Delivering an attractive smile is a key element in orthodontic patient satisfaction. Smile characteristics can be affected by the facial context. The purpose of this study was to investigate smile esthetics related to facial attractiveness and sex of the model.
Attractive, average, and unattractive model faces (2 of each; 3 male, 3 female) determined by peer ratings were combined with 10 smile variables (buccal corridor, smile arc, maxillary gingival discrepancy, gingival display, incisal-edge discrepancy, cant, overbite, central-incisor gingival margin discrepancy, and maxillary midline to face, and maxillary midline to mandibular midline). Each smile characteristic was altered digitally and presented with slider technology to allow a continuous range of choices. Raters chose the ideal and the limits of acceptability. The variables were divided into 6 separate surveys and rated 96 times. Reliability was assessed by answering each question twice.
Individual smile variable reliability ranged from fair to excellent, except for the buccal corridor. Clinically significant values were defined as greater than 1.0 mm with statistical significance ( P <0.05). Rater sex did not make a difference. Clinical significance was found for smile arc, gingival display, and maxillary midline to face. For females, accentuated smile arcs were preferred for the unattractive and attractive models compared with the average models. The opposite was found for male models. More gingival display was preferred for the attractive and unattractive male and female models compared with the average models. Attractive models were allowed less midline deviation.
Facial attractiveness and model sex impacted smile variables with a facial context, except for occlusal cant. These smile characteristics with a facial context should be considered when diagnosing and planning treatment for an orthodontic patient.
Facial and dental esthetics appear to play a role in the quality of life. Langlois et al believed that facial attractiveness was highly correlated with increased quality of life and interpersonal success. More attractive people were perceived as more intelligent, confident, and socially acceptable. Shaw and Shaw et al thought that facial attractiveness was the key to social success. They considered dental attractiveness an important factor contributing to facial and social attractiveness. Kiyak reviewed the effects of a pleasing dentition on the quality of life. Contrary to the beliefs of Shaw and Shaw et al, Kiyak thought that dental esthetics did not enhance social acceptance. She did, however, conclude that orthodontic treatment improved esthetics and the psychosocial well-being of patients. Adolescents who have completed orthodontic treatment reported fewer negative psychosocial influences, such as teasing and embarrassment from smiling. But long-term data do not substantiate that orthodontic treatment has a beneficial effect on psychological well-being or quality of life. In spite of the lack of resolution of the impact of the esthetics of the smile on a person, orthodontists’ focus on the smile and its components has grown in recent years.
Past investigations of smile characteristics have raised several issues. First is the method of presentation. Ker et al performed a computerized survey to identify the ideal and the range of acceptability for 10 smile characteristics from the layperson’s view using 1 image of the lower face. The computerized survey allowed the raters to digitally modify 1 variable at a time. The images rated appeared in a continuously morphed stream of choices. This was an improvement compared with previous studies because it eliminated the incremental nature of the choices and provided more precision.
A second issue is the facial perspective. Some previous studies used the circumoral or lower facial perspective for evaluation by the model raters; this might focus intensively on the smile. Other studies used a full-face perspective that could either reduce or enhance the raters’ ability to evaluate aspects of smile esthetics. Springer et al addressed this issue in part by using material and methods similar to those of Ker et al to study the lower face. Springer et al found statistical differences between the perspectives for several variables, but only 5 were clinically significant (ideal smile arc, ideal buccal corridor, maxillary midline to face, mandibular midline to maxillary midline, and occlusal cant). Of the 5 smile variables, only one was relative to the teeth (mandibular midline to maxillary midline); the rest had a facial context.
A third issue is the nationality or ethnicity of the model rater. Sharma et al compared the perspectives of 3 groups (white people and American-Asian Indians in the United States, and Indians residing in India) using study methods similar to those of Ker et al. They found differences between the 3 groups for certain smile variables (buccal corridor, smile arc, and upper midline). Fewer differences were found between Indians residing in the United States and those residing in India. McLeod et al compared perspective differences between residents of the United States (data from Ker et al ) and Canada for the following smile variables: buccal corridor, gingival display, occlusal cant, maxillary midline to face discrepancy, and central-lateral gingival discrepancy. They concluded that there were clinically significant differences for smile variables in all dimensions, except for buccal corridor. Canadians were more discriminating and allowed less deviation compared with US raters. Stockebrand et al distributed a dental esthetic questionnaire with different ethnic groups residing in Germany. Raters evaluated frontal smile photos of 8 male and female models, and the investigators reported differences between 3 ethnic groups (German, Russian, and Turkish). These studies indicate that nationality and ethnicity do influence dental esthetic perception.
