Background

Cannabis use is an increasing global health concern [1]. Although cannabis use may effectively address certain medical conditions [2], it may have negative health effects (e.g., immune function [3], respiratory function [4], mental health [5]) and socioeconomic (e.g., academic, employment) outcomes [6,7,8] and disproportionately impacts certain subpopulations (e.g., men, sexual minorities, racial/ethnic minorities) [9,10,11,12,13].

Despite these concerns, the global cannabis market has dramatically grown [14]. North American accounts for 96.8% ($22 billion) of the global legal cannabis market, with the US accounting for $20 billion (Canada $2 billion). Israel accounts for ~ 22% of the remainder of the market (second only to Germany, ~ 28%) [14]. Moreover, the US and Israel represent countries with the highest proportion of adults who consume cannabis [14]. Per 2020 data, 17.4% of US adults and 27.0% of Israeli adults reported past-year cannabis use [14]. Cannabis legislation in both countries has markedly changed over the past decades. As of November 2022, 38 states in the US have legalized medical cannabis [15]. Additionally, 21 states and the District of Columbia legalized non-medical (‘recreational’) cannabis use for adults (≥ 21 years-old) [15], the first of which date back to 2012 (Colorado, Washington) [15]. In Israel, in 2011, the Israel Medical Cannabis Agency was established to regulate medical cannabis [16], and non-medical cannabis use for adults (≥ 18 years-old) was decriminalized in 2019 [17, 18]. In 2020, 2 bills to allow non-medical cannabis sales passed preliminary readings in Israel’s parliament [19]. In 2021, another bill to legalize non-medical cannabis use was approved by the Ministerial Committee for Legislation, but rejected by parliament [20].

As legislative contexts have shifted in these 2 distinct countries with high cannabis use rates, cannabis-related perceptions and use-related behaviors are likely to change [21,22,23,24,25,26,27]. For example, cannabis legalization may be associated with more favorable perceptions (e.g., lower perceived risk, greater social norms) and increased use intentions among young adults [21], higher use prevalence and levels among youth [22, 28], greater use among adults with children living in the home [29], and changes in use motives (e.g., recreational vs. medical), modes of use, and product source among those using cannabis [13, 28, 30,31,32]. However, the evidence is mixed [33].

Understanding how adults address cannabis use within their homes or around children is critical in mitigating use-related risks among both adults and young people. One important consideration is the restrictiveness of home environments. For example, the literature regarding tobacco suggests that allowing use in the home is associated with greater secondhand smoke exposure and cigarette consumption, while prohibiting use is associated with lower consumption, more attempts to quit use, and higher quit rates [34]. Notably, children who live in homes that allow smoking are more likely to initiate smoking themselves [34]. Additionally, the literature underscores the importance of parenting, including parental substance use and monitoring, in influencing youth substance use [35, 36], including cannabis [37]. This is important given the literature indicating increased cannabis use among adults with children in the home post-legalization [29] and increased use among youth and young adults [21, 22, 28, 38].

Social Cognitive Theory (SCT) [39] highlights the dynamic interplay of one’s social context (e.g., social roles, social norms) and cognitions such as outcome expectancies (e.g., perceived risks or consequences vs. benefits) in relation to one’s behaviors. For example, an individual’s perceptions of how socially acceptable or normative cannabis use is likely depends on the nature and extent of their exposure to use within their social networks, and these perceptions ultimately impact their personal use [9,10,11,12, 37, 40,41,42]. Additionally, the expected outcomes of use – either positive or negative – influence whether one uses cannabis; if one expects positive outcomes, like a pleasurably psychological experience, they are more likely to use, while if one expects negative outcomes, like health risks or addiction, they may be less likely to use [9,10,11,12, 37, 40,41,42]. Moreover, if one perceives negative implications of others (e.g., family members or children in the home) being exposed to cannabis use (e.g., health risks, enticing youth), they may be more likely to implement restrictions about cannabis use in the home or in the presence of children. Indeed, in the tobacco literature, SCT has proven to be a useful framework for understanding the implementation of smoke-free homes to protect children and nonsmokers [43,44,45,46,47].

