The Modified Cigarette Evaluation Questionnaire: Psychometric Properties of the Originally-Proposed Five-Factor Structure and a Novel Four-Factor Structure for Use with Adults who Currently Smoke Cigarettes

Abstract

Background.

The Modified Cigarette Evaluation Questionnaire (MCEQ) measures self-reported, subjective reinforcing and aversive effects of smoking. The MCEQ is proposed as comprising five factors. However, its latent factor structure has been confirmed in only one published study, and key psychometric properties like measurement invariance and test-criterion validity have not been formally examined.

Methods.

We examined the latent structure of the MCEQ using confirmatory and exploratory factor analysis and evaluated internal consistency, measurement invariance, between-group differences, and relationships with smoking-related characteristics in a sample of 764 adult smokers. As a secondary test of construct validity, we fit identified latent factor structures to a sample of 131 adults from a prior trial who were biochemically-confirmed to currently smoke.

Results.

A four-factor structure that recently was validated for an e-cigarette-adapted version of the MCEQ outperformed the originally-proposed five-factor structure in several ways: it uniquely evidenced good model fit both within the primary and replication samples, reached scalar invariance for all participant subgroups tested, was sensitive to detecting numerous between-groups differences in subsamples of interest, and each of its subscales was associated with at least one smoking characteristic assessed.

Conclusions.

The four-factor MCEQ structure evidenced stronger psychometric properties in our sample than did the original five-factor MCEQ. However, until additional research is conducted on the utility of the five- and four-factor solutions in other samples, neither the five- nor the four-factor MCEQ structure should be assumed. In the meantime, researchers should examine the MECQ’s latent structure in their data prior to conducting subsequent analyses.

1.0. INTRODUCTION

Combustible cigarette use remains a leading cause of death worldwide (U.S. Department of Health and Human Services, 2014), and understanding factors that promote smoking and deter cessation continues to be an important topic of investigation. Subjective response to smoking, which reflects individual differences in the experience of positive and negative pharmacological effects of nicotine as well as smoking-related sensations like upper airway stimulation (e.g., throat-hit), is a risk factor for cigarette use, the development of dependence, and difficulty quitting smoking (e.g., Cohn et al., 2020). In addition, subjective response may play a role in smoking cessation, as cessation medications (e.g., varenicline) often decrease positive subjective effects like satisfaction and/or increase negative subjective effects like dizziness (e.g., Brandon et al., 2011; Cousins et al., 2001; King & Meyer, 2000). Despite the importance of subjective response, most measures of the subjective effects of smoking have not undergone sufficient psychometric evaluation to be considered reliable and valid instruments. The current study focuses on The Modified Cigarette Evaluation Questionnaire (MCEQ; Cappelleri et al., 2007), a revised version of The Cigarette Evaluation Questionnaire (Westman et al., 1992) that has been widely used to assess self-reported subjective effects of cigarette smoking.

Based on the original confirmatory factor analyses, which indicate the number of latent factors (i.e., subscales) that a measure comprises as well which items reflect each factor (Cappelleri et al., 2007), the 12-item MCEQ is proposed to comprise five subscales: Smoking Satisfaction, Psychological Reward, Enjoyment of Respiratory Tract Sensations, Craving Relief, and Aversion. Although we were unable to find evidence of the original theory underlying the MCEQ, the four subscales of the MCEQ with a positive valence presumably capture the immediately reinforcing effects of smoking that may prompt continued use, deter cessation efforts, and relate to addictive potential (Cappelleri et al., 2007). Similarly, experiencing low levels of aversion, especially in conjunction with experiencing reinforcing effects, may contribute to continued use (Cappelleri et al., 2007). In prior work (Cappelleri et al., 2007), each subscale had acceptable test-retest reliability, and internal reliability was good for Smoking Satisfaction and Psychological Reward. However, internal reliability was insufficient for Aversion and could not be calculated for the two single-item subscales (i.e., Enjoyment of Respiratory Tract Sensations and Craving Relief). Importantly, alternative factor solutions that may have evidenced superior fit were not tested (or at least published), and critical psychometric analyses were not examined including measurement invariance (i.e., the extent to which a construct is measured with sufficient equivalence across groups to ensure that the construct measures the same thing and is captured by the measure in the same way within subgroups of interest [Putnick & Bornstein, 2016]) and test-criterion validity (i.e., examining relationships between the measure and outcomes of interest).

Although numerous papers have examined the MCEQ in relation to a range of smoking characteristics, potentially adding credence to its test-criterion validity, no further papers explicitly examining the latent structure of the MCEQ have been published. This raises concerns about the validity of the originally proposed scoring for use in other samples and the interpretations that can be drawn from research using the MCEQ. Although the data were not published, a conference poster (Salzberger et al., 2018) presented the results of an exploratory factor analysis of the MCEQ among adults who smoke cigarettes. Three factors were permitted to be extracted to account for the originally-proposed, multi-item subscales (i.e., Smoking Satisfaction, Psychological Reward, Aversion). While the goal was not explicitly to validate the latent structure, the only subscale that performed exactly as proposed (Cappelleri et al., 2007) was “Aversion,” with both items loading strongly onto their primary factor (dizzy [0.87]; nauseous [0.81]) and no significant cross-loadings on the other factors. For Psychological Reward, four of the five proposed items loaded strongly onto the factor with no significant cross-loadings. However, the item “smoking calms me down” was cross-loaded, with a stronger loading on Smoking Satisfaction (0.51) than on Psychological Reward (0.43), meaning that the item did not strongly represent a single latent factor. As proposed, the three items for the Smoking Satisfaction subscale evidenced the strongest loadings on that factor (i.e., satisfying [0.96], taste good [0.94], and enjoy [0.90]) with no significant cross-loadings. However, the items originally proposed to comprise the two single-item subscales - Enjoyment of Respiratory Tract Sensations and Craving Reduction - also loaded strongly onto Smoking Satisfaction (i.e., 0.78; 0.68) with no significant cross-loadings. Given that the five-factor structure originally proposed by Cappelleri and colleagues (2007) was not supported in these data, scoring the measure as originally instructed would be problematic as it did not reflect the reality of the data.

