English HELP WITH ENGL ASSIGNMENTS DUE IN 24 HOURS - 48 HOURS
combs
retrieve11.pdf
Content analysis of cannabis vaping videos on YouTube
Carmen C. W. Lim1,2 , Janni Leung1,2 , Jack Yiu Chak Chung1,2 , Tianze Sun1,2 , Coral Gartner3,4 , Jason Connor1,5 , Wayne Hall1,5,6 , Vivian Chiu1,2 , Calvert Tisdale2 , Daniel Stjepanović1 & Gary Chan1
National Centre for Youth Substance Use Research, Faculty of Health and Behavioural Sciences, The University of Queensland, St Lucia, QLD, Australia,1 School of Psychology, Faculty of Health and Behavioural Sciences, The University of Queensland, St Lucia, QLD, Australia,2 School of Public Health, Faculty of Medicine, The University of Queensland, Herston, QLD, Australia,3 Queensland Alliance for Environmental Health Sciences, Faculty of Health and Behavioural Sciences, The University of Queensland, Woolloongabba, QLD, Australia,4 Discipline of Psychiatry, Faculty of Medicine, The University of Queensland, Herston, QLD, Australia5 and National Addiction Centre, Institute of Psychiatry, Psychology and Neuroscience, Kings College London, London, UK6
ABSTRACT
Background and Aims There has been an increase in the potency of cannabis during the last two decades and adoption of a novel method of administration—vaping. YouTube, a social media platform, has become a popular source to access cannabis-related information. This study aimed to identify cannabis vaping YouTube videos from 2016 to 2020 and ex- amine the themes and metrics. Design Cross-sectional sample of 200 YouTube videos. Setting YouTube, an on-line video sharing platform. Measurements Videos related to cannabis vaping were identified using the search terms: ‘vaping cannabis’, ‘vaping weed’, ‘vaping marijuana’ and ‘vaping THC’ [tetrahydrocannabinol]. Videos were indepen- dently coded by two researchers. The number of views, likes, dislikes and comments were also collected. Robust regression was used to analyse the relationship between identified video themes and video metrics. Findings Six themes were iden- tified: ‘advertisement’, ‘product review’, ‘celebratory’, ‘reflective’, ‘how-to’ and ‘warning’. The ‘how-to’ and ‘celebratory’ videos received the highest number of views and likes. The most popular video was viewed more than 4 000 000 times. Many videos portrayed risky behaviour (e.g. vaping a whole THC cartridge in a single setting). Fifty-two percent of these videos had no age access restrictions. The robust regression model also found that engagement metric was positively as- sociated with ‘reflective’ videos and negatively associated with ‘advertisement’ videos. Conclusions A large number of videos on cannabis vaping are available on-line without age-restriction. Videos that portrayed risky behaviour appear to be prevalent.
Keywords Cannabis, cannabis vaping, marijuana, social media, vaping marijuana, YouTube.
Correspondence to: Carmen C. W. Lim, National Centre for Youth Substance Use Research, The University of Queensland, St Lucia 4072 QLD, Australia.
E-mail: [email protected] Submitted 11 June 2020; initial review completed 5 October 2020; final version accepted 13 January 2021
INTRODUCTION
Cannabis is the most commonly used controlled substance world-wide after alcohol and tobacco [1]. With more juris- dictions adopting cannabis legalization [2] awider range of cannabis products with higher tetrahydrocannabinol (THC) content, such as cannabis concentrates, have emerged in the market [3]. Although smoking remains the primary route of administering cannabis [4,5], other modes of delivery such as vaping have risen in popularity due to the availability of these substances. A vaporizer is used to heat herbal cannabis or cannabis oil to release the active psychoactive compounds, including cannabidiol (CBD) and THC, into a vapour for inhalation [5,6].
Conversely, cannabis concentrates are often consumed using amethod called ‘dabbing’. Similar to vaping, dabbing uses a device (e.g. dab rig) where a small amount of extract is placed on a surface that has been heated to a high tem- perature and vapour is inhaled through the mouthpiece of the dab rig [7]. People who dab experience an intense high all at once, rather than a gradual build-up over time, as is experienced when vaping or smoking cannabis [7]. Worry- ingly, e-cigarettes or similar devices are also being used with concentrated cannabis oil or waxwith high THC con- tent [8,9]. The increasing accessibility to products with high THC content and novel methods of administering cannabis increases concern about the potential for misuse [9].
© 2021 Society for the Study of Addiction Addiction, 116, 2443–2453
RESEARCH REPORT doi:10.1111/add.15424
The use of cannabis products with high THC content raises a number of potential health problems. Concentrated products can be up to 30 times more potent than herbal cannabis [3]. The potency of cannabis, defined by the amount of THC content, remains largely unregulated [2,9]. The regular use of highly potent products is associ- ated with poorer mental health [7] and increased risks of health outcomes, such as impaired motor coordination, chronic bronchitis and psychosis [10,11]. In adolescents, early chronic cannabis use is linked to neuropsychological decline [12]. A recent meta-analysis showed that cannabis use among adolescents is linked to an increased risk in de- pression and suicidality in young adulthood [13].