While SCT provides a useful model for identifying factors associated with substance use related outcomes and cannabis use outcomes [9,10,11,12, 37, 40,41,42], existing literature regarding how adults address cannabis use within their homes or in the presence of children is limited. Furthermore, little cross-country research has examined theory-based constructs related to cannabis-related outcomes in differing sociopolitical contexts.

To advance the literature and inform regulatory and prevention efforts, this mixed-methods study identified: 1) theory-based correlates (risk perceptions, social norms) of past-month cannabis use, next-year use intentions, and intentions to use in the home or near children if non-medical cannabis was legal among US and Israeli adults; and 2) qualitative themes regarding perceptions of cannabis use and policies.

Methods

We analyzed data from a study of US and Israeli adults that used a sequential explanatory mixed-methods design and primarily focused on tobacco-related perceptions and use [48]. Eligibility criteria included: 1) age 18–45 years; and 2) able to speak English (US), or Hebrew or Arabic (Israel); in Israel, an additional criterion was having an Israeli ID. The study received ethical approvals from George Washington University (NCR213416) and Hebrew University (27062021). The current study analyzed: 1) cross-sectional survey data (collected in October-December 2021); and 2) semi-structured interviews (conducted in Spring 2022). This study adhered to STROBE guidelines for cross-sectional quantitative research and COREQ guidelines for qualitative research.

Quantitative Data

Participants

The US survey was conducted primarily using KnowledgePanel®, a probability-based web panel designed to be representative, recruited via random-digit dialing and address-based sampling, and incentivized via KnowledgePanel® points redeemable for cash. This approach was supplemented with off-panel participant recruitment (via banner ads, web pages) to meet subgroup recruitment targets (i.e., Asian individuals reporting tobacco use). Of 4,960 panelists recruited, 2,397 (48.3%) completed eligibility screening, and 1,095 (45.7%) completed the survey; of 353 off-panel individuals screened, 33 (9.3%) were eligible and completed the survey. The Israeli survey was conducted using opt-in sampling, as described above. Of 2,970 individuals screened and eligible, 1,094 (36.8%) completed the survey.

Measures

The survey took ~ 25 min to complete and was professionally translated to Hebrew and Arabic for Israeli participants.

Outcomes: Cannabis use, use intentions, and intentions to allow use in the home and/or near children if legalized

Participants were asked, “How old were you when you first used marijuana?” including the option of “I have never used marijuana” [49]. Those reporting lifetime use were asked, “During the past 30 days, on how many days did you use marijuana?” [49]. Among all participants, we assessed cannabis use intentions by asking, “How likely are you to try or continue to use marijuana in the next year?” (1 = not at all to 7 = extremely). We also asked, “If marijuana were legalized for recreational use, how likely would you be to allow marijuana use: in your home? in the presence of children?” (1 = not at all to 4 = very); responses to these 2 items were averaged to create an index score.

Cannabis use characteristics

Participants reporting lifetime use were asked, “How have you used marijuana in the last 12 months?” [50]. Response options (e.g., vaped in liquid form) are shown in Table 1. Those reporting past 30-day (current) use were asked, “Which is the one method you used most in the last 12 months?” with the same response options [50]. Those reporting lifetime use were also asked, “For which of the following reasons do you primarily use marijuana? recreational purposes; medical purposes; or both recreational and medical purposes” and “In the last 12 months, where have you most often obtained marijuana?” (response options in Table 1) [50]. Those reporting current use also reported times used per day on days used.

Table 1 Participant characteristics and cannabis-related factors among US and Israeli participants
Full size table

Perceived risk and social norms

To assess SCT-related social-cognitive constructs related to outcome expectancies [39], we assessed perceptions of risk and social norms. To assess perceived risk, participants were asked, “How [addictive; harmful to your health] do you think marijuana is?” (1 = not at all to 7 = extremely) [51]; responses were averaged to create an index score. To assess social norms, participants were asked, “Please indicate the extent to which people important to you would (or do) approve of you using marijuana?” (1 = all disapprove to 7 = all approve) and “How many of your closest connections (including your partner, friends, relatives, co-workers, and others) use marijuana?” (1 = none to 7 = all) [51]; responses were averaged to create an index score.

