This study aimed to compare the 2 most commonly used methods for assessing facial pleasantness, the Likert scale and the visual analog scale (VAS).
For assessing facial pleasantness, an album was assembled containing the front and profile view facial photographs of 10 patients (5 females, 5 males) who sought orthodontic treatment representing each of the different sagittal and vertical facial discrepancies (straight profile, convex profile, concave profile, long faces, and short faces). The facial pleasantness of the patients was judged by 3 groups of 90 evaluators (47 females and 43 males): 30 orthodontists (mean age, 32.41 years), 30 oral maxillofacial surgeons (mean age, 36.83 years), and 30 laypeople (mean age, 36.83 years). The evaluators judged the facial pleasantness using VAS and a 5-point Likert scale. At the end of the evaluation, the evaluators completed a questionnaire about their preferred scale. Analysis of variance and Tukey and Kruskal-Wallis tests were used to compare the data from VAS and Likert scale, respectively. The Spearman correlation coefficient was used to estimate the correlation between the 2 scales. Fisher transformation and z-statistic were used to estimate the correlation among the evaluator groups. In all tests, a 5% statistical significance level was adopted.
Both scales presented similar answers; only VAS values for the scores of pleasant and very pleasant according to the Likert scale could not be distinct. Most evaluators preferred the Likert scale as they found it easier to convey their opinion than VAS.
Although both scales can be used to assess facial pleasantness, the Likert scale seems more appropriate as the evaluators preferred it because of its simplicity.
Facial pleasantness was assessed with Likert and visual analog scales.
Oral surgeons, orthodontists, and laypeople evaluated sets of photographs.
Results were similar regardless of the scale used.
Most evaluators preferred the Likert scale because of its simplicity.
The primary goal of orthodontic treatment is to allow the patient to achieve a functional occlusion along with a harmonious face, thus achieving an adequate level of esthetics. In contemporary orthodontics, improving facial appearance is an important goal during patient planning because it is closely related to patient satisfaction after the treatment. In fact, esthetic improvement is one of the major reasons for searching for orthodontic treatment.
Beauty, perceived through physical appearance and facial esthetics, has always received great emphasis in human social interactions. Although the concept of beauty is not unique, it is extremely subjective and can be influenced by the socioeconomic and cultural context in which a patient is present. Their opinions are influenced by personal experiences and educational backgrounds, which form a plethora of knowledge and preferences for an orthodontist to consider while planning the treatment.
Because of this subjectivity, several studies sought to evaluate the perception of the laypeople, patients, and dental professionals regarding the esthetic patterns of the face and smile. What laypeople find attractive may not necessarily coincide with the opinion of a patient with malocclusion or with that of professional specialists in the field of dentistry. This perception of esthetics guides the professionals to propose treatment options to patients within their therapeutic possibilities.
In this context, facial pleasantness and its relationship with the quality of life and social interactions of patients have been the subject of much research. In these studies, 2 methods stand out for subjectively judging the facial esthetics: the visual analog scale (VAS) and the Likert scale.
The Likert scale is widely used in health studies and comprises 3-7 responses with different degrees of agreement and disagreement. , The VAS, a scale widely used in studies that quantify subjective experiences such as pain, consists of a 10-cm line, on which the extreme points are either opposite statements or minimum and maximum dimensions. The evaluators mark on this line according to the value of the response and later quantify it using a ruler.
Research comparing both scales in various areas such as physiotherapy, speech therapy, dentistry, psychology, nursing, and medicine showed different results, equivalent to the type of research addressed. When investigating the pros and cons of each scale, easiness in applying and interpreting have been in favor of the Likert scale. In contrast, VAS is considered more accurate for detecting small significant changes and more sensitive because of refined gradations of levels of response. However, participants seem to experience difficulties when filling out VAS, preferring to grade their responses closer to the edge of the line.
It is evident there is a lack of consensus on which method is the most appropriate, which is also a problem when facial pleasantness is considered. Therefore, this study aimed to compare 2 of the most common methods for judging facial pleasantness (VAS and Likert scales) from the perspective of orthodontists, oral maxillofacial surgeons, and laypeople. In addition, the ease of use and simplicity of each scale were also addressed.
Material and methods
This research was approved by the Research Ethics Committee of Research Ethics Committee of Sagrado Coração University (no. 3,222,772).
