Participants and procedures

Data were drawn from the ongoing 25-year Nicotine Dependence in Teens (NDIT) Study (n = 1294) [18]. NDIT was initiated in 1999–2000 with 1,294 Grade 7 students from 10 Montréal-area schools selected for linguistic, geographic, and socioeconomic diversity. In addition to its primary aim, NDIT has collected extensive sociodemographic, psychosocial, lifestyle, and health data.

For the present cross-sectional analysis, self-report data were collected online in cycle 25 in 2023 from 716 participants (Mage = 35.3, SD = 0.6). Data were accessed for research purposes on January 5, 2024. The study follows Transparency and Openness Promotion Guidelines to ensure integrity, reproducibility, and transparency. De-identified raw data, metadata (variable definitions, collection methods), and the analytic code are available on reasonable request, subject to ethical and legal constraints.

Ethics approval was obtained from the Montreal Department of Public Health, McGill University, and the Centre de recherche du centre hospitalier de l’Université de Montréal (CRCHUM) (2007–2384, 2017–6895, ND06.087). Written informed consent was provided by parents/guardians at study inception and by participants once they reached legal age. Detailed methods, data collection procedures, questionnaires, publications and data requests procedures [18] are available at https://www.celphie.ca/ndit.

Study variables

We examined eight body image-related indicators. Body-related emotions (shame, guilt, envy, embarrassment, authentic pride, hubristic pride) were measured using items from the Body and Appearance Self-Conscious Emotions Scale (BASES) [19], the Body-Related Embarrassment Scale (BREM) [20], and the Body-Related EnVy Scale (BREV) [21]. Items (e.g., “I feel ashamed of my appearance”) were rated on a 1 (never) to 5 (always) [22]. Higher scores for shame, guilt, envy, and embarrassment reflect negative emotional experiences, while higher scores for authentic and hubristic pride indicate positive emotions.

Internalized weight bias was assessed with the 3-item Weight Bias Internalization Scale short form (WBIS-3; α = 0.92) [14,23]. Items (e.g., “I dislike myself for my weight”) were rated from 1 (strongly disagree) to 7 (strongly agree); higher means indicate stronger internalized bias.

Worry about weight was measured with: “In the past 2 weeks, how often did you worry about your weight?” (1 = never to 5 = very often).

Weight commentary was assessed using two items: (i) “In the past 12 months, how often did people in your life make positive comments about your weight?” and (ii) “…negative comments…?” Response options (never, rarely, sometimes, often, always) were recoded as infrequent (never, rarely) or frequent (sometimes, often, always). Reliability and validity details are provided in Supplemental S1 File.

BMI (kg/m²) was derived from self-report height and weight, then categorized using WHO criteria [16]: underweight (<18.5), lower weight (18.5–24.9), higher weight (≥25.0). Participants categorized as underweight (n = 12) also experience weight commentary (i.e., often praise for thinness which can be linked to surveillance and disordered eating [4,12]), but were excluded due to the small sample. Analyses thus focused on lower weight vs. higher weight.

Covariates included baseline characteristics (age, sex, Canadian-born [yes/no], home language [French, English, other], mother was university-educated [yes/no]) and participant was university-educated (yes/no) in cycle 25.

Data analysis

Descriptive analyses examined distributions, identified missing data and outliers, and computed means and SDs for continuous variables, and proportions for categorical variables. To assess potential selection bias, we compared characteristics at NDIT inception between participants retained in cycle 25 and those lost-to-follow-up since inception. Following the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines [23], descriptive rather than statistical comparisons are presented.

Analyses proceeded in two steps: Within each sex, participants were grouped by weight status (lower weight vs. higher weight) and by frequency of positive or negative weight commentary (frequent vs. infrequent). For each subgroup, mean scores and standard deviations were calculated for body image emotions, internalized weight bias, and weight-related worry. Mean differences were estimated within weight strata (frequent minus infrequent), and “differences between mean differences” were calculated by subtracting the mean difference for the higher weight group from the corresponding value for the lower weight group. This descriptive “difference-in-differences” approach (which is presented in the Supplementary Material) provides unadjusted descriptive analyses of potential moderation and helps readers interpret the interaction (product term) estimates from subsequent multivariable linear regression models.