Recently, cannabis vaping has gained popularity around the world, particularly in the United States. The Monitoring the Future survey, a nation-wide US survey of 42 531 students from 396 public and private schools, showed an increasing trend in past month cannabis vaping, where the biggest increase was observed for 12th graders; from 5% in 2017 to 14% in 2019 [14]. A 2018 multi-national survey of 11 537 youths showed that the prevalence of cannabis vaping in the past month was highest in the United States (5.1%) and lowest in England (1.7%) [5]. A cross-sectional study of 32 678 German adults conducted between 2016 and 2019 found that one out of 14 e-cigarette users reported having vaporized cannabis [15]. Cannabis vaping has gained popularity be- cause it is perceived to bemore efficient in delivering higher doses of THC [16], and is associated with fewer respiratory symptoms [17] than smoking cannabis. Contemporary vaporizers and dabbing devices (e.g. vape and dab pens) allow individuals ready access to vaping [18] and the discreetness is particularly attractive for adolescents [19].
The internet has become a popular source of cannabis-related information, ranging from cannabis oil recipes to harm reduction advice on drug forums [9,20]. Although it is illegal for adolescents to possess cannabis, a 2018 survey showed that 80% of adolescents between ages 15 and 19 reported being exposed to cannabis marketing on social media platforms (e.g. YouTube) in US states where cannabis is legal [21]. Only two studies have exam- ined how cannabis vaping is portrayed on YouTube and were based on videos searched up to 2015. One study by, Krauss and colleagues, examined dabbing [22] and an- other by Yang et al. reviewed video content for cannabis vaping based on videos published between July 2014 and June 2015 [23]. A more recent study identified a number of videos that portrayedmodifications to electronic nicotine delivery systems that can be used to deliver substances other than nicotine. Although this study was not specific to cannabis vaping, the study found that videos on do-it- yourself (DIY) cannabis e-liquids received a high number of views [24]. Given recent findings [24] and rapid develop- ments in cannabis vaping [25,26], the results of the two
existing studies on cannabis vapingmay not reflect the cur- rent new vaping technology and associated viewers’ interests.
This study generates fresh insights into publicly avail- able videos related to cannabis vaping on YouTube. YouTube is an on-line video sharing platform that attracts more than 2 billion monthly users around the world [27], of whom 85% of 13–17-year-olds use YouTube in the United States alone [28]. Given the size and the reach of YouTube it is important to examine the content on this platform, as media exposure among adolescents is highly correlated to subsequent use [21,29]. We aim: (i) to collect a sample of cannabis vaping videos (2016–20) from YouTube, (ii) to identify the themes of the videos and (iii) to summarize the associated video metrics (e.g. number of views) and themes of the video.
METHODS
Sampling strategy
Although the risk profile for vaping cannabis and dabbing are different, both methods involve vaporization. The term ‘vaping’ is sometimes used interchangeably for dabbing. In this paperwe consider dabbing to be a subset of the broader vaping category. Four key phrases that included a combination of vaping and cannabis synonyms (‘vaping cannabis’, ‘vaping weed’, ‘vaping marijuana’, ‘vaping THC’) were selected based on their relative popularity on YouTube, according to Google Trends. Videos on dabbing were also captured using these key phrases. All video searches were performed in a single day on 1 April 2020. From the key phrase, 200 videos were retrieved between 2016 and 2020 using the default filter parameter (‘rele- vance’) on YouTube. Then the search was repeated using the ‘rating’ parameter to search for videos. Searches were conducted on the Chrome browser using the incognito mode with the watch and search history turned off to prevent YouTube’s auto-suggest video features. The initial sample contained 200 videos (four key phrases × 5 years × two sortingmethods in YouTube × five videos). The final sample contained 120 videos (see Supporting information, Table S1) after removing dupli- cates (n = 73), videos unrelated to cannabis vaping (n = 4) and videos not in the English language (n = 3) (see Fig. 1).
Development of the codebook for content themes
This codebook was created according to qualitative meth- odology for health research [30]. The development of the codebook was guided by (i) the themes in existing YouTube studies on cannabis vaping [22,23] and (ii) themes identi- fied in a small sample of videos (40 videos). First, two researchers independently watched and performed in-vivo
2444 Carmen C. W. Lim et al.
© 2021 Society for the Study of Addiction Addiction, 116, 2443–2453
coding on 20 videos [31]. This involves coding the audio and visuals in the videos to develop descriptive codes and definitions for the video themes. Then two researchers met to discuss and compare the coding. At this stage, both sets of coding were reconciled by combining or deleting where necessary. An additional 10 videos were coded using the refined codes and the researchers met again to recon- cile the codes. This coding process was repeated on another set of videos. After watching 40 videos, a final codebook was developed where the codes for each theme were deemed to be operational with a clear and concise definition.
Coding procedures
The general methodology of the coding was consistent with the methodology applied in previous research [22,32]. Two researchers watched 120 videos in full and independently coded the video themes, age group (< 25 years old, 25+ years old or could not tell) and gender of the primary presenter or person vaping in the video. Videos could be classified into one or more of these themes (not mutually exclusive). For analytical purposes all the themes were binary coded (e.g. yes or no). The kappa agreement, κ, was calculated on thewhole sample. The ini- tial intercoder reliability between the two researchers was acceptable (κ = 0.76). All the disagreements in coding were reconciled via discussion (i.e. κ = 1.00). Video title, description, web address, channel characteristics (number of channel subscribers, views, date established and geolocation) and video characteristics (number of likes, dis- likes, views and comments) were collected. The current study uses three different video metrics: (i) ‘number of
views’, (ii) ‘engagement’, which is the ratio of comments and views and (ii) ‘positivity’, which is the ratio of likes and views. The number of views was used as a proxy for popularity in keeping with prior studies [22,23]. Caution should be exercised when interpreting the results, as this metric is affected by the life-span of the video. The ‘engage- ment’ metric represents the users who engaged with the video by expressing a comment, while the ‘positivity’ met- ric represents the proportion of viewers who expressed a positive sentiment. These metrics were not affected by the life-span of the video.