Also speaking to the possible viability of alternate factor structures, a recently published paper (Morean & Bold, 2022) that used data from this parent study identified an 11-item, novel four-factor structure for an e-cigarette-adapted version of the MCEQ, with factors comprising Stimulant Effects (more awake, help concentrate, reduce hunger), Positive Reinforcement (taste good, enjoy the sensations in your throat and chest, enjoy smoking), Negative Reinforcement (calm down, less irritable, relieves craving), and Aversion (dizzy, nauseous). Of note, the originally-proposed five-factor solution also fit the data for e-cigarettes. Taken together, the results from the conference presentation and the e-cigarette-adapted MCEQ question the universality of the factor structure proposed by Cappelleri and colleagues (2007) and suggest that additional psychometric work on the MCEQ is needed.

To address the gaps in the literature, we conducted a series of psychometric analyses on the MCEQ in a sample of adults who smoke cigarettes. Primary analyses were conducted in the subsample of adults who reported currently smoking cigarettes (n = 764) from the larger parent study from which the psychometric properties of an e-cigarette adapted MCEQ also were evaluated (N = 857; Morean & Bold, 2022). With regard to examining the latent structure of the MCEQ, we first used confirmatory factor analysis (CFA) to determine whether the 12-item, five-factor structure (Cappelleri et al., 2007) was supported in our data. Next, we conducted a CFA to evaluate model fit for the novel 11-item, four-factor solution that recently was identified for the e-cigarette-adapted MCEQ (Morean & Bold, 2022). We then used exploratory factor analysis to determine whether an alternative factor solution(s) better reflected the latent structure of the MCEQ in our data than did the previously established five-factor (Cappelleri et al., 2007) or four-factor models (Morean & Bold, 2022). To provide further confirmation of all identified latent factor structure(s), we fit all plausible structures to the data from a previously conducted clinical trial of adults who were confirmed to smoke cigarettes and who were seeking treatment for alcohol use disorder (O’Malley et al., 2018).

For all latent structures that fit our data, we then tested measurement invariance to assesses the extent to which the subjective effects of cigarette smoking were measured with sufficient equivalence across groups (e.g., daily versus non-daily smokers) (Putnick & Bornstein, 2016). We subsequently examined between-group differences for all subgroups for which scalar invariance, the level of invariance that is required to justify conducting between-groups analyses, was established. Finally, as evidence of test-criterion validity, we examined unadjusted and adjusted relationships between the MCEQ variants and smoking-related characteristics (e.g., smoking frequency, cigarettes smoked per day, cigarette dependence).

Study findings will have important implications for research using the MCEQ moving forward. If the five-factor structure fit the data well and evidenced other strong psychometric properties, this would support scoring the MCEQ as originally proposed and would strengthen the interpretability of previously published studies. If the four-factor structure identified for the e-cigarette-adapted MCEQ (Morean & Bold, 2022) fit the data, this would suggest that this novel factor structure may be viable and requires additional study. Finally, if an alternative latent structure were observed, this would suggest that the theoretical organization and scoring of the MCEQ may need to be reconceptualized entirely. If any factor structure(s) other than the original five-factor model were supported, this would suggest that, at minimum, researchers should conduct a confirmatory factor analysis to examine fit of the MCEQ to their data before proceeding with scoring the measure and conducting further analyses.

2.0. MATERIAL AND METHODS

The Yale University Institutional Review Board approved all study procedures.

2.1. Participants and Procedures

A total of 857 adults participated in an anonymous, online, 20-minute survey in Summer 2021. Qualtrics Online Sample, a secure market research service, recruited and compensated participants. Eligibility criteria for the parent study included living in the US, being at least 21 years old, smoking cigarettes for at least 1 year, and using e-cigarettes in a quit smoking attempt in the past two years. The analytic sample for the secondary data analyses presented here comprised 764 adults who reported currently smoking cigarettes at the time of survey completion (Table 1).

Table 1.

Participant characteristics

Participant Characteristics	% (n) or Mean (SD)
Male Sex	54.7% (418)
Age	40.87 (12.17)
Non-Hispanic White	64.7% (494)
Age at Smoking Onset (in years)	18.83 (6.91)
Total Number of Years Smoked	17.47 (12.66)
Smoking Frequency (# Days/Week)	6.62 (0.79)
Daily Smoking (Yes)	78.0% (596)
Smoking Quantity (Cigs/Day)	11.57 (7.63)
Smoke at Least A Pack Per Day (Yes)	21.6% (165)
Menthol Cigarette Use (Yes)	57.1% (436)
Cigarette Dependence	2.48 (0.90)
Dual-Use of Cigarettes & E-cigarettes (Yes)	74.5% (569)
MCEQ (original 5 factor)
Satisfaction	5.13 (1.33)
Psychological Reward	4.90 (1.23)
Respiratory Tract	4.56 (1.81)
Craving Reduction	5.45 (1.37)
Aversion	2.38 (1.80)
MCEQ (revised 4 factor)
Stimulant Effects	4.63 (1.43)
Positive Reinforcement	4.85 (1.45)
Negative Reinforcement	5.35 (1.14)
Aversion	2.38 (1.80)

Note. Data for categorical variables are presented as percent. Data for continuous variables are presented as Mean (Standard Deviation). Scaling for variables was as follows: Cigarette Dependence (0 = never, 1 = rarely, 2 = sometimes, 3 = often, 4 = almost always; range 0–4); Modified Cigarette Evaluation Questionnaire (1 = not at all, 2 = very little, 3 = a little, 4 = moderately, 5 = a lot, 6 = quite a lot, 7 = extremely; range 1–7).

As noted, our analytic sample reflected a subsample of the full dataset used to evaluate the psychometric properties of the e-cigarette-adapted MCEQ. To mitigate concerns about participant overlap, we used previously collected MCEQ data from a clinical trial (O’Malley et al., 2018) as a secondary test of the validity of all identified latent structures. The sample comprised 131 adults who enrolled in a treatment study to examine the impact of Varenicline on Alcohol Use Disorder (AUD; 29.8% female, mean age 42.7 (SD =11.7) years, 52.7% Black/African American). All participants reported currently smoking cigarettes, reported not seeking smoking cessation treatment, and met diagnostic criteria AUD. Full details on participant recruitment and study findings are presented elsewhere (O’Malley et al., 2018).

2.2. Measures

2.2.1. Demographics.

Participants reported on age, sex (female/male), Hispanic ethnicity (no/yes), and race (select all that apply from White, Black, Asian, Native American, Pacific Islander/Native Hawaiian, Other). We dichotomized ethnicity/race as non-Hispanic White versus other for study analyses due to small sample sizes among specific ethnic/racial minorities.

