain participant-specific GPHRs with a simple, low-burden survey item.

Standardizing how cannabis consumption is measured and reported is now a central task for researchers (13). Despite the growing heterogeneity of cannabis products, most products still expose consumers to Δ9-THC, and thus standardized measurements and cross-product comparisons are achievable. Consequently, the field is moving toward a five-milligram “standard THC unit” (14) to facilitate comparison and communication of scientific results. However, researchers cannot adopt a five-milligram standard THC unit if they cannot measure milligrams of THC. Labels for popular flower and concentrate products do not communicate milligrams of THC directly to consumers. Thus researchers need alternative quantity measurement strategies. The present study sought to fill this gap by evaluating hits and grams as units for measuring flower and concentrate quantities and deriving milligrams of THC.

The unit-preference survey design allows respondents to choose the unit (e.g., hits unit or grams unit) for reporting their consumption quantities. The advantage of this approach is that it accommodates the growing diversity of cannabis product types, methods of administration, and use patterns. However, the disadvantage is that preferences for quantity units seem to introduce self-selection effects into the data. We encountered this self-selection in our prior study, which made it difficult to determine whether observed differences in mgTHC reflected actual differences in consumption or instead reflected differences in how consumption was measured (37). To resolve this issue, the present study required all participants to report quantities with both the hits and grams unit. We found that those who preferred the grams unit reported consuming larger quantities of more potent cannabis than those who preferred the hits unit – regardless of the unit used to report quantity. These behavioral differences translated into larger mgTHC estimates among those who prefer the grams unit. This finding supports the unit-preference survey design for estimating mgTHC by suggesting that measurement error (e.g., different psychometric properties of grams vs. hits units or differences in mgTHC formulas) is not entirely responsible for observed differences between grams-based and hits-based mgTHC estimates.

Formulas that convert “hits” quantities into mgTHC require a GPHR. Traditionally, researchers have assumed a single, universal GPHR (i.e., universal “hit size”) for all participants. However, we estimated individualized GPHRs for each study participant and observed a wide distribution of values – many of which deviated by more than 50% from the universal GPHR. This result is consistent with studies of cannabis inhalation topography (44,46) and indicates that universal GPHRs mask variability among consumers, which in turn, may reduce the accuracy and predictive utility of mgTHC estimates. Thus, future efforts to estimate population-level mgTHC consumption using the unit-preference survey design should consider estimating person-specific GPHRs.

The two approaches used to estimate person-specific GPHRs (“one-item approach” and “two-item approach”) in this study required different amounts of effort from participants. For a given product type (e.g., flower or concentrates), the two-item approach required eight separate responses (number of hits and grams in the morning, afternoon, evening, and night). Conversely, the one-item approach only required one response to a question with the following structure: “How many total hits would it take you to finish 1/2g of your [product type] by [method of administration]?” Although we cannot determine the amount of time participants spent answering individual items, we believe it is reasonable to assume that this one-item approach places less burden on participants than the two-item approach. Notably, both approaches yielded highly similar within-subject values despite requiring different amounts of effort from participants. We believe that this finding supports using the lower-burden one-item approach to estimate individualized GPHRs in future cannabis surveys to facilitate more accurate mgTHC estimates. However, these items should be used with caution as they have only been examined in a single study. Additional testing is needed to confirm psychometric properties and identify limitations.

Relatedly, it is important to remember that hits must be converted into grams to calculate mgTHC. Thus, the two-item approach yields redundant information; there is no need to convert hits into grams if the participant has already reported their grams. Similarly, when using the two-item approach, multiplying a participant’s total hits and GPHR will simply reduce to the original total grams responses, i.e., total hits × (total grams ÷ total hits) = total grams. This is another reason why using the one-item approach to estimate GPHRs may be advantageous.

There are several limitations to consider when interpreting the study results. First, the sample was recruited using social media and was comprised primarily of heavy cannabis consumers who were white. Thus, the observed interrelations among aspects of cannabis use patterns, differences among groups, and specific point estimates may be less applicable to low-frequency consumers or consumers from different racial or ethnic backgrounds. Results from this study need to be replicated with multiple heterogenous samples obtained using other sampling methods. Additionally, the study data are based on self-report, were not biochemically verified, and are likely affected by various forms of measurement error such as imperfect memory and recall, self-selection and social desirability effects, as well as overestimation and digit-preference biases (27,51,52). For example, the three smallest response options for reporting typical concentrate quantities were 1/16^th of a gram, 1/8^th of a gram, and 1/4^th of a gram. It is possible that participants had difficulty estimating such small amounts or that the denominations separating the response options were too wide. Additionally, we re-coded interval hits response options using the middle value (e.g., 11–15 hits re-coded as 13 hits). These types of issues may have compromised the accuracy of results. Thus we encourage readers to view these results as useful approximations with face validity rather than exact estimates.

Cannabis legalization also warrants comment. This study examined the relationship between preferences for cannabis quantity units and cannabis consumption. Legalization could impact this relationship by altering consumer preferences, perceptions, norms, knowledge, and behaviors (53). For example, increased availability and popularity of specific products or product packaging and labeling regulations could impact individuals’ preferred unit for reporting consumption quantities. Determining whether cannabis legalization can have such effects will require a thorough examination of the heterogenous legal cannabis laws among US States.

Researchers encounter many critical decision points when designing surveys to measure the various dimensions of cannabis use. The present study aimed to help researchers navigate portions of this decision-making process by shedding light on procedures for measuring cannabis quantities and the consequences of those procedures. Keeping the study limitations in mind, tentative recommendations include: (1) use detailed visual aids of different cannabis quantities and multiple reference objects (e.g., paper and coin currency, lighter) to help participants estimate quantities (images used in this study can be made available upon request); (2) if using the “hits” unit to measure quantities, consider also deriving individualized GPHRs (i.e., individualized “hit sizes”) by employing the “one-item approach” outlined in this study (see Table 2); (3) if survey participants can choose their preferred quantity unit (e.g., hits unit or grams unit) when reporting consumption, analyses should explore correlations between a unit preference indicator and patterns of consumption to consider how these relations may impact statistical models and interpretation of results. Note that these are tentative suggestions based on early-stage results that require further refinement and testing. However, with continued effort in this area of research, it may be possible to develop the next generation of best practices for measuring cannabis consumption in the population.