Data analysis
The video themes and the characteristics of the videoswere summarized using descriptive statistics. The distribution for each video metric was examined using a ridgeline plot. Seven videos had their comment feature turned off. Analy- sis was only performed on videos that allowed user com- ments. Robust regression was used to examine the association between video themes and engagement met- rics. In robust regression, unlike linear regression, cases with large residuals (potential outliers such as highly viewed videos) are down-weighted. A series of robust re- gressionmodels were developed to examine the association between each video theme and video metrics. These models initially adjusted for the number of channel sub- scribers and date the channel was established. Due to multicollinearity, these variables were excluded [variance inflation factor (VIF) > 5]. We conducted a set of analysis where video metrics were regressed on video themes. We additionally undertook a post-hoc analysis adjusting for characteristics of the primary presenter. The ‘number of
Figure 1 Flow diagram of the sampling strategy
Cannabis vaping videos on YouTube 2445
© 2021 Society for the Study of Addiction Addiction, 116, 2443–2453
views’ metric was additionally adjusted for time since the video was uploaded.We also performed sensitivity analyses using a log-transformation via an ordinary least-square model to test the robustness of our results. The robust re- gression models were conducted using the ‘robustbase’ package and the linear regression model via the ‘lm’ func- tion in R (version 3.6.2). We also adjusted for multiple comparisons by lowering our statistical significance level to 0.0083 (alpha = 0.05/6) to minimize concerns about the misuse of P-values. We did not pre-register the study because no hypotheses was tested in this exploratory anal- yses [33].
Ethics approval
Ethical clearance was obtained from the Office of Research Ethics at The University of Queensland (exemption ref: 20200011080).
RESULTS
Video themes
Six prominent video themes were identified from the data: (i) advertisement; (ii) product review; (iii) celebratory; (iv) reflective; (v) how-to; and (vi) warning (see Table 1).
Videos with a warning theme were the most dominant theme in our sample (24.6%), followed by the ‘how-to’ (20.8%), ‘advertisement’ (16.9%), ‘reflective’ (16.1%), ‘cel- ebratory’ (11.5%) and ‘product review’ (10.0%). An exam- ple of the video for each theme is shown in Fig. 2.
Video characteristics
The characteristics of the video content are shown in Table 2. One hundred and twenty videos from 2016 to 2020 were identified (see Table 2). These videos were
predominantly from the United States and Canada. After excluding videos with awarning theme, many videos were not age-restricted (52.3%). Approximately 66% of the videos feature someone vaping cannabis. The perceived age group and gender of the primary presenter or person vaping in the videos were young adults (47.7%) and males (66.9%).
The video metrics are summarized in Table 3. Overall, the median (25th percentile, 75th percentile) number of viewswas 1592 (247.75, 35 630.25), themedian number of likes and dislikes (25th percentile, 75th percentile), respectively, were 26 (3.5, 357) and 1 (0, 39), while the median number of comments (25th percentile, 75th per- centile) was approximately 7 (1, 108). The most popular videowas viewedmore than 4 000 000 times.Within each video theme, ‘celebratory’ and ‘how to’ videos received the highest median number of views and median number of likes. The number of warning videos in our sample was large, but the median number of views and likes was lower than that of other themes.
Distribution of video metrics
Figure 3 shows the distribution of video metrics (frequency counts) in a ridgeline plot. Like a histogram, the height of each bar indicates the frequency of data points within the specific bin. For the number of views metric, the distribution is right-skewed because some videos in the ‘celebratory’, ‘reflective’ and ‘how-to’ categories were highly viewed. For the engagement metric the distribution was relatively skewed to the right, with a small proportion of videos in the ‘product review’ and ‘celebratory’ themes having a large number of comments. The positivity metric is right-skewed, with a large spread and many influential data points.
Table 1 Video themes and counts.
Themesa,b Definition Number of videos Percentage
I. Advertisement
A clear intent to promote a brand by a commercial enterprise 22 16.9%
II. Product review
Review of one or more products with no clear intention to promote or sell a particular product (e.g. 9 weed vapes under $150)
13 10.0%
III. Celebratory
Focus on pleasure-seeking or portrayed risky behaviour (e.g. getting high via vaping) 15 11.5%
IV. Reflective Thoughtful discussion about the drug style and experience without the hedonistic component found in celebratory videos
21 16.1%
V. How-to Step-by-step instructions on how to use, maintain a vaporizer, do-it-yourself cannabis e-liquid recipes
27 20.8%
VI. Warning Warning audience the dangers and negative health consequences associated with cannabis vaping
32 24.6%
a Videos could be classified into one or more of these themes (not mutually exclusive);
b all the themes were binary coded (e.g. yes or no).
2446 Carmen C. W. Lim et al.
© 2021 Society for the Study of Addiction Addiction, 116, 2443–2453
Video metrics
Table 4 shows the association between each video theme and video metric adjusting for characteristics of the pri- mary presenter in the video. After adjusting for multiple comparison, ‘advertisement’ videos were associated with lower engagement [beta = �0.0014, 95% confidence in- terval (CI) = �0.0027, �0.0046], while ‘reflective’ videos were associated with higher engagement (beta = 0.0019, 95% CI = 0.0004, 0.0015). The findings for the associa- tions between video themes with the number of views and the positivity metric were inconclusive. A set of sensi- tivity analyses using log-transformation were performed and the results were broadly consistent with the robust re- gression (see Supporting information, Table S2).