2.2.2. Cigarette Use and Dependence.

To assess smoking history, we queried participants about age at smoking onset (“How old were you when you tried a cigarette for the first time, even 1 or 2 puffs?” ____# years) and their total duration of cigarette use (“For how many years have you smoked cigarettes?” ____# years). To assess current cigarette use, we assessed both smoking frequency (“How many days per week do you typically smoke cigarettes?” [0–7 days/week]) and quantity (“How many cigarettes do you typically smoke per day?” [1 cigarette per day to more than 40]). Variables reflecting daily smoking (no/yes) and smoking at least a pack per day (no/yes) were calculated from cigarette smoking frequency and quantity, respectively, for invariance testing. We also assessed whether participants currently smoke menthol cigarettes (“What kind of cigarettes do you usually smoke?” [menthol, regular/non-menthol], both menthol and regular/non-menthol)]. A variable reflecting the any menthol cigarette use (no/yes) was calculated for invariance testing. Finally, we assessed cigarette dependence using the psychometrically sound, PROMIS Short Form v1.0 - Smoking Nicotine Dependence for All Smokers 4a (Hansen et al., 2014).

2.2.3. Subjective Response.

The Modified Cigarette Evaluation Questionnaire was used to assess self-reported reinforcing and aversive effects of smoking. The originally proposed 12-item, five-factor solution includes four positive subscales (i.e., Smoking Satisfaction, Psychological Reward, Enjoyment of Respiratory Tract Sensations, Craving Reduction) and one negative subscale (i.e., Aversion; Cappelleri et al., 2007). All items are included in Table 2. The measure uses a 7-point response scale ranging from “not at all” to “extremely.” Each multi-item subscale is scored by taking the mean of its items. For the two single-item subscales, the value of the single item is the subscale score.

Table 2.

Confirmatory factor analysis for the original five-factor and new four-factor structures of the MCEQ

Original Five-Factor Model			New Four-Factor Model
Model Fit Indices			Model Fit Indices
Root Mean Square Error of Approximation	0.074 (0.065–0.085)		Root Mean Square Error of Approximation	0.053 (0.042–0.064)
Bentler’s Comparative Fit Index	0.943		Bentler’s Comparative Fit Index	0.973
Standardized Root Mean Square Residual	0.045		Standardized Root Mean Square Residual	0.034
χ2 (41) = 216.85***			χ2 (35) = 108.82***
MCEQ Items by Subscale	Loading	S.E.	MCEQ Items by Subscale	Loading	S.E.
Smoking Satisfaction	α = 0.82		Stimulant Effects	α = 0.76
1. Cigarettes are satisfying	0.79	0.03	5. Smoking makes me feel more awake	0.73	0.03
2. Cigarettes taste good	0.83	0.02	7. Smoking helps me concentrate	0.71	0.03
12. I enjoy smoking	0.80	0.02	8. Smoking reduces my hunger for food	0.60	0.03
Psychological Reward	α = 0.80		Positive Reinforcement	α = .82
4. Smoking calms me down	0.68	0.03	2. Cigarettes taste good	0.81	0.02
5. Smoking makes me feel more awake	0.74	0.03	3. I enjoy the sensations of smoking in my throat and chest	0.79	0.03
6. Smoking makes me feel less irritable	0.59	0.03	12. I enjoy smoking	0.82	0.03
7. Smoking helps me concentrate	0.71	0.03	Negative Reinforcement	α = 0.71
8. Smoking reduces my hunger for food	0.60	0.03	4. Smoking calms me down	0.81	0.03
Enjoyment of Respiratory Tract Sensations	--		6. Smoking makes me feel less irritable	0.64	0.04
3. I enjoy the sensations of smoking in my throat and chest	1.00	0.00	11. Smoking immediately relieves my craving for a cigarette	0.60	0.04
Craving Reduction	--		Aversion	α = 0.90
11. Smoking immediately relieves my craving for a cigarette	1.00	0.00	9. Smoking makes me dizzy	0.85	0.04
Aversion	α = 0.90		10. Smoking makes me nauseous	0.96	0.03
9. Smoking makes me dizzy	0.86	0.04
10. Smoking makes me nauseous	0.95	0.04

Note. N = 764

^***p < .001 = Cronbach’s alpha; S.E. = Standard Error

Factor loadings are standardized. The values in parentheses represent the 95% confidence interval for RMSEA. Suggested cutoffs for good model fit are: Bentler’s Comparative Fit Index > 0.95, Root Mean Square Error of Approximation < 0.06, and Standardized Root Mean Square Residual < 0.05.

Recently, the psychometric properties an e-cigarette-adapted MCEQ were tested (Morean & Bold, 2022). The same response scale and scoring were employed as in the original MCEQ, but the wording of items was altered to reflect e-cigarette use (e.g., “vaping” instead of “smoking”). The original five-factor solution was supported, but a novel, psychometrically-sound, four-factor structure also was supported. The novel four-factor solution comprised 11-items and 4 subscales: Stimulant Effects, Positive Reinforcement, Negative Reinforcement, and Aversion. Given the good fit to the data of the four-factor e-cigarette-modified MCEQ (Morean & Bold, 2022), we examined the fit of a four-factor structure in the current data using the original cigarette item content.

2.2.4. E-cigarette Use.

Participants reported on past-30-day e-cigarette use. Those who reported any use in the past month were characterized as dually using both cigarettes and e-cigarettes. Those who reported no e-cigarette use were characterized as exclusively using cigarettes.

2.3. Ensuring Data Quality

Regarding data quality, Qualtrics Online Sample guarantees “good completes,” meaning quality, usable data. Qualtrics automatically includes several functions to help eliminate poor-quality responders. First, a participant must reach the end of the survey (although they may skip questions per IRB requirements). Second, Qualtrics prohibits participants, who have unique panelist IDs, from taking the survey multiple times even if done from different IP addresses. Third, Qualtrics excludes participants who complete the survey in an unreasonably short amount of time. In addition, Qualtrics allows researchers to embed attention check questions into surveys to identify inattentive responders. We included two of attention questions in this study, and have found that attention check questions work best when the item stem mirrors the real items and is matched for length, making them difficult to identify. Qualtrics also will replace respondents at the request of the researcher if issues are identified once the data have been collected. MM reviewed all responses for indications of poor responding (e.g., nonsensical write-in responses, straight-lining, evidence of random responding) and requested replacements with “good completes” as necessary.

2.4. Data Analytic Plan

We used Mplus 8.6 for conducting all analyses to examine the latent structure and measurement invariance of the MCEQ. We used SPSS 27 to conduct all remaining analyses.

