Psychometric properties of the French and English short form of the Protective Behavioural Strategies for Marijuana Scale in Canadian university students
Centre de Recherche du CHUM, Montreal, Québec, Canada
Université de Montréal, Montreal, Québec, Canada
Research Center, Ottawa Hospital Research Institute Clinical Epidemiology Program, Ottawa, Ontario, Canada
Faculty of Medicine, University of Ottawa, Ottawa, Ontario, Canada
Université Laval, Quebec City, Québec, Canada
University of Southern California, Los Angeles, California, USA
Correspondence to Professor José Côté; jose.cote@umontreal.caAbstract
Background
The Protective Behavioural Strategies for Marijuana (PBSM-17) scale serves to identify and measure strategies employed by young adults before, during or after cannabis use. After the adaptation and translation of the PBSM-17 into French, a methodological study was conducted to evaluate the psychometric properties of this French version (FV) and of the original English version (EV) in a sample of bilingual Canadian university students.
Methods
A total of 211 cannabis users (mean age=22.1 years) completed a sociodemographic questionnaire, a question on frequency of cannabis use (four categories: 1–3 times a month, once a week, more than once a week, everyday) and both versions (FV and EV) of the PBSM-17.
Results
Both versions had similar internal reliability (α=0.91; α=0.88). The one-factor solution explained 36.46% of the variance for the FV and 42.26% for the EV. As hypothesised, greater use of protective behavioural strategies was related to lower frequency of cannabis use. One-way ANOVA test results revealed a statistically significant difference in use of strategies by frequency of cannabis use for both the FV (F(3, 207)=27.38, p<0.001) and EV (F(3, 207)=29.32, p<0.001). Post hoc comparisons showed that everyday users employed fewer strategies on average than lower-frequency users.
Conclusion
The FV and EV of the PBSM-17 demonstrated satisfactory psychometric properties. The proposed FV of the PBSM-17 is a reliable instrument that could be used for research and clinical purposes. Protective behavioural strategies can serve as indicator of lower-risk cannabis use and could be targeted in prevention interventions.
Boxed Text
- The legalisation of cannabis use affords the opportunity to set up prevention initiatives to promote safer consumption behaviours.
- The use of protective behavioural strategies is a robust predictor of outcomes such as frequency of cannabis use and cannabis-related harms.
- This article reports on full psychometric properties of the English version of the Protective Behavioural Strategies for Marijuana short form (PBSM-17) and its French translation in a Canadian context.
- This validation of psychometric properties was conducted among a sample of university students, thus they might not necessarily be representative of the entire adult population.
- The PBSM-17 is a reliable instrument that could be used for research and clinical purposes in the context of the prevention and promotion of safe cannabis use.
Introduction
Prospective cohort studies suggest that substance use generally begins in adolescence and peaks in young adulthood.1 In Canada, 18–years old constitute the age group with the highest proportion of cannabis users.2 Canada was the first G7 country to legalise cannabis consumption for recreational purposes nationwide. According to Degenhardt et al,1 substance availability and access as well as regulatory environment are factors that affect substance use in young people. Legalisation, then, affords the opportunity to put forth initiatives aimed at reducing cannabis-related harms such as risk behaviours, social-interpersonal consequences and impaired control.3
Bravo et al stressed that using protective behavioural strategies (PBS) could help regulate cannabis use and even reduce the associated negative consequences.4 PBS were initially tied to the context of young adult alcohol use and conceptualised as behaviours used immediately prior to, during, and/or after a drinking episode to reduce alcohol misuse and/or alcohol-related harm.5 In a systematic review, Pearson6 concluded that PBS were promising ways to reduce alcohol misuse and the negative consequences of drinking. Transferring PBS to the context of cannabis use, Bravo et al carried out a cohort study of 2000 students from 11 universities in the USA. Findings showed PBS use to be a robust predictor of outcomes such as frequency of cannabis use and cannabis-related harms (eg, social-interpersonal consequences, impaired control and risk behaviours).7 Examples of PBS include avoiding using marijuana while spending time with family, avoiding mixing marijuana with other drugs, and purchasing marijuana from a trusted source.8 Considering that PBS use could mitigate the harms related to cannabis use, it is of great importance to develop a reliable tool for the purpose of measuring the extent to which these strategies are used by young adults.