Sociodemographic covariates

We included country, age, sex, education, marital status, and children in the home. (Note: Given differences in state cannabis legalization in the US, participants were also coded as residing in a state with legal non-medical cannabis versus not to explore this factor in US-specific preliminary analyses.)

Data analysis

First, descriptive and bivariate analyses (Chi-square for categorical variables, t-tests or ANOVAs for continuous variables) characterized participants across countries with regard to sociodemographics and cannabis use characteristics and outcomes. Next, bivariate analyses assessed correlates of interest in relation to cannabis use, use intentions, and intentions to allow use in the home/near children if legalized, separately. Then, multivariable regression analyses examined correlates of these 3 outcomes (binary logistic regression for current use, linear regressions for use intentions and intentions to allow use in the home/near children if legalized). Country, sociodemographics, and perceived risk and social norms were included in the models; current cannabis use was also included in the models predicting use intentions and intentions to allow use in the home/near children if legalized.

Exploratory analyses also assessed country-specific models, which yielded similar findings to the overall models; further, no significant interactions between country and correlates of interest were found in relation to any outcome. We also examined US state non-medical cannabis legalization in relation to cannabis use characteristics and outcomes among US participants. Those in legalized states reported greater social norms, lifetime use, and legal sources; no other differences were found, and state legalization were not significantly associated with cannabis-related outcomes in US-specific models. Thus, we presented the overall models, including country as a covariate. Quantitative analyses were conducted by SPSS (26.0), using alpha = 0.05.

Qualitative data

Participants

Participants in both countries were purposively recruited for representation by sex and race/ethnicity. In the US, participants were recruited from the online survey sample and were called and/or emailed an invitation to participate. In Israel, the opt-in sample for the online survey precluded our ability to re-contact survey participants; instead, we promoted the study via ads on Facebook; potential participants were provided a study description, consented, and screened for eligibility (i.e., ≥ 18 years old, speak Hebrew or Arabic).

Assessment

Interviews were guided by standard principles of qualitative methods. [52, 53] Each interview was conducted online via Zoom in English (US) or Hebrew/Arabic (Israel; participant’s choice), audio-recorded, ~ 45 min long, and incentivized (USD$25 or 100 NIS). The interview guide included various questions about cannabis and tobacco use. Questions relevant to the current study assessed perceptions of cannabis use (e.g., “If marijuana were legalized for recreational use, how likely would you be to allow marijuana use in your home? Why or why not?”) and cannabis policies (e.g., “What do you think about recreational marijuana policies?”). Interviews were transcribed by a professional transcription service.

Qualitative analysis

Qualitative data were analyzed using standard principles of qualitative methods [52,53,54] and deductive-inductive thematic analysis. [55] A preliminary set of deductive codes were compiled based on the interview guide and a preliminary review of the US-based transcripts. Then, a subsample of 8 US transcripts and 8 translated Arabic and Hebrew interviews (n = 4 each) were independently reviewed by 2 US-based coders and 2 Israeli-based coders (1 from each team per transcript). An iterative process was used to assess inter-rater reliability, reach consensus, inform revisions, and yield additional codes based on emergent themes. [55] These codes were compiled into a codebook. After ensuring sufficient inter-rater reliability (> 80%), the remainder of the interviews were coded. Representative quotes were selected for inclusion in the manuscript.