In the regression analyses, moderation by weight status was tested using product terms in linear regression models with Hayes’ PROCESS macro (Model 1, version 4.0) for SPSS [24]. Each model tested whether weight status modified the association between frequent vs. infrequent positive or negative commentary and a body image-related indicator. This hypothesis-by-hypothesis approach [25] yielded 16 models per sex (8 indicators × 2 commentary types (positive, negative)). All models adjusted for age and “participant was university-educated”. Bootstrapping with 5,000 resamples produced bias-corrected 95% CIs.

Participant characteristics

A total of 687 participants at cycle 25 were included in the analyses (392 females, 295 males). Twenty-nine participants who completed questionnaires but were missing data on weight commentary were excluded. Participants’ mean (SD) age was 35.3 (0.6) years; 57% were female, 94.1% were born in Canada, and 70.5% spoke English at home. Nearly half (46.5%) had mothers with a university education, 16.6% reported annual household income below $50,000 CAD, 17.9% lived alone, and 60.7% lived with children (excluding siblings). Most were employed (88.5%) and 61.8% reported having a university education.

Compared with those lost to follow-up since inception, retained participants showed few baseline differences, except that a higher proportion of retained males had mothers with a university education (Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Characteristics at NDIT inception of females and males retained and not retained for analysis in cycle 25 by sex, NDIT, 1999-2000 to 2023 (n = 1294).

https://doi.org/10.1371/journal.pone.0337951.t001

Females

Weight status and body image-related indicators.

Among females, 47% had lower weight and 53% had higher weight. Six of eight body image-related indicators reflected less favorable body image indicators among those with higher weight, including higher shame, guilt, envy, embarrassment, internalized weight bias, and worry about weight. Authentic and hubristic pride scores were lower in this group, also indicating more negative body-related emotional states (Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Mean (SD) levels of body image-related indicators in participants with lower weight or higher weight by sex, NDIT, 2023 (n = 716)*.

https://doi.org/10.1371/journal.pone.0337951.t002

Positive weight commentary and body image-related indicators.

Overall, 44% of females reported frequent positive weight commentary (47% of those with lower weight and 41% of those with higher weight).

S1 Table presents unadjusted descriptive analyses of potential moderation. For positive commentary, difference-in-differences were generally small, with the possible exceptions of guilt and internalized weight. However, adjusted linear regression models (Table 3) provided formal tests of moderation and showed significant moderation by weight status for shame, guilt, embarrassment, and internalized weight bias. This indicates that positive weight commentary was more strongly associated with these indicators among females with higher weight than among those with lower weight, despite relatively modest crude differences.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Estimated beta coefficients and 95% confidence intervals for weight status x positive weight commentary product terms in the relationship between frequent positive weight commentary and body image-related indicators in females, NDIT, 2023 (n = 392)*.

https://doi.org/10.1371/journal.pone.0337951.t003

Negative weight commentary and body image-related indicators.

Thirteen percent of females reported frequent negative weight commentary (10% among those with lower weight and 16% among those with higher weight). The differences-in-differences (S1 Table) were larger than those observed for positive commentary, particularly for shame, embarrassment, internalized weight bias, and worry about weight, suggesting stronger adverse associations in females with higher weight.

Adjusted linear regression models testing weight status × negative commentary product terms (Table 4) supported these descriptive findings. Significant interactions for shame, embarrassment, internalized weight bias, and worry about weight indicated that negative commentary was more strongly associated with adverse body image-related indicators among females with higher weight than among those with lower weight.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Estimated beta coefficients and 95% confidence intervals (CI’s) for weight status X negative weight commentary product terms in the relationship between frequent negative weight commentary and body image-related indicators in females, NDIT, 2023 (n = 392)*.

https://doi.org/10.1371/journal.pone.0337951.t004

Taken together, these results suggest that while positive commentary was modestly more beneficial for females with higher weight, negative commentary had markedly stronger adverse links in this group, highlighting a possible disproportionate impact of weight-related remarks on body image-related indicators among females with higher weight.