Pedersen et al developed the Protective Behavioural Strategies for Marijuana scale (PBSM-17) to identify and measure the strategies employed by young adult cannabis users.8 They developed the PBSM scale in an initial exploratory study with 210 college student marijuana users by reducing a pool of 50 items down to a single-factor 39-item measure through iterative principal component analysis. According to psychometric evaluations, the unique structure of this scale explained 34% of the variance in protective behaviours related to cannabis use. Initial psychometric properties revealed excellent internal consistency (α=0.95) and convergent validity with a measure of alcohol PBS.8 In their subsequent work, Pedersen et al sought to validate the PBSM further with a larger and more diverse sample (n=2117, 40% men, 30% non-white) of college students from 11 different universities across the USA.9 Two versions of the scale, the 36-item version (PBSM-36) and the 17-item short form (PBSM-17), were developed and tested. Advanced factor analytical techniques (ie, confirmatory factor analysis) and advanced item response theory techniques confirmed the unidimensionality and good content coverage of both the PBSM-36 and the PBSM-17.9 The correlation between the two versions of the scale was high (r=0.98, p<0.001). Both versions demonstrated appreciable criterion-related validity and were strongly and negatively associated with past-month marijuana use and consequences. They were found to be free of bias in terms of gender, race and ethnicity.9
The short form of the scale (PBSM-17) can be of great value as it requires less time to complete. This renders it highly appealing for use in both research and clinical settings where professionals have limited time and interactions with research participants and patients.
However, the PBSM-17 was not available in French and had never been adapted to the Canadian context until recently. The aim of this methodological study was to evaluate the psychometric properties of the English and French short forms of the PBSM scale in a sample of bilingual university students in the province of Quebec, Canada.
Methods
The study was conducted according to the methodology proposed by Sousa and Rojjanasrirat for translating, adapting and validating the psychometric properties of scales.10 The procedure of French translation and adaptation of the PBSM-17 to the Canadian context is covered elsewhere.11 Briefly, the process involved members of the research team who were bilingual native French speakers, four translators, researchers with expertise in the development and validation of scales, and researchers with expertise in the area of cannabis use among young adults. It comprised five steps: (1) the original English instrument was translated to French by two translators separately (forward translation); (2) the two translations were compared and differences were resolved by consensus (synthesis 1); (3) this version was translated back to English by two translators blindly and separately (backward translation); (4) the two back translations were compared against one another and against the original English version (synthesis 2) and (5) the prefinal French version was pilot-tested.11
This article reports specifically on the fidelity (internal consistency and interitem correlation), construct validity (factor analysis) and criterion-related validity of the English version of the PBSM-17 and its French translation in a sample of bilingual university students.
Following a rigorous adaptation and validation process, we expected to obtain alpha coefficients for the French version of the PBSM-17 similar to those obtained by the English versions. Regarding the criterion-related validity, a negative relationship was hypothesised: We expected a higher PBSM-17 mean score (indicating a greater use of PBS) to be associated with a lower frequency of cannabis use.
Selection and description of participants
The study was conducted with a sample of university students from the province of Quebec, Canada. A convenience sampling strategy was used to recruit participants through social media, specifically the Université de Montréal Students Facebook Groups. Interested students were invited to visit the study webpage using a mobile device connected to the Internet (smartphone, tablet or laptop) in order to complete the online survey. Students 18 years old or over, who used cannabis at least once in the past month and were bilingual (ie, proficiency in written French and English) were included.
Online data collection
The study was conducted entirely online. LimeSurvey, an online survey system, was used to complete consent forms and questionnaires. After accepting the conditions and consenting, students were invited to complete a short sociodemographic questionnaire, one question on frequency of cannabis use, the French version of the PBSM-17 and the original English version of the PBSM-17. Eligibility criteria evaluation was integrated in the online questionnaire: participants who reported no cannabis use in the past year or less than once a month were not invited to complete the rest of the survey (ie, completion of the PBSM-17 was conditional on current cannabis use). Participants who completed the survey were compensated for their time with a US$10 Amazon.ca gift certificate.