Results

Quantitative results

Participant characteristics

In this sample (N = 2,222; US n = 1,128, Israel n = 10.94), participants were an average age of 32.19 (SD = 7.74), and 50.3% were female; further, 36.4% reported lifetime cannabis use and 16.7% current use (Table 1). The average score for next-year use intentions was 2.07 (SD = 1.95, scale: 1–7), with 70.5% indicating “not at all likely”. Average scores for intentions to use (if legalized) in the home and in the presence of children were 1.87 (SD = 1.15, scale: 1–4) and 1.30 (SD = 0.73), with 56.3% and 82.6% indicating “not at all likely”, respectively.

Bivariate results characterizing factors associated with current use, use intentions, and intentions to allow use in the home or near children if legal are shown in Table 2. Participants residing in the US (vs. Israel) more likely reported lifetime (49.9% vs. 22.5%) and current use (22.0% vs. 11.2%) and reported lower risk perceptions and greater social norms (p’s < 0.001).

Table 2 Bivariate analyses examining participant characteristics in relation to past 30-day cannabis use, intentions to use, and intentions to allow use in the home or near children if legalized among US and Israeli adults, N = 2,222
Full size table

Cannabis use and related characteristics

In multivariable regression analyses (Table 3), correlates of current cannabis use included lower perceived risk (aOR = 0.80, 95%CI = 0.74, 0.87) and greater norms (aOR = 2.39, 95%CI = 2.17, 2.63), as well as being male (aORfemale = 0.6, 95%CI = 0.52, 0.93) and sexual minority (aOR = 1.67, 95%CI = 1.15, 2.40; p’s < 0.05; Nagelkerke R-square = 0.432). Bivariate analyses (Table 1) indicated that, among participants reporting lifetime use, US (vs. Israeli) participants reported younger first age of use, more likely using via smoking without tobacco, vaping in liquid form, dabbing/concentrates, and edibles, but less likely via smoking with tobacco or vaping dried leaves or herbs (p’s < 0.05). US participants reporting lifetime use were more likely using primarily recreationally and obtaining it from retailers (p’s < 0.05). Among those reporting current use, US participants reported more days of use (p < 0.001).

Table 3 Multivariable regression models examining correlates of past 30-day cannabis use, intentions to use cannabis, and intentions to allow use in the home or near children if legalized among US and Israeli adults, N = 2,222
Full size table

Cannabis use intentions

In multivariable regression (Table 3), correlates of greater use intentions included lower perceived risk (B = -0.04, 95%CI = -0.07, -0.01), greater norms (B = 0.43, 95%CI = 0.39, 0.47), and current cannabis use (B = 2.80, 95%CI = 2.63, 2.96), as well as being from Israel (B = 0.13, 95%CI = 0.02, 0.24) and male (Bfemale = -0.17, 95%CI =—0.27, -0.06; p’s < 0.05; Adjusted R-square = 0.629).

Intentions to allow use in the home and near children if legalized

In multivariable regression analysis (Table 3), correlates of intentions to allow use in the home or near children if legal included lower perceived risk (B = -0.07, 95%CI = -0.09, -0.06), greater norms (B = 0.21, 95%CI = 0.19, 0.23), and current use (B = 0.45, 95%CI = 0.37, 0.54), as well as being from Israel (B = 0.23, 95%CI = 0.17, 0.29), male (Bfemale = -0.06, 95%CI = -0.12, -0.01), and more educated (B = 0.06, 95%CI = 0.001, 0.12; p’s < 0.05; Adjusted R-square = 0.391).

Qualitative results

US participants were 36.5 (SD = 6.3) years-old, 42.5% female, 32.5% non-Hispanic (NH) White, 32.5% NH Black, 12.5% NH Asian, and 22.5% Hispanic; 52.5% reported current cannabis use. Israeli participants were on average 29.35 (SD = 6.2) years old, 52.3% female, 88.6% Jewish, and 11.4% Arab. Several themes emerged with regard to perceived risks and benefits of cannabis use and non-medical cannabis legalization (Table 4).