DISCUSSION
This study found a large variety of cannabis vaping videos on YouTube from 2016 to 2020. The six themes found were how-to, celebratory, advertisement, warning, reflec- tive and product review. Consistent with prior studies, can- nabis vaping videos (including dabbing) had a high number of views and likes [22,23], particularly videos within the ‘how-to’ and ‘celebratory’ theme. Themost pop- ular videowas viewed 4 148 392 times. A large proportion of the videos were US-based. Many were also not age-re- stricted for viewers under the age of 18. This is of concern,
due to the frequent internet use with which adolescents engage. Furthermore, adolescence is a period of develop- ment that is often associated with impulsivity and experi- mentation, some of which involve risks that can be harmful [34]. It is possible that videos which show users in distress after vaping may discourage adolescent use; however, normalization of cannabis vaping can still occur when the behaviour is modelled, particularly if it is enter- taining [22,35].
Videoswith a ‘how-to’ theme received a high number of views and likes. In general, these videos were a beginner’s guide to cannabis vaping, tutorials and DIYvideos onmak- ing THC oil. These videos do not appear to violate YouTube’s community guidelines, which do not allow post- ingof content that depicts abuse or provides instructions on how to create hard drugs such as cocaine or opioids [36]. YouTube defines hard drugs as drugs that mainly lead to physical addiction. It is unclear whether cannabis falls into this category. This is concerning, as the THC level in some DIY cannabis concentrates videos claimed to contain more than 70%THC. Some of the ‘how-to’ videos alsomade side- by-side comparisons between vaping and smoking canna- bis, where vaping was often being portrayed as a more effi- cient and healthier way to administer cannabis.
Another theme that received a high number of views and likes was the ‘celebratory’, where the primary focus is on pleasure-seeking. Many videos with a celebratory theme portrayed risky behaviour with people indulging in
Figure 2 Screenshots of each video theme in the current study. [Colour figure can be viewed at wileyonlinelibrary.com]
Cannabis vaping videos on YouTube 2447
© 2021 Society for the Study of Addiction Addiction, 116, 2443–2453
activities that could endanger health (e.g. attempts to vape a large amount of THC oil in 4 hours). Many of these videos also showed the use of highly potent THC oil-based car- tridges with 70–90% THC. The use of high-potency THC among first-time inexperienced users has been shown to cause anxiety, dysphoria and paranoia [37]. In some cases, this can also lead to unintentional injury such as a motor vehicle crash while intoxicated [37]. We found multiple short clips of first-time/inexperienced young adults vaping becoming distressed after vaping THC cartridges.
We also found that videos with a ‘warning’ theme re- ceived fewer views, likes and lower engagement compared to other themes, despite fewer age restrictions (< 10%). These videos were largely comprised of educational, cam- paign and harm reduction materials from various govern- ment agencies and news outlets. The warning videos (26.7%) were more prevalent in this sample than in Yang et al.’s sample in 2015 (1.4%). The difference in sampling methodology could have contributed to the disparity. Many
videos warned viewers about the danger of consuming THC oil from the black market after the outbreak of vaping-related lung injuries associated with cannabis vapes adulterated with vitamin E in the United States. Some ‘reflective’ videos also had a warning theme (e.g. videos that encourage appropriate dosing as vaping can amplify the effects of THC compared to smoking cannabis). The increase in the visibility of these warning videos in our study compared to Yang et al.’s study performed 5 years ago may indicate an increasing awareness of the risks of cannabis vaping. However, the low number of views and level of engagement suggests that different strategies may need to be implemented to increase views by young audiences.
Another noteworthy finding was that ‘reflective’ videos were associated with high levels of engagement, while ‘advertisement’ videos were associated with lower levels of engagement. ‘Reflective’ videos typically involved discussion of the content creator’s own drug experience.
Table 2 Video content characteristics.