2.4.1. Confirming the five-factor latent structure of the MCEQ.

To determine if the originally proposed MCEQ structure (Cappelleri et al., 2007) was supported in our data, we conducted a CFA. We specified full-information maximum likelihood (FIML) with robust standard errors to handle non-normally distributed data and to generate model fit indices. Although FIML also can be used to handle missing data, there were no missing data. The following were used as indices of model fit: Bentler’s Comparative Fit Index (CFI; mediocre: > 0.90; good: > 0.95; Hu & Benter, 1999), Root Mean Square Error of Approximation (RMSEA; mediocre: < 0.08; good: < 0.06; Hu & Bentler, 1999; MacCallum et al., 1996), and Standardized Root Mean Square Residual (SRMR; mediocre: < 0.08; good: < .05; Byrne, 1998; Hu & Bentler, 1999). Although chi-square statistics are included in Table 2, we did not interpret these values given that large samples yield significant yet unreliable chi-square statistics (Chen, 2007)

2.4.2. Confirming a novel, four-factor latent structure for the MCEQ.

To test whether the novel 11-item, four-factor structure that recently was identified for the e-cigarette-adapted MCEQ (Morean & Bold, 2022) was supported in our data, we conducted a second CFA. The analytic approach mirrored that described above.

2.4.3. Using exploratory/confirmatory factor analysis to identify alternative factor solutions for the MCEQ.

Within a randomly selected 50% of the dataset, we conducted an EFA with an oblique rotation (i.e., goemin) to identify plausible alternative factor structures. To identify viable alternative solutions, we relied on a combination of model fit indices, model comparisons, factor interpretability, and the number of items per factor. If a novel, alternative solution were generated, we planned to use CFA to fit the model to the remaining data.

2.4.4. Confirming all plausible models in an independent dataset.

Given that the four-factor structure for the e-cigarette MCEQ was derived from the full dataset of the parent study, participant responses for the MCEQ and, subsequently, for the e-cigarette MCEQ may have been similar given that the same participants completed both measures (although all participants completed the MCEQ first). As a secondary test of the validity of the identified latent structures, we fit all viable models using CFA to the separate clinical trial data of 131 adults who were biochemically-confirmed to currently smoke cigarettes (O’Malley et al., 2018). The models were fit to the baseline data obtained prior to treatment.

2.4.5. Evaluating the internal reliability of the MCEQ subscales.

To assess internal consistency, we calculated Cronbach’s alpha values for all subscales comprising multiple items.

2.4.6. Evaluating relationships between different versions of the MCEQ subscales.

We used bivariate correlations to examine relationships between the subscales of the different variants of the MCEQ (e.g., the five-factor vs the four-factor).

2.4.7. Evaluating measurement invariance of the latent structures of the MCEQ.

We used a multigroup CFA approach to test invariance by sex (female/male), ethnicity/race (Non-Hispanic White/other), menthol cigarette use (no/yes), smoking frequency (daily/non-daily), smoking at least a pack per day (no/yes), and current exclusive cigarette use versus dual-use of cigarettes and e-cigarettes. For each subgroup, we tested three tiers of measurement invariance.

First, we tested configural invariance by fitting the same basic latent structure (i.e., number of factors; items per factor) across groups to determine whether the same latent factors, comprising the same items, were supported in each group. Second, we tested metric invariance by constraining the factor loadings to be equivalent to determine whether each item was reflected by the latent construct (i.e., the factor/subscale) similarly across groups. Finally, we tested scalar invariance by additionally constraining the item intercepts to be equivalent across groups to ensure that differences in group means were not due to differential scale properties.

Configural invariance was achieved if model fit indices suggested acceptable fit to the data, with all items loading significantly onto their designated factors. Given that each successive level of invariance imposes increasingly stringent constraints on the model, model fit may worsen as constraints are imposed at the levels of metric and scalar invariance. Thus, the acceptability of model fit at each level is based on minimized decrement in model fit. Metric invariance was achieved if decrements in model fit from the configurally-invariant model did not exceed RMSEA ≥ .015, CFI ≥ .01, or SRMR ≥ .03 when we constrained the factor loadings to equality (Chen, 2007). Scalar invariance was achieved if decrements in model fit from the metrically-invariant model did not exceed CFI ≥ .01 (plus a decrement in SRMR ≥ .01 or RMSEA ≥ .015) when we constrained both factor loadings and intercepts to equality (Byrne, 1998). An MCEQ version was considered sufficiently invariant if scalar invariance was achieved, given that this is a requirement for conducting between-group differences (Chen, 2008; Steenkamp & Baumgartner, 1998); if scalar invariance is not established, between-groups differences may be attributable to error from various unknown sources (e.g., the items fundamentally mean something different to the different groups) and cannot be interpreted meaningfully.

2.4.8. Evaluating between-group differences in MCEQ scores.

For all subgroups for which scalar invariance was achieved, we used independent-samples t-tests to evaluate between-groups differences in MCEQ subscale scores. Given that 54 analyses were conducted, we used a Bonferroni-adjusted alpha value < .001 as the threshold for statistical significance to account for familywise error (calculated as the standard alpha value of 0.05/54).

2.4.9. Examining unadjusted and adjusted relationships between MCEQ scores and smoking characteristics.

We used univariate general linear modeling (GLM) to examine unadjusted relationships between the MCEQ subscales and age at smoking onset, total number of years smoked, smoking frequency (days/week), cigarettes smoked per day, and cigarette dependence to determine the impact of the subscales themselves. This is akin to assessing basic concurrent validity. However, we included smoking frequency in the model for cigarette dependence to ensure that findings were not driven by the unaccounted impact of smoking frequency. An adjusted alpha value of < 0.01 served as the threshold for statistical significance (i.e., 0.05/5 models).

To examine adjusted relationships, we repeated the GLM analyses to examine whether the subscales accounted for unique variance in smoking characteristics after accounting for model covariates. This is a form of testing cross-sectional, incremental validity. Sex, age, and race/ethnicity were included as covariates based on research suggesting that men, older individuals (i.e., 24–64 years compared to 18–24 years), and individuals from multiple ethnic/racial minority groups are more likely to smoke than are their respective counterparts (Cornelius et al., 2022). Menthol use was included as a covariate based on research suggesting that menthol cigarette use is associated with increased addiction potential, dependence, and failed quit attempts leading to a longer duration of smoking (e.g., Villanti et al., 2017). For the model predicting cigarette dependence, we again adjusted for smoking frequency.