Table 4 Qualitative findings regarding perceptions of cannabis use and non-medical legalization among US and Israeli adults
Full size table

Perceptions of cannabis use

Most participants perceived cannabis use as prevalent (“Almost the entire world smokes it.” –27 year-old Arab female, Israel). Participants reported several concerns regarding potential risks of cannabis use. While a few commented that there is no need to shield children from cannabis use (“Children can be aware of what is happening in the world and around them. There is no need to conceal such information from them.” –38 year-old Jewish female, Israel), many reported concerns about using cannabis around children (“Even if it's legal, there's still some boundaries or respect that you still should have. I would not do that in front of others and in front of children.” –24 year-old Hispanic male, US). Some were concerned that using cannabis around children would socially normalize use among youth and expose children to cannabis byproducts (e.g., smoke). Another concern among many was use among young people, because of their developmental period and potentially leading to addiction and/or other drug use. Many also commented on cannabis’ general health risks. Relatedly, some reported concerns regarding driving under the influence of cannabis.

Regarding cannabis’ benefits, many participants commented on its potential to address physical and mental health symptoms. Some compared cannabis to opioids, alcohol, or tobacco, underscoring that they perceived cannabis to be less harmful than these other substances.

Perceptions of non-medical cannabis legalization

Some participants reported no substantial concerns about legalizing non-medical cannabis (“As long as [cannabis] is well regulated, I don't have concerns.” –39 year-old NH Asian female, US). However, some were concerned about the impact on society, including families, communities, the economy, and crime. Some also raised concerns about potential inequitable economic benefits (“People who've been selling weed for forever and getting in trouble for it are going to be left out of the opportunity to make money now when it's legal.” –37 year-old NH Black female, US). Another concern raised by a few participants was the price of cannabis products in a legalized market. Several also noted concerns about how cannabis products are marketed, for example, edibles appealing to young people and health claims in advertisements. Some also raised concern regarding possible insufficient regulation and quality control of legalized cannabis products.

Many commented on potential benefits of non-medical cannabis legalization (see Table 4 for example quotes). Many indicated the positive impact of decriminalization on society, including for communities disproportionately impacted by criminalization, as well as more broadly in terms of financial resources (e.g., revenue, employment, less enforcement costs). Some commented that the market might facilitate realization of cannabis’ potential medical benefits. Many also indicated that increased access to cannabis products through legal retailers (versus illegal sources) and the regulatory oversight of the legal cannabis market would reduce consumers’ overall risk in terms of products accessibility and contents. Many highlighted the importance of individual rights and freedom to choose to use. A couple suggested that legalizing cannabis use might reduce use among youth, by reducing its taboo or enforcing legal age limits.

Discussion

Despite differences in cannabis use and related perceptions among US and Israeli adults, theory-based factors, specifically perceived risks and social norms [39], may be important targets for interventions to mitigate use-related risks among adults and youth. Such intervention efforts are critical and timely, given high cannabis use rates in the US and Israel [14], the evolving cannabis legislative context in the US [15], Israel [16], and globally [1], the potential impact of cannabis legalization on cannabis use among young people [21, 22, 28] and adults [29], and the key roles of parents and the home environment in shaping youth cannabis use [34,35,36,37].

Lower perceived risk and greater perceived social norms were associated with current use, greater use intentions, and greater intentions to use in the home or near children if legal. Furthermore, although US participants more likely reported cannabis use and favorable perceptions, Israeli participants reported greater use intentions and intentions to use in the home or among children if legalized. Particularly noteworthy is that the variables included in models tested in this study accounted for ~ 40% to ~ 63% of the variability in these outcomes, with perceived risk and social norms significantly contributing to each of the models, beyond demographic factors and cannabis use. Thus, SCT-driven interventions targeting perceived risks and social norms may help adults determine how to address cannabis use within their homes or among children and potentially mitigate use-related risks among adults and young people.

Qualitative findings suggested mixed perceptions regarding the potential impact of cannabis exposure on youth, with some participants reporting no concerns but more participants being concerned due to the potential impact on social norms and health. SCT-driven interventions targeting constructs like outcome expectancies have been shown effective in promoting rules banning tobacco smoking in private settings like homes [43,44,45,46,47], which reduce cigarette consumption, promote quit attempts, increase quit rates, and reduce youth tobacco use initiation [34]. These interventions provide a basis for interventions targeting cannabis-related restrictions that may lead to favorable outcomes, such as abstinence or limited use among adults and youth.