Video themesa
Total (n = 120)
Advertisement Product review Celebratory Reflective How to Warning
(n = 22) (n = 13) (n = 15) (n = 21) (n = 27) (n = 32)
I. Year uploaded, n (%) 2016 4 (18.2) 5 (38.5) 3 (20.0) 4 (19.0) 10 (37.0) – 24 (20) 2017 9 (40.9) – 3 (20.0) 5 (23.8) 7 (25.9) 2 (6.3) 25 (20.8) 2018 6 (27.3) 3 (23.1) 4 (26.7) 2 (9.5) 3 (11.1) 7 (21.9) 23 (19.2) 2019 1 (4.5) – 1 (6.7) 2 (9.5) 6 (22.2) 14 (43.8) 23 (19.2) 2020 2 (9.1) 5 (38.5) 4 (26.7) 8 (38.1) 1 (3.7) 9 (28.1) 25 (20.8)
II. Geolocation of the video, n (%) United States 13 (59.1) 9 (69.2) 9 (60.0) 11 (52.4) 19 (70.4) 23 (71.9) 74 (61.7) Canada 1 (4.5) 3 (23.1) 2 (13.3) 2 (9.5) 2 (7.4) 3 (9.4) 11 (9.2) United Kingdom – 1 (7.7) 1 (6.7) 2 (9.5) 1 (3.7) – 5 (4.2) Australia – – – – 1 (3.7) 1 (3.1) 3 (2.5) China 1 (4.5) – – – – – 1 (0.8) Not specified 7 (31.8) – 3 (20.0) 6 (28.6) 4 (14.8) 5 (15.6) 26 (21.7)
III. Non-age-restricted videos, n (%) 19 (86.4) 8 (61.5) 10 (66.7) 15 (71.4) 16 (59.3) 29 (90.6) 97 (74.6) IV. Mean video length (minutes) 1.45 8.47 8.38 9.18 8.54 14.38 8.42 V. Features someone vaping in the video, n (%)
13 (59.1) 10 (76.9) 15 (100) 16 (76.2) 18 (66.7) 14 (43.8) 86 (66.1)
VI. Perceived age groupb, n (%) < 25 years 8 (36.4) 9 (69.2) 11 (73.3) 12 (57.1) 12 (44.4) 10 (31.2) 62 (47.7) ≥ 25 years 5 (22.7) 4 (30.8) 4 (26.7) 9 (42.9) 11 (40.7) 20 (62.5) 53 (40.8) Hands only 4 (18.2) – – – 4 (14.8) 2 (6.2) 10 (7.7) No information 5 (22.7) – – – – – 5 (3.8)
VI. Perceived genderc, n (%) Male 13 (59.1) 12 (92.3) 10 (66.7) 19 (90.4) 18 (66.7) 15 (46.9) 87 (66.9) Female 3 (13.6) 1 (7.7) 4 (26.7) 2 (9.5) 7 (25.9) 7 (21.9) 24 (18.4) Mix of bothd 4 (18.2) – 1 (6.7) – 2 (7.4) 10 (31.2) 17 (13.1) Not applicable 2 (0.1) – – – – – 2 (1.5)
a A video can have multiple themes;
b age of the person vaping, running demo tutorials, talking about their experience;
c gender of the person vaping, running
demo tutorials, talking about their experience; d group vaping.
2448 Carmen C. W. Lim et al.
© 2021 Society for the Study of Addiction Addiction, 116, 2443–2453
Ta bl e 3
V id eo
m et ri cs
fo r ea ch
vi de o th em
e. V id eo
th em
es a
To ta l( n = 12
0)
A dv er tis em
en t
Pr od uc t re vi ew
Ce le br at or y
R efl ec tiv e
H ow
to W ar ni ng
(n = 22
) (n
= 13
) (n
= 15
) (n
= 21
) (n
= 27
) (n
= 32
)
I. V ie w s
To ta l
63 7 63
8 84
0 29
0 7 10
9 77
8 1 52
4 39
9 5 04
0 73
9 34
9 82
8 13
60 3 79
1 M ed ia n
77 4
41 35
22 98
8 13
64 56
15 6
39 8
15 92
R an
ge (1 0–
44 0 47
5) (1 4–
28 8 94
0) (6 –4
14 8 39
2) (2 9–
1 30
3 84
8) (7 6–
1 65
1 53
2) (1 0–
11 7 64
1) (6 –4
14 8 39
2) (2 5t h pe rc en til e, 75
th pe rc en til e)
(2 98
,3 61
3. 5)
(4 70
,1 12
82 6)
(1 03
.5 ,2
03 44
0) (2 48
,1 6 88
1) (9 71
7. 5,
16 5 73
5. 5)
(2 33
.5 ,2
2 17
7. 5)
(2 47
.7 5,
35 63
0. 25
) II .L ik es
M ed ia n
4 64
34 0
48 34
9 20
26 R an
ge (0 –2
35 8)
(3 –4
76 4)
(1 –1
9 14
1) (1 –1
9 14
1) (1 –2
3 00
7) (0 –2
3 00
7) (0 –2
3 00
7) (2 5t h pe rc en til e, 75
th pe rc en til e)
(1 .2 5,
10 .5 )
(2 0,
68 3)
(3 ,2
34 8)
(6 ,1
32 )
(6 3,
13 67
) (3 ,2
89 )
(3 .5 ,3
57 )
II I. D is lik es
M ed ia n
0 1
32 2
29 1
1 R an
ge (0 –4
94 )
(0 –3
47 )
(0 –1
80 8)
(0 –1
80 8)
(0 –1
36 8)
(0 –1
80 8)
(0 –1
80 8)
(2 5t h pe rc en til e, 75
th pe rc en til e)
(0 ,0
) (0 ,8
0) (0 ,1
37 .5 )
(0 ,1
6) (4 ,1
16 .5 )
(0 ,3
2) (0 ,3
9)
IV .C
om m en ts
M ed ia n
7 58
7 7
55 6
7 R an
ge (0 –3
41 )
(2 –8
59 )
(0 –4
89 5)
(0 –4
89 5)
(0 –2
12 5)
(0 –4
89 5)
(0 –4
89 5)
(2 5t h pe rc en til e, 75
th pe rc en til e)
(1 ,1
17 )
(6 ,1
55 )
(1 ,1
17 )
(1 ,1
10 .2 5)
(1 5. 5,
23 9. 5)
1, 96
.5 )
(1 ,1
08 )
a A vi de o ca n ha
ve m ul tip
le th em
es .
Cannabis vaping videos on YouTube 2449
© 2021 Society for the Study of Addiction Addiction, 116, 2443–2453
Reflective videos may guide more informed purchasing de- cisions [38]. Users can potentially use them to enhance their own drug experiences and identify the ideal cannabis product to vape. Conversely, ‘advertisement’ videos were less engaging but still highly prevalent, and received dou- ble the number of views received by warning videos. Cur- rently, unregulated substances content, such as cannabis advertisement, is prohibited on YouTube, with the excep- tion of cannabis videos which are for educational pur- poses or in the form of a documentary. As a consequence, many advertisements in our study were ad- vertising the vaporizing device, not the substance being vaped.