2.4.10. Accounting for familywise error.

Within the manuscript text, we only discuss findings that were significant at the thresholds described in the sections above (i.e., p-values < .001 for between-groups differences [section 2.3.8], p-values < .01 for test-criterion validity [section 2.3.9]). However, we present findings that were significant at all levels (i.e., p < .05, p < .01, and p < .001) in the corresponding tables for reference given that Bonferroni corrections may increase Type II error and future studies may use the MCEQ to address narrower study aims that require fewer or no adjustments to alpha.

3.0. RESULTS

3.1. Descriptive Statistics

Descriptive statistics are presented in Table 1.

3.2. Confirming the five-factor and four-factor latent structures of the MCEQ

The five-factor latent structure evidenced mediocre fit to our data (Table 2), with two of the three fit indices assessed (i.e., CFI and RMSEA) falling outside of the range for good fit. The four-factor model evidenced good fit to the data across all fit indices. For both the five-factor and four-factor models, each item loaded significantly onto its designated factor (p-values < .001).

3.3. Exploratory/confirmatory factor analysis to identify an alternative factor solution for the MCEQ

The EFA including all 12 items produced one-, two-, three-, and four-factor solutions. The one- and two-factor models fit the data poorly. Model comparisons showed that the four-factor solution evidenced superior fit (p < .001; RMSEA = 0.032, CFI = 0.994, SRMR = 0.011) to the three-factor solution (RMSEA = 0.66, CFI = 0.965, SRMR = 0.026), suggesting that only the four-factor model should be retained. Upon examination, the four factors mirrored those obtained for the e-cigarette-adapted MCEQ (Morean & Bold, 2022; Supplemental Table 1).

3.4. Confirming all plausible models in an independent dataset

The five-factor model did not evidence adequate fit in the independent dataset (RMSEA = 0.085, CFI = 0.931, SRMR = 0.051). The four-factor model fit the data well (RMSEA = 0.052, CFI = 0.974, SRMR = 0.051).

3.5. Evaluating the internal reliability of the MCEQ subscales

For the five-factor solution, each multi-item subscale evidenced good internal reliability (Smoking Satisfaction [α = 0.82]; Psychological Reward [α = 0.80]; Aversion [α = 0.90]). For the four-factor solution, each subscale evidenced acceptable to good internal consistency (Stimulant Effects [α = 0.76]; Positive Reinforcement [α = 0.82]; Negative Reinforcement [α = 0.71]; Aversion [α = 0.90]).

3.6. Evaluating relationships between the two versions of the MCEQ subscales.

The subscales of the five-factor MCEQ were significantly correlated with the subscales of the four-factor MCEQ (Supplemental Table 2).

3.7. Evaluating measurement invariance of the latent structures of the MCEQ

The five-factor model failed to reach scalar invariance for sex and smoking frequency (daily vs. non-daily), two key subgroups of interest. The four-factor model was scalar invariant for all subgroups tested (Supplemental Table 3).

3.8. Evaluating between-group differences in MCEQ scores

3.8.1. The original five-factor model.

Because scalar measurement invariance was not achieved for sex or daily smoking status, conducting between-groups analyses for these subgroups was not supported. Individuals who reported smoking less than a pack per day reported more Aversion than did those who smoked a pack per day or more (Table 3). Additionally, menthol users reported more Smoking Satisfaction, Psychological Reward, and Enjoyment of Respiratory Sensations than did non-menthol users. Finally, adults who reported using both cigarettes and e-cigarettes reported more Smoking Satisfaction, Psychological Reward, Enjoyment of Respiratory Sensations, and Aversion, but less Craving Reduction than did those who exclusively used cigarettes. No other significant between-groups differences were observed.

Table 3.

Between-groups comparisons of the five-factor and four-factor MCEQ subscales

			Original Five-Factor Structure
			Smoking Satisfaction	Psychological Reward	Respiratory Tract	Craving Reduction	Aversion
Sex	Male	418	5.30 (1.23)***	4.98 (1.25)*	4.88 (1.65)***	5.32 (1.37)	2.72 (2.01)***
	Female	346	4.93 (1.41)	4.80 (1.21)	4.16 (1.92)	5.60 (1.35)***	1.96 (1.41)
Non-Hispanic White	Other	270	5.10 (1.24)	4.87 (1.26)	4.57 (1.69)	5.26 (1.41)	2.27 (1.62)
Non-Hispanic vs Other Ethnicity/Race	White	494	5.15 (1.37)	4.92 (1.22)	4.55 (1.88)	5.55 (1.33)**	2.44 (1.89)
Daily Smoking	No	168	5.19 (1.04)	4.96 (1.10)	4.86 (1.51)*	5.01 (1.27)	2.93 (1.96)***
	Yes	596	5.12 (1.39)	4.88 (1.27)	4.47 (1.88)	5.56 (1.37)***	2.23 (1.72)
Pack Per Day Smoking	No	599	5.12 (1.30)	4.89 (1.21)	4.59 (1.80)	5.38 (1.36)	2.48 (1.84) ***
	Yes	165	5.19 (1.43)	4.93 (1.33)	4.43 (1.87)	5.67 (1.37)*	2.00 (1.60)
Menthol Cigarette Use	No	328	4.92 (1.41)	4.72 (1.20)	4.20 (1.92)	5.50 (1.36)	2.18 (1.67)
	Yes	436	5.29 (1.24) ***	5.04 (1.24) ***	4.83 (1.68) ***	5.40 (1.38)	2.52 (1.89)**
Cigarette Only vs Dual Use of Cigarettes and E-cigarettes	Cigarette Only	195	4.72 (1.49)	4.55 (1.29)	3.89 (1.95)	5.64 (1.27) *	1.99 (1.51)
	Dual Use	569	5.27 (1.24) ***	5.02 (1.19) ***	4.79 (1.71) ***	5.38 (1.39)	2.51 (1.87) ***
			New Four-Factor Structure
			Stimulant Effects	Positive Reinforcement	Negative Reinforcement	Aversion
Sex	Male	418	4.79 (1.42) ***	5.12 (1.34) ***	5.28 (1.13)	2.72 (2.01) ***
	Female	346	4.43 (1.42)	4.54 (1.52)	5.43 (1.16)**	1.96 (1.41)
Non-Hispanic White vs Other Ethnicity/Race	Other	270	4.63 (1.45)	4.83 (1.33)	5.23 (1.17)	2.27 (1.62)
	Non-Hispanic White	494	4.63 (1.42)	4.87 (1.52)	5.42 (1.13)	2.44 (1.89)
Daily Smoking	No	168	4.88 (1.21) ***	5.06 (1.12)*	5.06 (1.06)	2.93 (1.96) ***
	Yes	596	4.56 (1.48)	4.80 (1.53)	5.43 (1.16) ***	2.23 (1.72)
Pack Per Day Smoking	No	599	4.65 (1.40)	4.87 (1.43)	5.30 (1.11)	2.48 (1.84) ***
	Yes	165	4.57 (1.54)	4.80 (1.54)	5.53 (1.24)*	2.00 (1.60)
Menthol Cigarette Use	No	328	4.36 (1.44)	4.55 (1.53)	5.33 (1.15)	2.18 (1.67)
	Yes	436	4.83 (1.40) ***	5.08 (1.35) ***	5.37 (1.14)	2.52 (1.89)**
Cigarette Only vs Dual Use of Cigarettes and E-cigarettes	Cigarette Only	195	4.09 (1.52)	4.28 (1.59)	5.37 (1.19)	1.99 (1.51)
	Dual Use	569	4.82 (1.35) ***	5.05 (1.35) ***	5.35 (1.13)	2.51 (1.87) ***

