The TLFB continues to be a useful tool for assessing substance use quantity and frequency patterns, within the broader scope of validated substance use assessments that each vary on their research‐based and clinical utility (Carey, 2002). Importantly, the TLFB is not a diagnostic tool, but reveals frequency and quantity patterns and has the potential to measure fine‐grained details of use, especially for substances that are newly emerging (such as e‐cigarettes or edible cannabis products for instance) and for which the field is still exploring the best way to examine and quantify.

While popular and well‐validated, the TLFB requires a relatively labor‐intensive in‐person administration. Further, while situations remain where an in‐person administration of this assessment is more appropriate than an online or computer‐based version (e.g., among patient groups or with respondents who have low computer literacy), an online assessment has many advantages. It may be more appropriate for a large community sample, or for multiple administrations across many months or years, and it increases participant convenience and privacy. The online format is easy and intuitive to use with basic Internet literacy possessed by nearly all Americans (Ryan, 2016), and finally, it offers uniformity of data capture and seamlessly integrates with REDCap, a method of data management that is increasingly utilized in biological and behavioral research, including clinical settings (Bahr, Christensen, Agarwal, George, & Bhutani, 2019; Blough, Mansfield, & Kondapalli, 2014; Campion, Sholle, & Davila, 2017; Cantor, Plint, Kamil, & Zemek, 2016; Dunn et al., 2016; Gabriel, Finlay, & Weiner, 2012; Obeid et al., 2013; Yamamoto et al., 2017). Data obtained from the O‐TLFB demonstrated a high level of consistency with the in‐person TLFB interview and strong concurrent validity with the MDS and AUDIT. Our findings extend previous research on the feasibility and validity of administering the TLFB in an online format (Hjorthøj, Fohlmann, Larsen, Arendt, & Nordentoft, 2012; Hoeppner et al., 2010; Rueger et al., 2012).

Of utmost importance, this O‐TLFB has the ability to query an increased diversity of substances (e.g., prescription drugs and illicit drugs) in greater detail in terms of method of administration (e.g., cannabis could be endorsed through flower, concentrate, or edible use and nicotine/tobacco could be endorsed through cigarettes, e‐cigarettes, chew or dip, cigars, hookah, or other tobacco), dosage (differentiated according to mode of administration, e.g., pinch, puff, and drag), and potency (e.g., THC vs. CBD proportions in cannabis products), and in flexible combinations based on the need of the study (i.e., one study may query prescription medications while another might omit them, and one study may query 14 retrospective days awhile another may query 60). Capturing these specifics online as accurately as the in‐person version of the TLFB is important in today's climate of increased legalization and accessibility of cannabis, use of prescription medications recreationally, and rapidly changing forms of nicotine consumption. It is interesting to note, however, that in the course of the validation of the O‐TLFB, an ancillary finding regarding marker date use (a retrospective calendar as a memory “trigger” for instances of substance use) became apparent. While it was shown that participants made greater use of marker dates online than in person, overall the results indicated that the two modalities are highly correlated across primary and secondary substance use measures despite this difference.

Our study should be interpreted in the context of the following limitations. First, disclosures of the true intent to compare the TLFB methods were given to participants at the end of their in‐person interview. Therefore, participants who interviewed before taking the O‐TLFB had the chance to read the disclosure and understand the purpose of the study before they completed the O‐TLFB, potentially influencing the study's results. Despite this, however, substance use reporting levels across both groups were consistently related across all domains.

Second, the location of the study in Colorado—a state that has legalized the medical and recreational use of cannabis—and the use of a convenience sample of college students are also limitations that may have also influenced results. For example, while we demonstrated preliminary reliability for the reporting of both quantity and frequency across multiple forms of substances, the differences between the online and in‐person measures for nicotine/tobacco total days and alcohol total drinks were significantly different. This could be due to the fact that use in our sample was notably limited, though it should be noted that the actual differences were small and do not appear to be clinically meaningful and could benefit from further validation among community or clinical samples. Also, while our measure demonstrated reliable reporting of days and amounts used for different modes of cannabis administration, such as flower, edibles, and concentrates, the majority of participants interestingly did not report on the specific THC or CBD contents of their cannabis as demonstrated by our ancillary findings. This limitation could be due to our reliance on a sample that was on average below the legal age for purchasing and using cannabis (21), and thus more likely to use cannabis in groups socially where potency is not known by users. Therefore, these potency measures could be more powerful in studies of older or medical cannabis users or heavier substance users. Again, this validation study using a convenience sample is a first step toward demonstrating the validity of these novel potency and administration measures among diverse samples.

Finally, participants may also have been more comfortable and had more knowledge reporting on their cannabis and other substance use in our state, as a result of the state cannabis packaging regulations and general social climate surrounding use, which in turn may limit the replicability of this study in other states. Future directions would include additional validation with a community or clinical sample to further explore the validity of substance categories measured by the O‐TLFB. Overall, however, the data suggest that this new O‐TLFB demonstrates reliability across a wide variety of substances in greater detail. It delivers a useful addition to the broader substance use measurement arsenal that is relevant to a variety of biomedical and psychological research contexts.