The fourth issue is the attractiveness of the full face. Springer et al used average male and female faces for their study. Other previous full-face smile studies did not identify the attractiveness of the face. Flores-Mir et al found that the full-face context led to few differences, and almost all smile variables affected were measured relative to the face. It appeared that models did have an effect on the results. This could be due in whole or in part to model attractiveness. From the full-face perspective, does model sex affect the rating of smile variables? Second, does the level of attractiveness of the model affect our current perceptions of ideal and acceptable values of a smile variable?
The purpose of this study was to investigate the effects of varying facial attractiveness (unattractive, average, and attractive) and sex of the model on smile esthetics from laypersons’ perspectives.
Material and methods
This was an observational cross-sectional study approved by the institutional review board of Ohio State University in Columbus with data collection at a central campus facility. The first part of the study was identification of model faces (unattractive, average, or attractive) for each sex. The second part focused on interactions between model faces (unattractive, average, and attractive), model raters (sex), model sex, and selected smile characteristics.
Twenty male and 20 female young adult models were selected to participate in the first portion of this study. They were initially selected because they appeared upon screening to provide a large range of general facial attractiveness. All models consented to participate and signed a model release form, which allowed frontal posed smiling facial images to be obtained.
The frontal images were bisected and mirrored to eliminate any minor facial asymmetries by using Photoshop (CS2; Adobe Systems, San Jose, Calif) similar to the procedures used by Parekh et al. These images were rated for attractiveness with a visual analog scale from “very unattractive” (scale rating, 0) to “very attractive” (scale rating, 100). Model faces were presented in random order. Twenty percent of the model face images were repeated for reliability purposes.
One hundred raters evaluated the model faces. The raters were at least 18 years of age and nondental professionals. Average ratings for each of the 20 male and 20 female model faces were calculated. One male and 1 female face were selected from the central 20% of the visual analog scale’s average ratings and called average attractiveness. Attractive faces (1 male, 1 female) were those with the highest average ranks on the visual analog scale. The unattractive male face chosen was ranked lowest. The unattractive female face selected was ranked second to the lowest. The additional consideration in model choice was that the model face smile selected had to accommodate the dental template of each smile variable.
The 6 model faces (3 male, 3 female) had their smiles blocked out by using Adobe Photoshop CS2 in a procedure modified from Parekh et al. The smile characteristics evaluated in this study were similar to previous studies. Each smile characteristic was altered by using digitally modified templates separated from the initial image. This method eliminated any obvious incremental changes of dental variations and allowed a continuous gradient of changes.
The 10 smile variables assessed in this survey were identical to those examined by Springer et al, but with different model faces. These included buccal corridor (the amount of dark space between the inner corner of the mouth and the buccal surface of the molars), smile arc (the curvature formed by the incisal edges of the anterior teeth relative to the curvature of the lower lip ), central-lateral gingival margin discrepancy (the gingival margin difference between the maxillary central and lateral incisors), maxillary gingival display (the amount of gingival show between the zenith of the crown [central incisor] and the lower curtain of the upper lip), lateral incisal step incisal edge discrepancy (the difference between the incisal edges of the central and lateral incisors), overbite (the projection of the maxillary front teeth overlapping the mandibular teeth), occlusal cant (the transverse rotation of the occlusal plane), central-central incisor gingival margin discrepancy (the gingival margin difference between the maxillary central incisors), maxillary midline to face (the horizontal discrepancy between the central incisors and the philtrum of the upper lip), and maxillary to mandibular midline discrepancy (the horizontal discrepancy between the maxillary and mandibular midlines).
The model raters used 1 to 3 of the following directions per model face depending on the smile characteristic in question.
Please move the slider to the left to select the first image you find unattractive.
Please move the slider to the right to select the first image you find unattractive.
Select the image you find most ideal. Move the slider to the extremes to see the possible range of options, and then choose the most appropriate position.
The main challenges in this study were model rater reliability and fatigue. Model rater reliability was assessed by randomly repeating each question twice. Model rater fatigue was managed by dividing the smile characteristics into 6 surveys. Each model rater judged 1 or 2 smile characteristics for all model faces of varying attractiveness (unattractive, average, and attractive). Six of the 10 smile characteristics had 3 questions: smile at ideal, and upper and lower limits of acceptability. This generated 36 questions per smile characteristic, including model rater reliability. Four of the 10 smile characteristics had 1 question: deviation from the ideal. This generated 12 questions per smile characteristic, including model rater reliability. The surveys were divided so that 4 surveys contained 1 large and 1 small variable, and 2 surveys contained 1 large variable only. Each model rater answered fewer than 50 questions per survey.