Outcome measures
Participants completed a brief sociodemographic questionnaire covering age, gender self-identification, ethnicity and student status (full-time/part-time, graduate/undergraduate).
They were asked to rate their frequency of cannabis use by answering the following question from the Canadian Community Health Survey12: ‘How often did you use cannabis in the past 12 months?’. Frequency was measured on a six-point scale: (1) never, (2) less than once a month, (3) 1–3 times a month, (4) once a week, (5) more than once a week and (6) every day.
Participants were asked how often they resorted to employing 17 PBS when they used cannabis. Frequency of PBS use was rated on a six-point Likert scale: (1) never, (2) rarely, (3) occasionally, (4) sometimes, (5) usually and (6) always. As suggested by Pedersen et al9 the overall score was calculated by tallying the numbers corresponding to the answers given and converting this raw total to a t-score. There were no subscales. The higher the score, the greater the use of PBS.
For this study, participants were asked to complete the original English and the translated French versions of the PBSM-17. For the former, the order of the items was shuffled, as recommended by Sousa and Rojjanasrirat.10
Sample size
Following Sousa and Rojjanasrirat’s guideline10 of about 10 participants per item for general psychometric approaches (17 items × 10 participants=170) and projecting that 15% of participants might have missing data, we targeted a rounded-up total sample size of 200 participants.
Statistics
Descriptive statistics were computed for each item (frequency distribution for categorical data and means with SD for continuous variables). Fidelity was assessed by examining internal consistency (Cronbach’s alpha) and interitem correlation. As a general rule of thumb, a Cronbach’s alpha higher than α=0.70 is considered adequate. Gliem and Gliem describe α=0.80 as a reasonable goal for internal consistency, and that higher alpha values can suggest higher internal consistency. These authors also describe an adequate corrected item-total correlation having a value of at least r=0.40.13 For descriptive purposes, the preliminary psychometric properties of the English and French versions of the scale were compared.
As in the original study,8 a factor analysis with a one-factor solution structure was performed to evaluate the construct validity. Criterion-related validity was evaluated by conducting a one-way analysis of variance (ANOVA) to determine whether PBSM scale mean scores differed by frequency of cannabis use.
All statistical analyses were run on SPSS V.26 (IBM, Released 2019. IBM SPSS Statistics for Windows, V.26.0, IBM).
The creators of the PBSM scale were contacted for permission to translate and adapt the instrument. They were consulted during the translation process and reviewed the final manuscript. The tool is in the public domain and available free of charge (https://www.rand.org/health-care/surveys_tools.html).
Patient and public involvement statement
The participants in this study were not patients, but rather members of the general population (university students). Some members of the research team, involved in the development of the protocol for this study, are undergraduate and graduate students. They contributed to inform the research question, outcome measures and analyses. The research assistant who contributed to participant recruitment was an undergraduate student. Results will be disseminated to study participants using a plain language summary available on email request.
Results
Sample description
In September/October of 2020, 375 students provided consent, completed the sociodemographic questionnaire and answered the question on cannabis use frequency. Of these, 164 were excluded: 101 participants did not complete the two versions of the PBSM because they reported no cannabis use in the past year or using cannabis less than once a month (and were therefore not considered current cannabis users), 2 did not adhere with study procedures, and 61 had missing data. As a result, a total of 211 questionnaires were considered for analysis.
Mean age of participants was 22 years (range of 18–44). Most of them self-identified as women (61.1%) and as Caucasian (83.9%). Most were full-time students (88.2%) and at the undergraduate level (84.4%). Regarding frequency of cannabis use in the past 12 months, 46.9% indicated 1 to 3 times a month, 15.2% once a week, 22.3% more than once a week, and 15.6% every day.
As proposed by Pedersen et al,8 a total raw score was calculated for each participant by summing the scores obtained on the response scales of all 17 items. This total raw score was then converted to a T-score with a possible range of 15 to 73. In our sample, participants presented a mean T-score of 47.95 (SD: 8.53) on the French version and a mean T-score of 47.73 (SD: 9.45) on the original English version of the PBSM-17 scale.