Note.

^*p < .05

^**p < .01

^***p < .001

Data in the table reflect independent-samples t-tests. Gray highlighting indicates comparisons that were not supported because scalar invariance was not achieved for these participant subgroups. Data are presented as Mean (Standard Deviation). Sex (female/male), Daily Smoking (no/yes), Pack Per Day (less than a pack vs a pack or more per day), Menthol Cigarette Use (no/yes), Cigarette Only vs Dual Use (adults who currently only smoke cigarettes vs adults who currently smoke cigarettes and use e-cigarettes).

3.8.2. The new four-factor model.

Men and adults who reported using both cigarettes and e-cigarettes reported stronger Stimulant Effects, Positive Reinforcement, and Aversion than did women and adults who reported exclusively using cigarettes, respectively (Table 3). Those who smoked daily reported more Negative Reinforcement but less Stimulant Effects and Aversion compared to those who smoked less than daily. Individuals who smoked less than a pack per day reported more Aversion than did those who smoked more. Finally, adults who smoked menthol cigarettes reported stronger Stimulant Effects and Positive Reinforcement than did adults who smoked non-menthol cigarettes. No other significant between-groups differences were observed.

3.9. Examining unadjusted relationships between MCEQ scores and smoking characteristics

3.9.1. The original five-factor model.

Craving reduction and aversion were associated significantly with each smoking characteristic assessed (Supplemental Table 4). Stronger craving reduction was associated with an earlier age of onset, a longer total smoking duration, more frequent smoking, smoking more cigarettes per day, and greater dependence. Stronger aversion was associated with an older age of onset, a shorter total smoking duration, less frequent smoking, smoking fewer cigarettes per day, and greater dependence. Greater enjoyment of respiratory tract sensations was associated only with a shorter total smoking duration, while greater psychological reward was associated only with greater dependence. Smoking satisfaction was not significantly associated with any smoking characteristic.

3.9.2. The new four-factor model.

Aversion and negative reinforcement were associated significantly with each smoking characteristic assessed. The effects for aversion were the same as described above for the five-factor model. Experiencing stronger negative reinforcement was associated with an earlier age at smoking onset, a longer total smoking duration, more frequent smoking, smoking a greater number of cigarettes per day, and greater dependence. Experiencing more stimulant effects was associated with smoking fewer cigarettes per day but with increased dependence symptoms. Experiencing stronger positively reinforcing effects was associated with a later age at smoking onset and a shorter total smoking duration.

3.10. Examining adjusted relationships between MCEQ scores and smoking characteristics

3.10.1. The original five-factor model.

After adjusting for covariates, the pattern of findings remained the same as described above with two exceptions. The relationship between craving reduction and cigarettes smoked per day and the relationship between enjoyment of respiratory tract sensations and smoking duration no longer met the adjusted threshold for statistical significance (Table 4).

Table 4.

Adjusted relationships between the MCEQ subscales and smoking-related characteristics