Regarding sociodemographics, males more likely used cannabis and reported greater use intentions and intentions to allow use in the home or near children if legal; bivariate analyses also indicated that not having children was associated with these cannabis-related outcomes. Identifying as a sexual minority was also associated with cannabis use. These findings align with previous studies examining correlates of use [9,10,11,12,13] and extend them to other cannabis-related outcomes. Interestingly, those more educated reported greater intentions to allow use in the home or near children if legal, which warrants further study.

This study also documented various perceived risks and benefits regarding cannabis use and non-medical legalization. As in previous research, qualitative results indicated perceptions that cannabis is less harmful than other substances [42, 51] and has medical benefits [56], and that legalization has various societal benefits (e.g., positive economic impact) and promotes individual rights [57]. Additionally, participants had mixed perceptions regarding whether sufficient regulation may increase safety of use and the economic impact of non-medical legalization. Moreover, participants reported additional concerns (e.g., increased crime) and benefits (e.g., increased accessibility). Collectively, these perceptions are likely influenced by various factors, including exposure to anti- and pro-legalization media [58, 59], and should be considered in future efforts to effectively communicate about changes in policy and potential risks and benefits.

Regarding use characteristics, in this sample of US and Israeli adults, US participants more likely reported lifetime (~ 50% vs. ~ 23%) and current use (22% vs. ~ 11%), and more likely obtained cannabis from legal sources (e.g., retail) and used primarily recreationally and via forms alternative to smoking (e.g., vaping, dabbing, edibles). These findings may reflect differences in access, product types, and marketing in the US resulting from legalization (as nearly half of US states have legalized non-medical cannabis [15]), proximity to legal markets, and/or shifts in social norms that have resulted from legalization [13, 28, 30,31,32]. Compared to 2020 data indicating past-year use prevalence of 17% among US adults and 27% among Israeli adults [14], current results indicated lower use rates among Israeli participants (likely due to different assessment timeframes, i.e., past 30-day vs. past-year) but higher use rates among US participants; this may reflect the sample’s restricted age range [18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45], as 2021 national data indicated 19.6% past-year use, with rates highest among 18–25 and 26–49 year-olds (35.4% and 24.6%) [49].

Limitations

Despite study strengths (e.g., mixed-methods design, theory-based, cross-country), survey and interview findings may have limited generalizability, given the use of web panels and opt-in sampling to recruit the survey sample and the small sample size involved in the qualitative data collection (despite n = 84 considered a large sample for semi-structured interviews [60,61,62,63]); thus, these data are subject to selection bias and may not reflect all possible perspectives. However, our samples of survey and interview participants in each country were designed to ensure representation of the sexes, racial/ethnic groups, and tobacco use characteristics – across each subgroup (e.g., White females who reported current tobacco use vs. no use). Additionally, the cross-sectional design limits the ability to establish causal relationships between variables or assess changes over time; however, our hypotheses and analytic approach were driven by the existing literature and SCT [39]. Finally, self-report assessments of cannabis use and related characteristics introduces potential recall and social desirability biases. Cognizant of such concerns, study assessments were derived from existing published measures, neutrally worded, translated/back-translated in Israel, pilot tested for comprehension, and created to allow “refusal” to answer.

Conclusions

The relatively high rates of cannabis use in the US and Israel [14] and rapidly shifting cannabis legislation in the US [15], Israel [16], and globally [1] underscore the need and timeliness of intervention efforts to mitigate cannabis use-related risks among adults and youth. Despite differences in cannabis use and use characteristics across countries, theory-based factors, specifically perceived risk and social norms, were shown to be relevant potential targets for interventions to mitigate cannabis use-related risks among adults and youth in the US and Israel, highlighting the importance of theory-based research across differing sociopolitical contexts.