Fidelity: internal consistency and interitem correlation
Cronbach’s alpha showed that the French version of the scale reached acceptable reliability, α=0.88. All of the items deserved retention; withdrawing any of them resulted in a lower alpha. Item-to-total corrected correlations ranged from r=0.38 to r=0.68 (table 1). Cronbach’s alpha showed that the original English version of the short form reached acceptable internal reliability (α=0.91). Item-to-total corrected correlations ranged from r=0.45 to r=0.72 (table 1).
| Item 1 | Item 2 | Item 3 | Item 4 | Item 5 | Item 6 | Item 7 | Item 8 | Item 9 | Item 10 | Item 11 | Item 12 | Item 13 | Item 14 | Item 15 | Item 16 | Item 17 | |
| Item 1 | 0.409 | 0.470 | 0.380 | 0.463 | 0.495 | 0.490 | 0.431 | 0.513 | 0.459 | 0.311 | 0.248 | 0.308 | 0.390 | 0.487 | 0.501 | 0.526 | |
| Item 2 | 0.442 | 0.663 | 0.291 | 0.416 | 0.384 | 0.460 | 0.311 | 0.419 | 0.377 | 0.325 | 0.250 | 0.202 | 0.323 | 0.422 | 0.342 | 0.483 | |
| Item 3 | 0.456 | 0.619 | 0.376 | 0.384 | 0.436 | 0.411 | 0.273 | 0.408 | 0.398 | 0.271 | 0.200 | 0.176 | 0.341 | 0.358 | 0.371 | 0.626 | |
| Item 4 | 0.303 | 0.272 | 0.351 | 0.483 | 0.284 | 0.497 | 0.270 | 0.285 | 0.170 | 0.313 | 0.351 | 0.203 | 0.225 | 0.264 | 0.374 | 0.312 | |
| Item 5 | 0.450 | 0.392 | 0.354 | 0.389 | 0.316 | 0.659 | 0.514 | 0.452 | 0.439 | 0.480 | 0.331 | 0.355 | 0.329 | 0.477 | 0.544 | 0.399 | |
| Item 6 | 0.347 | 0.326 | 0.370 | 0.180 | 0.236 | 0.371 | 0.331 | 0.455 | 0.328 | 0.233 | 0.124 | 0.224 | 0.338 | 0.340 | 0.451 | 0.452 | |
| Item 7 | 0.442 | 0.363 | 0.425 | 0.363 | 0.659 | 0.285 | 0.474 | 0.399 | 0.361 | 0.622 | 0.402 | 0.298 | 0.300 | 0.502 | 0.490 | 0.490 | |
| Item 8 | 0.294 | 0.255 | 0.301 | 0.234 | 0.425 | 0.292 | 0.517 | 0.413 | 0.347 | 0.487 | 0.337 | 0.306 | 0.350 | 0.417 | 0.615 | 0.343 | |
| Item 9 | 0.387 | 0.325 | 0.343 | 0.250 | 0.410 | 0.379 | 0.398 | 0.399 | 0.446 | 0.312 | 0.168 | 0.278 | 0.394 | 0.422 | 0.473 | 0.403 | |
| Item 10 | 0.308 | 0.293 | 0.295 | 0.033 | 0.386 | 0.274 | 0.334 | 0.415 | 0.415 | 0.268 | 0.231 | 0.392 | 0.376 | 0.443 | 0.474 | 0.441 | |
| Item 11 | 0.290 | 0.375 | 0.361 | 0.284 | 0.397 | 0.141 | 0.544 | 0.408 | 0.315 | 0.239 | 0.473 | 0.276 | 0.272 | 0.440 | 0.431 | 0.395 | |
| Item 12 | 0.205 | 0.182 | 0.212 | 0.285 | 0.219 | 0.093 | 0.301 | 0.195 | 0.157 | 0.189 | 0.471 | 0.294 | 0.365 | 0.255 | 0.374 | 0.278 | |
| Item 13 | 0.241 | 0.198 | 0.178 | 0.091 | 0.277 | 0.171 | 0.291 | 0.251 | 0.299 | 0.343 | 0.303 | 0.227 | 0.309 | 0.308 | 0.484 | 0.204 | |
| Item 14 | 0.317 | 0.308 | 0.304 | 0.133 | 0.180 | 0.298 | 0.235 | 0.275 | 0.324 | 0.330 | 0.238 | 0.360 | 0.334 | 0.343 | 0.421 | 0.372 | |
| Item 15 | 0.308 | 0.309 | 0.286 | 0.111 | 0.253 | 0.238 | 0.287 | 0.272 | 0.384 | 0.271 | 0.320 | 0.151 | 0.231 | 0.307 | 0.566 | 0.416 | |
| Item 16 | 0.356 | 0.289 | 0.308 | 0.222 | 0.464 | 0.359 | 0.494 | 0.460 | 0.396 | 0.380 | 0.460 | 0.187 | 0.331 | 0.345 | 0.508 | 0.428 | |