Males

Interpretation of findings

Cultural context may help explain these differences. Women, particularly those with higher weight, are frequently subjected to objectification and scrutiny [6]. According to Objectification Theory, appearance-based evaluations heighten body surveillance and self-conscious emotions such as shame and anxiety, mechanisms that contribute to body dissatisfaction [26,27]. This may explain greater reactivity among women with higher weight. Negative weight commentary may elicit stronger effects because it reinforces societal stigma and triggers self-conscious emotions tied to appearance surveillance, whereas positive commentary may be perceived as less credible or fail to counteract entrenched negative beliefs about body weight. Thus, negative remarks may more powerfully activate internalized weight bias and self-evaluative processes.

Although the role of positive commentary is debated [3,28], our findings suggest that such remarks may be interpreted as more supportive by women with higher weight. Several explanations are possible. First, because women with higher weight experience more stigma, weight-related comments that are perceived as complimentary may provide temporary affirmation and reduce self-conscious emotions [12]. Second, when interpreted as autonomy-supportive recognition of health efforts rather than appearance judgments, positive comments may promote authentic pride (although we did not find an effect on authentic pride) [10]. Third, a “headroom” effect may operate: women with higher weight started from less favorable emotional profiles, leaving more room for positive commentary to correspond with improvement. However, while possibly offering short-term emotional relief, caution should be exercised in viewing these associations as an endorsement of weight-focused praise. Positive commentary may be interpreted as superficial, and women with higher weight may be less responsive to appearance-based validation due to longstanding experiences of objectification, internalized weight bias, and societal stigma. In appearance-valuing cultures, such remarks may also reinforce links between worth and body size and encourage body surveillance, with effects varying by context and intent [4,6].

Although body-image concerns were higher among males with higher weight, commentary varied little by weight status. On 1–5 scales, means clustered around 2–3, with higher weight linked to modestly higher negative scores (+0.2 to +0.9) and little difference in pride. Frequent negative commentary (~12% of men) was linked to poorer body-image across weight groups, suggesting a uniform effect. Possible explanations include norms emphasizing muscularity over weight [2,17], BMI misclassification of muscular men [16], weight-centric rather than muscularity-focused items [17], and limited power due to few men reporting frequent commentary.

Commentary context and source

The meaning of commentary likely depends on context and source, which were not captured here. Prior work shows that teasing from family or partners strongly predicts body-image disturbance [1,2,22], while comments from health professionals are often experienced as stigmatizing [12,15]. Even compliments, although correlated with lower dissatisfaction [3,5], may reinforce appearance-contingent worth [4,28]. Future studies should explicitly examine source, intent, and context using designs such as ecological momentary assessment [22], vignette-based experiments, or dyadic approaches.

Age and developmental context

Most prior research has focused on adolescents and emerging adults [3,5,22], whereas participants in this study were in their mid-30s. Commentary meanings likely evolve with age: in adolescence, remarks often reflect peer or family dynamics [1,2,22], whereas in adulthood they may arise in intimate, workplace, or healthcare settings [12]. Thus, these findings are most applicable to adults in their 30s and may not generalize to younger populations [18].

Public health and practical implications

Because these are cross-sectional data from a single cohort, we refrain from practice recommendations. Synthesizing our results with prior work [12,15,28], two provisional points emerge: negative commentary consistently aligns with poorer body-image, suggesting caution with critical or teasing remarks; and when weight discussions are unavoidable, weight-neutral, autonomy-supportive language focused on behaviors and well-being may reduce appearance-based evaluation [4,28]. Responses may also vary by ethnicity, socioeconomic background, and sexual orientation, highlighting the need for replication in more diverse samples.

Strengths and limitations

Strengths include validated measures of both emotional and cognitive aspects of body image [19,21,24], sex-stratified analyses, and explicit tests of moderation. The large, community-based sample enhances generalizability to young adults.

Limitations include the cross-sectional design, which precludes causal inference and raises the possibility that individuals with negative body image recall commentary more readily. Self-report introduces recall and social desirability bias, and those exposed to both positive and negative remarks may have been misclassified. Reaction to weight commentary likely varies by age, ethnicity, socioeconomic position, and sexual orientation, but our largely Canadian-born, English-speaking sample now in their mid-30s may limit generalizability [18]. Attrition and missing BMI data may also introduce bias and limit generalizability. BMI does not capture body composition, possibly misclassifying muscular participants [16]. About 11% of women were pregnant, which may have influenced BMI classification and perceptions. For men, few reported frequent commentary, limiting statistical power. Finally, multiple tests increase the risk of Type I error [29]; findings near significance should be interpreted cautiously.