In many product review videos, the video creators disclosed that they were reviewing a product sent to them by cannabis or vaping companies. The cannabis industry is leveraging the power of social media to reach mass audi- ences by sending free products to YouTube influencers in exchange for a review of their products. This indirect form of advertising is efficient, because messages conveyed by digital influencers with thousands of followers and mes- sages are considered to be more ‘credible’ and ‘attractive’ for their audiences [39]. Consequently, regulating market- ing on social media is challenging [40]. These challenges need to be considered for future policy development around cannabis vaping.
The use of statistical models that appropriately accounted for the distribution of the video metrics is a
strength of the current study. All three video metrics were right-skewed and did not follow a standard probability dis- tribution. Fitting ordinary least-squares regression models would be inappropriate in this situation due to the presence of many influential points that may incorrectly lead to er- roneous conclusions. We used robust regression in our study because this method is superior to transforming the data via a mathematical function to handle highly influen- tial data points and outliers [41].
Limitations of the current study should be acknowl- edged. Only including English language videos on one video platform limits generalizability. Our results, that ad- justed for characteristics of the primary presenter, were based on a set of post-hoc analyses. The direction of find- ings in the post-hoc analyses were similar to the results that we originally planned without adjusting for presenter characteristics (see Supporting information, Table S3). The number of views metric is affected by the life-span of the video; caution should be exercised when interpreting the results. Our study did not analyze user comments that would allow further understanding of how the content is being received by the audience, despite the provenance of the comments being unknown. Our study is cross-sectional in nature. Tracking the popularity of each video in terms of number of views over time might provide an indication of the trajectory of cannabis use and likeli- hood of engaging in risky behaviour, as exposure is corre- lated to consumption.
Figure 3 Distribution of video metrics. [Colour figure can be viewed at wileyonlinelibrary.com]
2450 Carmen C. W. Lim et al.
© 2021 Society for the Study of Addiction Addiction, 116, 2443–2453
CONCLUSIONS
This study demonstrates that a large number of cannabis vaping videos, many without age restriction, can be found on YouTube. Videos that portray risky behaviour are prev- alent in our study. Adolescent exposure to these videosmay normalize cannabis vaping. The use of digital influencers to review cannabis products on YouTube is also becoming in- creasingly common. Together with vigorous monitoring of social media, a policy reform on YouTube is urgently needed with tighter regulations around age restrictions to minimize the prevalence of cannabis vaping among adoles- cents, especially for videos portraying activities that en- courage harmful use and product reviews.
Declaration of interests.
None.
Acknowledgements
C.L. is supported by a National Health Medical Research Council (NHMRC) of Australia Postgraduate Scholarship (APP2005317), The University of Queensland Graduate School Scholarship (UQGSS) and a National Centre for Youth Substance Use Research (NCYSUR) top-up scholar- ship. J.L. is supported by The University of Queensland de- velopment fellowship. G.C. is supported by a National Health and Medical Research Council (NHMRC) of Australia Investigator Fellowship. NCYSUR and the Lives LivedWell group are supported by Commonwealth funding from the Australian Government provided under the Drug and Alcohol Program. The funding bodies had no role in the study design, collection, analysis or interpretation of the data, writing the manuscript or the decision to sub- mit the paper for publication.
Author contributions
Carmen C. W. Lim: Conceptualization; data curation; for- mal analysis; investigation; methodology; project adminis- tration. Janni Leung: Conceptualization; investigation; methodology; resources; supervision. Jack Yiu Chak Chung: Data curation. Tianze Sun: Investigation; re- sources. Coral Gartner: Investigation; supervision. Jason Connor: Investigation; supervision.Wayne Hall: Investiga- tion; supervision. Vivian Chiu: Data curation; investiga- tion; resources. Calvert Tisdale: Data curation; resources. Daniel Stjepanović: Investigation. Gary Chan: Conceptual- ization; formal analysis; investigation; methodology; pro- ject administration; supervision.
References
1. Peacock A., Leung J., Larney S., Colledge S., Hickman M., Rehm J., et al. Global statistics on alcohol, tobacco andTa
bl e 4
A ss oc ia tio
ns be tw
ee n ca nn
ab is va pi ng
vi de o th em
es an
d vi de o en ga ge m en t m et ri cs
us in g ro bu
st re gr es si on
.