| Item 17 | 0.424 | 0.426 | 0.566 | 0.242 | 0.316 | 0.312 | 0.424 | 0.267 | 0.332 | 0.282 | 0.445 | 0.241 | 0.214 | 0.311 | 0.374 | 0.379 |
Construct validity: factor analysis
A principal component analysis using a one-factor solution was performed to assess construct validity. In this sample, for the French version, it explained 36.46% of the variance, with factor loadings of 0.432–0.741. For the English version of the scale, the 17 items explained 42.26% of the variance, with factor loadings of 0.495–0.763 (table 2). The Kayser-Mayer-Olkin (KMO) test and Bartlett’s test of sphericity showed that the data were suited for factor analysis (KMO=0.886, Bartlett 1270.65, df=130, p<0.001 for the French version and KMO=0.915, Bartlett 1590.15, df=136, p<0.001 for the English version).
| PBSM items | French version factor loadings | English version factor loadings |
| Item 1 | 0.652 | 0.725 |
| Item 2 | 0.633 | 0.647 |
| Item 3 | 0.670 | 0.655 |
| Item 4 | 0.452 | 0.546 |
| Item 5 | 0.684 | 0.740 |
| Item 6 | 0.513 | 0.598 |
| Item 7 | 0.741 | 0.757 |
| Item 8 | 0.621 | 0.662 |
| Item 9 | 0.643 | 0.664 |
| Item 10 | 0.566 | 0.633 |
| Item 11 | 0.652 | 0.628 |
| Item 12 | 0.432 | 0.497 |
| Item 13 | 0.471 | 0.495 |
| Item 14 | 0.531 | 0.576 |
| Item 15 | 0.549 | 0.687 |
| Item 16 | 0.691 | 0.763 |
| Item 17 | 0.652 | 0.695 |
Criterion-related validity
To evaluate the criterion-related validity of the instruments, we examined the association between PBSM scale mean score and frequency of cannabis use. A negative relationship was hypothesised with higher mean scores on the PBSM (ie, greater use of PBS) related to lower frequency of cannabis use. The one-way ANOVA test results revealed a statistically significant difference in use of marijuana PBS by frequency of cannabis use for the French version (F (3, 207)=27.38, p<0.001) and the English version (F (3, 207)=29.32, p<0.001). Post hoc comparisons (using Bonferroni correction) suggested that everyday cannabis users employed fewer marijuana PBS, on average, than did lower-frequency cannabis users. Additionally, people who used cannabis 1–3 times per month made significantly greater use of marijuana PBS than did those who used cannabis more than once a week (figure 1).
Discussion
The objective of our methodological study was to evaluate the full psychometric properties of the English and French versions of the PBSM-17 in a sample of bilingual university students. The two versions had similar internal reliability (α=0.91 and α=0.88, respectively). To our knowledge, few studies have reported on the reliability of the short form of the scale. In Pedersen et al too, the reliability of the PBSM-17 was high (α=0.93).9 Similarly to what was observed with the original structure of the scale8 9 the single-factor solution explained 42.26% of the variance for the English version and 36.46% for the French. Criterion-related validity, too, was established for both versions. PBS use differed by frequency of cannabis use, with overall greater use of PBS being associated with lower frequency of cannabis use.