The model raters were recruited by using a poster at a central campus facility. Volunteers who were interested approached the investigators for more information. The raters were not solicited. They were given a brief script on the purpose of this study. The participants were required to be conversant in English and able to control a computer mouse. Dental professionals and dental students were excluded. All model raters met the criteria and consented to participate by agreeing with the terms of the waiver of informed consent. Age, sex, ethnicity, and sociodemographic status were voluntarily provided. Upon completion of the survey, a $10 gift card was provided.
Five hundred seventy-six adult (18 years or older) model raters were included in this study. This sample size requirement was determined by power analysis. For an experiment-wise, nondirectional alpha risk of 0.05 and assuming a standard deviation of 3.5, a sample size of 87 model raters per smile characteristic yielded a power of 0.86. To accommodate a nonparametric analysis, 10% was added to this calculated sample size. The final sample size needed 96 raters per smile characteristic and model face. There were 6 surveys, and each survey had to be rated 96 times. Therefore, a total of 576 model raters were required to complete this study.
Reliability for all smile variables measurements was evaluated by using the Fleiss-Cohen weighted kappa statistic.
Medians and their corresponding ranges were calculated for ideal, and upper and lower limits of acceptability for all smile characteristics. Comparisons between model sex and among attractiveness levels were assessed by using multiple Wilcoxon matched-pairs signed rank tests with P values adjusted with the step-down Bonferroni method of Holm.
There were no statistically significant differences between the ratings of male and female model raters ( P >0.13). The Fleiss-Cohen weighted kappa statistic showed that reliability was fair to excellent when judging individual smile variables, except for buccal corridor ( Table I ). The buccal corridor variable was excluded because of poor reliability.
|Measurement||Kappa ∗||95% CI, lower||95% CI, upper||Interpretation †|
|Gingival display||0.84||0.83||0.86||Almost perfect|
|Maxillary midline to face deviation||0.60||0.53||0.67||Moderate|
|Maxillary to mandibular midline deviation||0.48||0.40||0.57||Moderate|
|Central-central gingival discrepancy||0.58||0.51||0.66||Moderate|
|Central-lateral gingival discrepancy||0.83||0.81||0.85||Almost perfect|
|Lateral incisal step||0.80||0.79||0.82||Substantial|
In this study, an alpha level less than 0.05 was considered statistically significant, and a difference of 1.0 mm or greater was considered clinically meaningful. Differences that were both statistically significant and clinically meaningful were designated as clinically significant .
There were clinically significant differences between the model sexes and among the levels of facial attractiveness. For all smile variables, there were no differences between the male and female models with average facial attractiveness.
Table II refers to the 5 smile variables that had no facial context—variables that could be judged only in the context of the dentition or surrounding soft tissues. Although some statistically significant differences appeared with maximum overbite, none was clinically meaningful. Model sex and model attractiveness did not have a clinically significant effect on these dental smile variables.
|Model sex||Attractiveness level|
|Median (mm)||Quartile range||Median (mm)||Quartile range||Median (mm)||Quartile range|
|Central-lateral gingival difference ideal ∗|
|Central-lateral gingival difference minimum|
|Central-lateral gingival difference maximum|
|Lateral incisal step ideal|
|Lateral incisal step maximum|
|Central-central gingival discrepancy|
|Maxillary to mandibular midline|
Table III refers to smile variables that had a facial context—variables that could only be judged relative to the face. Of these facial smile characteristics, only occlusal cant was not affected by model sex or model attractiveness. All others were affected by either model sex, attractiveness, or both. Attractiveness had more statistical and clinical significance than model sex.
|Model sex||Attractiveness level|
|Median (mm)||Quartile range||Median (mm)||Quartile range||Median (mm)||Quartile range|
|Smile arc ideal ¶|
|Female||−1.5 ∗ †||3||−2.5||1.5||−2 ∗||2.5|
|Male||−3 †||2||−2 §||2.5||−3||2|
|Smile arc minimum ¶|
|Smile arc maximum ¶|
|Female||2.5 ∗ †||1.5||1.5||2||1.75||2.5|
|Gingival display ideal ∗∗|
|Female||1 † ‡||1.5||2.4 §||1.0||−0.5||0.5|
|Male||1.5 ‡||0.75||2.25 §||1.5||0||1|
|Gingival display minimum ∗∗|
|Female||−0.5 † ‡||0.8||0.8 §||1.5||−1.5||1.0|
|Male||0.3 ‡||1.0||0.4 §||1.5||−1.5||1.5|
|Gingival display maximum ∗∗|
|Female||2.9 † ‡||1.0||5.0 §||1.7||0.5||1.0|
|Male||3.0 † ‡||1.0||4.5 §||1.7||1.0||1.8|
|Maxillary midline to face|
|Female||2.0 † ‡||1.3||3.0||1.7||3.4||1.1|
|Male||3.0 † ‡||2.3||3.6||1.7||3.8||2.3|
|Occlusal cant (°)|