En ga ge m en ta
N um
be r of vi ew
sb Po si tiv ity
a
N um
be r of co m m en ts /n um
be r of vi ew
s N um
be r of vi ew
s N um
be r of lik es /n um
be r of vi ew
s
B et a
Lo w er
U pp er
p- va lu e
B et a
Lo w er
U pp er
P- va lu e
B et a
Lo w er
U pp er
P- va lu e
V id eo
th em
es A dv er tis em
en t
�0 .0 01
4 �0
.0 02
7 �0
.0 04
6 0. 02
77 63
9. 40
�3 94
.8 1
�2 46
.1 7
0. 22
81 �0
.0 05
4 �0
.0 09
6 �0
.0 01
2 0. 01
32 Pr od uc t re vi ew
0. 00
19 �0
.0 00
1 0. 00
08 0. 06
22 53
9. 37
�5 80
.6 1
�5 66
.5 8
0. 34
72 0. 00
38 �0
.0 01
6 0. 00
93 0. 16
89 Ce le br at or y
0. 00
00 �0
.0 01
6 0. 00
23 0. 99
91 �5
67 .4 9
�1 27
4. 14
�2 13
6. 78
0. 11
83 0. 00
23 �0
.0 04
5 0. 00
91 0. 51
03 R efl ec tiv e
0. 00
19 0. 00
04 0. 00
15 0. 01
81 �2
34 .3 5
�1 09
3. 73
�1 70
5. 25
0. 59
41 0. 00
57 �0
.0 00
6 0. 01
19 0. 07
69 H ow
to �0
.0 00
5 �0
.0 01
6 �0
.0 02
6 0. 43
85 23
2. 07
�4 45
.7 5
�5 27
.8 5
0. 50
40 �0
.0 03
7 �0
.0 08
0 0. 00
06 0. 09
10 W ar ni ng
0. 00
27 �0
.0 01
5 �0
.0 00
8 0. 21
02 �2
31 .0 0
�1 08
1. 78
�1 68
6. 21
0. 59
57 0. 00
19 �0
.0 03
1 0. 00
68 0. 45
50
a A dj us te d fo r ag e gr ou
p, ge nd
er of th e pe op le th at
pr es en t or
va pe
in th e vi de o;
b a dj us te d fo r ag e gr ou
p, ge nd
er of th e pe op le th at
pr es en t or
va pe
in th e vi de o, tim
e si nc e vi de o w as
up lo ad ed .
Cannabis vaping videos on YouTube 2451
© 2021 Society for the Study of Addiction Addiction, 116, 2443–2453
illicit drug use: 2017 status report. Addiction 2018; 113: 1905–26.
2. Hall W., Stjepanović D., Caulkins J., Lynskey M., Leung J., Campbell G., et al. Public health implications of legalising the production and sale of cannabis for medicinal and recrea- tional use. Lancet 2019; 394: 1580–90.
3. ElSohly M. A., Mehmedic Z., Foster S., Gon C., Chandra S., Church J. C. Changes in cannabis potency over the last 2 de- cades (1995–2014): analysis of current data in the United States. Biol Psychiatry 2016; 79: 613–9.
4. Smith D. M., Miller C., O’Connor R. J., Kozlowski L. T., Wadsworth E., Fix B. V. et al. Modes of delivery in concurrent nicotine and cannabis use (‘co-use’) among youth: findings from the international tobacco control (ITC) survey. Subst Abuse 2020. https://doi.org/10.1080/ 08897077.2019.1709603
5. Fataar F., Hammond D. The prevalence of vapingand smoking as modes of delivery for nicotine and cannabis among youth in Canada, England and the United States. Int J Environ Res Public Health 2019; 16: 4111.
6. Lee D. C., Crosier B. S., Borodovsky J. T., Sargent J. D., Budney A. J. Online survey characterizing vaporizer use among cannabis users. Drug Alcohol Depend 2016; 159: 227–33.
7. Chan G. C. K., Hall W., Freeman T. P., Ferris J., Kelly A. B., Winstock A. User characteristics and effect profile of butane hash oil: an extremely high-potency cannabis concentrate. Drug Alcohol Depend 2017; 178: 32–8.
8. Kowitt S. D., Osman A., Meernik C., Zarkin G. A., Ranney L. M., Martin J., et al. Vaping cannabis among adolescents: prev- alence and associations with tobacco use from a cross-sectional study in the USA. BMJ Open 2019; 9: e028535–e028535.
9. Giroud C., de Cesare M., Berthet A., Varlet V., Concha-Lozano N., Favrat B. E-cigarettes: a review of new trends in cannabis use. Int J Environ Res Public Health 2015; 12: 9988–10008.
10. Hall W., Solowij N. Adverse effects of cannabis. Lancet 1998; 352: 1611–6.
11. Volkow N. D., Swanson J. M., Evins A. E., DeLisi L. E., Meier M. H., Gonzalez R., et al. Effects of cannabis use on human behav- ior, including cognition, motivation, and psychosis: a review. JAMA Psychiatry 2016; 73: 292–7.
12. Meier M. H., Caspi A., Ambler A., Harrington H., Houts R., Keefe R. S. E., et al. Persistent cannabis users show neuropsy- chological decline from childhood tomidlife. Proc Natl Acad Sci USA 2012; 109: e2657–e2664.
13. Gobbi G., Atkin T., Zytynski T., Wang S., Askari S., Boruff J., et al. Association of Cannabis use in adolescence and risk of depression, anxiety, and suicidality in young adulthood: a sys- tematic review and meta-analysis. JAMA Psychiatry 2019; 76: 426–34.
14. Miech R. A., Patrick M. E., O’Malley P. M., Johnston L. D., Bachman J. G. Trends in reported marijuana vaping among US adolescents, 2017–2019. JAMA 2020; 323: 475–6.
15. Kastaun S., Hildebrandt J., Kotz D. Electronic cigarettes to va- porize cannabis: prevalence of use and associated factors among current electronic cigarette users in Germany (DEBRA study). Subst Use Misuse 2020; 55: 1106–12.
16. Aston E. R., Scott B., Farris S. G. A qualitative analysis of can- nabis vaporization among medical users. Exp Clin Psychopharmacol 2019; 27: 301–8.
17. Earleywine M., Barnwell S. S. Decreased respiratory symp- toms in cannabis users who vaporize. Harm Reduct J 2007; 4: 11.