Conducting rigorous psychometric assessments of instruments prior to their use in large-scale experimental studies is crucial. Multiple factors may impact the reliability obtained by an instrument such as the PBSM-17. These include scoring system, instrument’s characteristics (eg, item formulation, language), physical and emotional state of respondents at time of measure, and characteristics of context of administration.14 This study addressed factors related to the instrument’s characteristics, including item formulation and language, relevant to using the French version of the PBSM-17 with a sample of bilingual university students.
Both the English and the French versions of the PBSM-17 demonstrated satisfactory psychometric properties, in particular fidelity, construct validity and criterion-related validity. The PBSM-17 serves to assess strategies that can be applied prior to, during and after using cannabis. We validated the instruments in a population of university students, as did the creators of the PBSM. It is important to bear in mind that this student population is not necessarily representative of the adult population, which includes adults in the labour market. Also, the French version of the scale, like the English version, was developed and validated with current cannabis users as they were the ones who could resort to using the strategies under study.
It should be noted that the study took place in the context of the COVID-19 pandemic. Data collection was carried out at the start of the academic year in the fall of 2020, between the first and second waves of the pandemic in Canada.
When applied, PBS may limit frequency of marijuana use and the negative consequences associated with cannabis use. To our knowledge, PBS are not often used in research interventions. We found only two digital interventions that had PBS as an intervention target. The team led by Prince et al15 created a brief in-person intervention, including a smartphone application (MApp) to distribute tips related to PBS in order to reduce cannabis use. Another team, based in the USA, developed the Marijuana eCHECKUPTO GO, a web-based personalised feedback tool with normative information and PBS.16 Their study results revealed a decrease in frequency of cannabis use among heavy-user college students who participated in the intervention. In the experimental group, females increased their use of PBS more than males did.16 This suggests that PBS use should be considered as a target when developing interventions to encourage lower-risk cannabis use.
In the context of the prevention and promotion of safe cannabis use, the PBSM scale could be useful to both health professionals and community/social workers. It can be easily used in various specialised but also non-specialised settings and contexts, such as primary care services, family practices, schools and community settings and for various purposes, such as evaluation, screening, personalised feedback and intervention. It is important to offer tools to better help professionals who might feel ill equipped in the face of cannabis legalisation and of the adoption of the harm reduction paradigm in response to lifting of the prohibition on the use of this substance.
Limitations
This article reports on the full psychometric properties of the English version of the PBSM-17 and its French translation in a Canadian context. Though we applied a systematic procedure and high standards to adapt, translate and evaluate the PBSM-17, our study is not without limitations. First, the PBSM was validated among a sample of university students with particular sociodemographic characteristics, which is to say a population not necessarily representative of the entire adult population. Unfortunately, analyses of the differences between PBSM scales by gender and race have not been carried out. Second, we could have evaluated criterion validity using other indicators instead of just frequency of cannabis use, such as number of days of cannabis use, quantity of cannabis use, cannabis use consequences and substance abuse diagnosis. The fact that frequency of cannabis use was measured dichotomously rather than continuously can also be considered as a limitation. Third, there was a reduction in accounted variance for the French version of the scale compared with the English version. Though the difference is small, it suggests that the one-factor structure may be less well represented in French although this was the first attempt to evaluate the factor structure of the French version of the PBSM-17. This possibility should be investigated in future research, for instance using CFA.
Conclusion
The French-Canadian version of the 17-item short form of the PBSM-17 and its English version were found to be valid with Canadian university students. The legalisation of cannabis use affords the opportunity to set up prevention initiatives to promote safer consumption behaviours such as those measured by the PBSM. A significant research effort is critically needed for new, evidence-based public health prevention interventions to target PBS, using innovative strategies to encourage cannabis users to apply them.
Supplementary Material
Acknowledgements
We want to thank Emma Février for her assistance with data collection and Paul Di Biase for linguistic revision. We also wish to thank the participants who contributed to this study.
Data availability statement
Data are available on reasonable request. Data are available on reasonable request by contacting the corresponding author, JC.
Ethics statements
Patient consent for publication
Not applicable.
Ethics approval
This study was approved by the Centre Hospitalier de l’Université de Montréal (CHUM) Research Ethics Board (20.172) and by the Comité d’éthique de la recherche en sciences et en santé (CERSES) of the Université de Montréal (CERSES-20–114-D). All participants provided informed consent.