	Original Five-Factor Model												New Four-Factor Model
	Age at Smoking Onset						Total Smoking Duration							Age at Smoking Onset						Total Smoking Duration
	B	SE	t	95% CI		ηp2	B	SE	t	95% CI		ηp2		B	SE	t	95% CI		ηp2	B	SE	t	95% CI		ηp2
Intercept	16.43	1.64	10.04	13.22	19.64	0.12***	−15.97	1.76	−9.08	−19.42	−12.51	0.10***	Intercept	17.73	1.65	10.74	14.49	20.97	0.13***	−17.10	1.78	−9.62	−20.59	−13.61	0.11***
Female Sex	1.47	0.51	2.88	0.47	2.47	0.01**	−1.16	0.55	−2.11	−2.23	−0.08	0.01	Female Sex	1.28	0.51	2.52	0.28	2.27	0.01*	−0.99	0.55	−1.81	−2.06	0.09	0.00
Race other than NH White	1.15	0.51	2.23	0.14	2.16	0.01*	−0.52	0.55	−0.93	−1.60	0.57	0.00	Race other than NH White	1.08	0.51	2.12	0.08	2.08	0.01*	−0.45	0.55	0.83	−1.53	0.62	0.00
Age	0.06	0.02	2.79	0.02	0.10	0.01**	0.77	0.02	33.92	0.72	0.81	0.60***	Age	0.05	0.02	2.30	0.01	0.09	0.01*	0.78	0.02	34.95	0.74	0.82	0.62***
Menthol Use	−1.60	0.51	−3.14	−2.60	−0.60	0.01**	2.85	0.55	5.21	1.78	3.92	0.04***	Menthol Use	−1.40	0.51	−2.75	−2.39	−0.40	0.01**	2.66	0.55	4.86	1.58	3.73	0.03***
Satisfaction	0.36	0.28	1.31	−0.18	0.90	0.00	−0.04	0.30	−0.12	−0.62	0.55	0.00	Stimulant Effects	0.49	0.25	1.98	0.00	0.97	0.01*	−0.40	0.26	−1.50	−0.92	0.12	0.00
Psychological Reward	−0.19	0.29	−0.66	−0.77	0.38	0.00	0.34	0.32	1.09	−0.28	0.96	0.00	Positive Rein forcement	0.57	0.23	2.54	0.13	1.01	0.01**	−0.51	0.24	−2.12	−0.99	−0.04	0.01*
Respiratory Tract	0.23	0.18	1.24	−0.13	0.59	0.00	−0.40	0.20	−2.04	−0.79	−0.02	0.01*	Negative Rein forcement	−1.42	0.27	−5.29	−1.95	−0.90	0.04***	1.62	0.29	5.59	1.05	2.19	0.04***
Craving Reduction	−0.72	0.20	−3.59	−1.11	−0.33	0.02***	0.79	0.22	3.67	0.37	1.21	0.02***	Aversion	0.53	0.15	3.64	0.25	0.82	0.02***	−0.88	0.16	−5.58	−1.19	−0.57	0.04***
Aversion	0.62	0.15	4.21	0.33	0.91	0.02***	−0.95	0.16	−6.03	−1.26	−0.64	0.05***
	Smoking Frequency (Days/Week)						Cigarettes Per Day							Smoking Frequency (Days/Week)						Cigarettes Per Day
	B	SE	t	95% CI		ηp2	B	SE	t	95% CI		ηp2		B	SE	t	95% CI		ηp2	B	SE	t	95% CI		ηp2
Intercept	6.31	0.18	34.32	5.95	6.67	0.61***	4.89	1.81	2.70	1.34	8.45	0.01**	Intercept	6.17	0.19	33.29	5.80	6.53	0.60***	4.24	1.84	2.30	0.63	7.86	0.01*
Female Sex	−0.12	0.06	−2.05	−0.23	−0.01	0.01*	−0.73	0.56	−1.29	−1.84	0.38	0.00	Female Sex	−0.10	0.06	−1.69	−0.21	0.02	0.00	−0.62	0.57	−1.10	−1.73	0.49	0.00
Race other than NH White	−0.37	0.06	−6.33	−0.48	−0.25	0.05***	−1.94	0.57	−3.40	−3.05	−0.82	0.02**	Race other than NH White	−0.36	0.06	−6.23	−0.47	−0.24	0.05***	−1.88	057	−3.30	−2.99	−0.76	0.01**
Age	0.00	0.00	1.08	0.00	0.01	0.00	0.10	0.02	4.11	0.05	0.14	0.02***	Age	0.00	0.00	1.79	0.00	0.01	0.00	0.11	0.02	4.57	0.06	0.15	0.03***
Menthol Use	0.06	0.06	1.04	−0.05	0.17	0.00	0.68	0.56	1.21	−0.43	1.79	0.02	Menthol Use	0.04	0.06	0.63	−0.08	0.15	0.00	0.57	0.57	1.00	−0.55	1.68	0.00
Satisfaction	0.03	0.03	0.87	−0.03	0.09	0.00	0.56	0.31	1.84	−0.04	1.16	0.00	Stimulant Effects	−0.07	0.03	−2.60	−0.13	−0.02	0.01**	−0.18	0.27	−0.67	−0.72	0.35	0.00
Psychological Reward	0.01	0.03	0.29	−0.06	0.07	0.00	0.37	0.32	1.13	−0.27	1.00	0.00	Positive Rein forcement	−0.01	0.03	−0.46	−0.06	0.04	0.00	−0.02	0.25	−0.09	−0.52	0.47	0.00
Respiratory Tract	−0.03	0.02	−1.26	−0.07	0.02	0.00	−0.42	0.20	−2.05	−0.81	−0.02	0.01*	Negative Rein forcement	0.19	0.03	6.16	0.13	0.25	0.05***	1.10	0.30	3.67	0.51	1.69	0.02***
Craving Reduction	0.09	0.02	4.01	0.05	0.14	0.02***	0.37	0.22	1.64	−0.07	0.80	0.00	Aversion	−0.07	0.02	−4.12	−0.10	−0.04	0.02***	−0.49	0.16	−3.00	−0.81	−0.17	0.01**
Aversion	−0.08	0.02	−4.67	−0.11	−0.05	0.03***	−0.51	0.16	−3.15	−0.83	−0.19	0.01**
	Dependence													Dependence
	B	SE	t	95% CI		ηp2								B	SE	t	95% CI		ηp2
Intercept	−0.68	0.29	−2.35	−1.25	−0.11	0.01*							Intercept	−0.69	0.29	−2.41	−1.26	−0.13	0.01*
Female Sex	−0.09	0.06	−1.61	−0.20	0.02	0.00							Female Sex	−0.07	0.06	−1.30	−0.18	0.04	0.00
Race other than NH White	−0.12	0.06	−2.11	−0.24	−0.01	0.01*							Race other than NH White	−0.12	0.06	−2.07	−0.23	−0.01	0.01*
Age	0.00	0.00	−1.00	−0.01	0.00	0.00							Age	0.00	0.00	−0.52	−0.01	0.00	0.00
Menthol Use	0.02	0.06	0.33	−0.09	0.13	0.00							Menthol Use	0.00	0.06	0.03	−0.11	0.11	0.00
Smoking Frequency	0.16	0.04	4.43	0.09	0.23	0.03*							Smoking Frequency	0.14	0.04	3.96	0.07	0.21	0.02*
Satisfaction	0.04	0.03	1.37	−0.02	0.10	0.00							Stimulant Effects	0.09	0.03	3.27	0.04	0.14	0.02***
Psychological Reward	0.25	0.03	7.74	0.19	0.31	0.07***							Positive Rein forcement	0.04	0.03	1.58	−0.01	0.09	0.00
Respiratory Tract	0.01	0.02	0.52	−0.03	0.05	0.00							Negative Rein forcement	0.29	0.03	9.57	0.23	0.35	0.11***
Craving Reduction	0.11	0.02	5.09	0.07	0.16	0.03***							Aversion	0.08	0.02	4.86	0.05	0.11	0.03***
Aversion	0.07	0.02	4.42	0.04	0.11	0.03***

Note.

^*p < .05

^**p < .01

^***p < .001

B = Beta; SE = Standard Error; CI = Confidence Interval; η_p² = Partial Eta Squared; NH = Non-Hispanic. Underlined values indicate differences from the unadjusted models.

3.10.2. The new four-factor model.

After adjusting for covariates, the pattern of findings remained the same as described for the unadjusted models except that the relationship between positive reinforcement and smoking duration no longer met the adjusted threshold for statistical significance (Table 4).