18. Centers for Disease Control and Prevention (CDC). E- Cigarette, or Vaping, Products Visual Dictionary. Atlanta, GA: CDC; 2020.
19. Jones C. B., Hill M. L., Pardini D. A., Meier M. H. Prevalence and correlates of vaping cannabis in a sample of young adults. J Soc Psychol Add Behav 2016; 30: 915–21.
20. Chiauzzi E., Dasmahapatra P., Lobo K., Barratt M. J. Participa- tory research with an online drug forum: a survey of user characteristics, information sharing, and harm reduction views. Subst Use Misuse 2013; 48: 661–70.
21. Whitehill J. M., Trangenstein P. J., Jenkins M. C., Jernigan D. H., Moreno M. A. Exposure to cannabis marketing in social and traditional media and past-year use among adolescents in states with legal retail cannabis. J Adolesc Health 2020; 66: 247–54.
22. Krauss M. J., Sowles S. J., Mylvaganam S., Zewdie K., Bierut L. J., Cavazos-Rehg P. A. Displays of dabbing marijuana extracts on YouTube. Drug Alcohol Depend 2015; 155: 45–51.
23. Yang Q., Sangalang A., Rooney M., Maloney E., Emery S., Cappella J. N. How is marijuana vaping portrayed on YouTube? Content, features, popularity and retransmission of vaping marijuana YouTube videos. J Health Commun 2018; 23: 360–9.
24. Massey Z. B., Li Y., Holli J., Churchill V., Yang B., Henderson K., et al. Modifications to electronic nicotine delivery systems: content analysis of YouTube videos. J Med Internet Res 2020; 22: e17104–e17104.
25. Dai H. Self-reported marijuana use in electronic cigarettes among US youth, 2017 to 2018. JAMA 2020; 323: 473–4.
26. Johnston L. D., Miech R. A., O’Malley P. M., Bachman J. G., Schulenberg J. E., Patrick M. E. Monitoring the Future Na- tional Survey Results on Drug Use, 1975–2018: Overview, Key Findings on Adolescent Drug Use. Ann Arbor, MI: Insti- tute for Social Research; 2019.
27. YouTube. YouTube for Press; 2020. Available at: https:// www.youtube.com/about/press/ (accessed 12 October 2020).
28. SproutSocial. YouTube age and gender demographics; 2020. Available at: https://sproutsocial.com/insights/new-social- media-demographics/ (accessed 25 December 2020).
29. Trangenstein P. J., Whitehill J. M., Jenkins M. C., Jernigan D. H., Moreno M. A. Active cannabis marketing and adolescent past-year cannabis use. Drug Alcohol Depend 2019; 204: 107548.
30. Neuendorf K. A. Content analysis—a methodological primer for gender research. Sex Roles 2011; 64: 276–89.
31. Manning J. In Vivo Coding. In: Matthes, J., editors. The International Encyclopedia of Communication Research Methods. Hoboken, NJ: John Wiley & Sons; 2017. https:// doi.org/10.1002/9781118901731.iecrm0270
32. Primack B. A., Colditz J. B., Pang K. C., JacksonK.M. Portrayal of alcohol intoxication on YouTube. Alcohol Clin Exp Res 2015; 39: 496–503.
33. Moore D. A. Preregister if you want to. Am Psychol 2016; 71: 238–9.
34. Glicksohn J., Naor-Ziv R., Leshem R. Chapter 8—Sensation- seeking and risk-taking. In: Martel M. M., editor. Developmen- tal Pathways to Disruptive. Impulse-Control and Conduct Disorders: Academic Press; 2018, pp. 183–208.
35. Cavazos-Rehg P. A., Zewdie K., Krauss M. J., Sowles S. J. ‘No high like a brownie high’: a content analysis of edible mari- juana tweets. Am J Health Promot 2018; 32: 880–6.
2452 Carmen C. W. Lim et al.
© 2021 Society for the Study of Addiction Addiction, 116, 2443–2453
36. YouTube. Harmful or dangerous content policy, YouTube: YouTube; 2020. Available at: https://support.google.com/ youtube/answer/2801964?hl=en (accessed 16 May 2020).
37. Hall W.What has research over the past two decades revealed about the adverse health effects of recreational cannabis use? Addiction 2015; 110: 19–35.
38. Burton J., Khammash M. Why do people read reviews posted on consumer-opinion portals? J DestinMark Manag 2010; 26: 230–55.
39. Lim X. J., Radzol A. M., Cheah J., Wong M. The impact of so- cial media influencers on purchase intention and the mediation effect of customer attitude.Asian J Bus Res2017; 7: 19–36.
40. Brodmerkel S., Carah N. Alcohol brands on Facebook: the challenges of regulating brands on social media. J Public Aff 2013; 13: 272–81.
41. Field A. P., Wilcox R. R. Robust statistical methods: a primer for clinical psychology and experimental psychopathology re- searchers. Behav Res Ther 2017; 98: 19–38.
Supporting Information
Additional supporting information may be found online in the Supporting Information section at the end of the article.
Table S1 List of cannabis vaping video in this study Table S2 Associations between cannabis vaping video themes and video engagement metrics using log-transfor- mation Table S3 Associations between cannabis vaping video themes and video engagementmetrics using robust regres- sion.
Cannabis vaping videos on YouTube 2453
© 2021 Society for the Study of Addiction Addiction, 116, 2443–2453
This document is a scanned copy of a printed document. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material.