4.0. DISCUSSION

In the current study, we evaluated psychometric properties of the originally proposed five-factor scoring for the MCEQ (Cappelleri et al., 2007) and a novel, four-factor approach. We found the strongest psychometric support for the four-factor structure comprising Stimulant Effects, Positive Reinforcement, Negative Reinforcement, and Aversion. Importantly, while the four factors were data-derived, each mapped onto well-established constructs that have been shown to relate to cigarette smoking behavior in prior research. First, cigarettes are known to produce stimulant effects that often are experienced positively by users and drive continued use including wakefulness (e.g., Boutrel & Koob, 2004), improved attention and memory (e.g., Rezvani et al., 2001; Valentine & Sofuoglu, 2018), and appetite suppression (e.g., Audrain-McGovern & Benowitz, 2011). Second, there are many well-documented positively reinforcing effects of smoking including experiences related to pleasant taste, enjoyment, and physical sensations like throat-hit (e.g., Carter et al., 2009). Third, negatively reinforcing effects like craving relief and reduced irritability are integral to the cycle of nicotine withdrawal and continued smoking (Carter et al., 2009). Finally, experiencing aversive effects like nausea and dizziness may deter further cigarette use (Carter et al., 2009), although experiencing such symptoms during early smoking experiences can predict developing nicotine dependence in the future (e.g., DeFranza et al., 2004). Thus, the four-factors appear theoretically justified despite being data-derived.

With regard to specific psychometric properties, confirmatory factor analyses revealed that the four-factor solution uniquely evidenced good fit to the data from the primary analytic sample and the clinical trial data. In the primary analytic sample, internal consistency for each of the four multi-item subscales was adequate to excellent. Further, scalar measurement invariance uniquely was established for all subgroups tested, supporting the examination of between-groups differences for all subgroups of interest. Finally, the four-factor structure evidenced test-criterion validity with all smoking-related outcomes, with each subscale accounting for significant variance in at least one smoking-related characteristic. Thus, when considered in concert, the study findings suggest that a four-factor structure for scoring the MCEQ may be a viable alternative moving forward.

It is important to consider several study limitations when interpreting the findings. All participants in the primary analytic sample were market research panelists recruited via Qualtrics Online sample who self-reported current smoking, a history of smoking for at least one year, and use of e-cigarettes in a smoking cessation attempt in the past two years. Our reliance on online panels and the inability to confirm current cigarette smoking status, in particular, may limit the generalizability of our findings. However, this concern is mitigated to some extent by the fact that mediocre/poor model fit for the five-factor latent structure and good model fit for the four-factor latent structure was replicated in an independent sample of adults who were biochemically confirmed to smoke cigarettes. An additional limitation, the inclusion of two single-item subscales (i.e., enjoyment of respiratory tract sensations, craving reduction) in the five-factor version of the MCEQ may limit the ability to effectively evaluate the theoretical structure it is designed to assess. Specifically, it is not possible to evaluate internal reliability or to estimate model parameters like factor loadings for single-item subscales. Therefore, future research should consider testing additional items for the two single-item domains before judgments about the theoretical structure can be made. In addition, the good fit of the four-factor model to the data for both smoking and vaping may be linked to the fact that all participants (who had used both cigarettes and e-cigarettes within the past 2 years) completed both versions of the measure. However, this concern is reduced for three reasons. First, all participants completed the cigarette version of the MCEQ prior to the e-cigarette version, so carry-over effects could not impact the cigarette version. Second, model fit for the four- and five-factor latent structures for the MCEQ was tested in an independent sample that was very different from the primary analytic sample (i.e., adults who were confirmed to smoke cigarettes and who were seeking treatment for Alcohol Use Disorder;; O’Malley et al., 2018). Third, scalar measurement invariance was achieved when comparing adults who reported current exclusive use of cigarettes and those who reported using both cigarettes and e-cigarettes, suggesting that the measure performs comparably for assessing the subjective effects of cigarette smoking irrespective of whether an individual also uses e-cigarettes. That said, future research is needed to evaluate the fit of the five- and four-factor models in other samples (e.g., youth, those who exclusively smoke cigarettes with no prior e-cigarette experience). Finally, although the current sample included a substantial number of adults who reported currently using both cigarettes and e-cigarettes, the study design was not appropriate for conducting measurement invariance analyses of the cigarette and e-cigarette versions.A future study that is explicitly designed to evaluate invariance of the five-factor and four-factor versions by product type (i.e., cigarettes vs e-cigarettes) among adults who use both cigarettes and e-cigarettes is required. In this future work, it will be critical that the order of presentation of the cigarette and e-cigarette versions of the MCEQ is randomized so that order effects (impacting carryover) can be addressed.

In sum, the current study provided modest support for the originally proposed, five-factor structure, with stronger psychometric support observed for a novel, four-factor MCEQ structure. As such, conceptualizing and scoring the MCEQ as an 11-item, four-factor scale reflecting Stimulant, Positively Reinforcing, Negatively Reinforcing, and Aversive subjective effects of smoking ultimately may be comparable or even superior to conceptualizing the measure as a 12-item, five-factor scale reflecting Smoking Satisfaction, Psychological Reward, Craving Reduction, Enjoyment of Respiratory Tract Sensations, and Aversion. However, care must be taken when considering these two scoring approaches. Just as the five-factor model was less supported in our data, the four-factor structure proposed here may not be universally supported in all other samples. Until there is sufficient evidence that one factor solution is superior to the other, researchers are encouraged to examine the latent structure of all MCEQ data prior to scoring the measure and utilizing it in further analyses. Further, it is important to remember that even if one structure is determined to be optimal in the present day, the latent structure may need to be revisited over time as smoking behaviors and the population of people who smoke change. Finally, while we have now observed strong psychometric support for the four-factor versions of the MCEQ and the e-cigarette adapted version (i.e., the EMCEQ; Morean & Bold, 2022), respectively, until invariance analyses can be conducted properly, direct comparisons of cigarette and e-cigarette effects using the two versions of the measure are not yet supported.

Highlights.

• The MCEQ is widely used to assess subjective effects of cigarette smoking.

• The original five-factor structure has never been independently confirmed.

• Additional psychometrics (e.g., measurement invariance) had not been examined.

• Limited support for the original 5-factor structure was observed in this study.

• A 4-factor structure evidenced superior psychometric properties in this sample.

Data Availability

Data will be made available upon reasonable request to the corresponding author, Meghan Morean, PhD (meghan.morean@yale.edu).