An album was made containing photographs of 10 adult patients (5 of each gender) who sought orthodontic treatment for different sagittal and vertical facial discrepancies. Each patient had 2 front photographs of the face, at rest and smiling, and another profile view photograph. To form the photograph album, 1 representative of each sex was selected with sa gittal (straight, convex, or concave face ) and vertical (long or short face) facial discrepancies. Subjective facial analysis was used to define the characteristics of the 5 facial types assessed in this study ( Figs 1-5 ). The following inclusion criteria were adopted for this selection: White patients; aged 18-30 years with no major facial asymmetries; and no piercing, beard, makeup, tattoos, or other accessories visible on facial photographs that might compromise the assessment. After the patients were selected, they were informed about the objectives of the research, and those who agreed to participate signed the informed consent form.
A power calculation was performed to determine the sample size of the evaluators. Using the significance level threshold of 5% and study power of 80%, demonstrating a minimum correlation of 0.5 would have required 29 evaluators in each group. It was decided that it would be appropriate to have 1 group of orthodontists and another of oral maxillofacial surgeons because these are the specialties that routinely perform a facial analysis to diagnose and treat facial complaints. In addition, a group of laypeople with no knowledge of dentistry was included. In this way, it would be possible to compare their opinions and observe whether they agree with regard to facial pleasantness.
The cohort of 90 evaluators (mean age, 35.35 years; 47 females and 43 males) consisted of 3 groups of 30 evaluators. These groups consisted of laypeople (mean age, 36.83 years; range, 20-67 years; 22 females and 8 males), orthodontists (mean age, 32.41 years; range, 27-45 years; 17 females and 13 males), and oral and maxillofacial surgeons (mean age, 36.83 years; range, 26-51 years; 7 females and 23 males). At the time of data collection, evaluators had at least 5 years of clinical experience after completion of their postgraduate training. Each evaluator made a judgment about facial pleasantness on the basis of patient photographs in the album evaluated at 2 different times 15-30 days apart; alternating assessment method each time (ie, 5-point Likert scale or VAS).
For VAS, the evaluator was instructed to mark their opinion with a vertical line anywhere along a 100-mm line segment, assigning a subjective grade to the images. A mark closer to the right side of the scale meant a more pleasant face and less pleasant if the mark was closer to the left side of the scale. Once finalized, the marked value would be measured in millimeters with a ruler (from left to right) to quantify the subjective evaluation. On the 5-point Likert scale, the examiner had 5 response options (1, very unpleasant; 2, unpleasant; 3, acceptable; 4, pleasant; and 5, very pleasant), and the option that best suited the evaluator’s opinion was marked with an “X.” To eliminate possible biases, half of the examiners in each group initially received the questionnaire with the Likert scale, whereas the other half received the questionnaire with VAS. At the end of the second visit, the evaluators completed a 3-question questionnaire, in which they were asked about the scale that they preferred considering the simplicity and ease of use.
In addition, 30% of the evaluators (10 per group) were randomly selected to repeat the assessment on each scale 30 days later to check for intraexaminer errors.
The data were described using mean, standard deviation, median, first quartile, third quartile, semiquartile range, and minimum-maximum values and were reported in the form of tables and graphs.
VAS measurements by the evaluator type passed the Kolmogorov-Smirnov normality test, whereas VAS measurements, when evaluated by the Likert scale rating, did not pass the normality criterion.
Intraexaminer errors analysis was performed using the weighted-kappa test statistic for the Likert scale and the intraclass correlation coefficient for VAS.
One-way analysis of variance and Tukey post-hoc test were used to compare VAS values among the 3 evaluator groups, whereas the Kruskal-Wallis test was used for comparing the Likert scale values. The Kruskal-Wallis test was used to compare the 5 categories of the Likert scale with the scores given with VAS. The Spearman correlation coefficient was used to determine the correlation between the 2 scales. Fisher transformation and z-statistics were used to compare the correlation coefficients among the evaluator groups. For the comparison of evaluator gender, t tests were used for VAS, and Mann-Whitney U tests were used for the Likert scale.
A 5% significance level was adopted for all tests. All statistical procedures were carried out using SPSS software (version 25.0; IBM Corp, Armonk, NY).
The agreement values for intraexaminer error were 0.70 and 0.69 for the Likert scale and VAS, respectively .
|Likert scale||Mean||Median||First quartile||Third quartile||Minimum||Maximum||P value|
|Very unpleasant a||16.66||11.00||5.00||26.00||0||78||<0.001 ∗|
|Very pleasant d||69.71||70.00||62.50||81.00||30||99|
Table I and Figure 6 show the comparison of facial pleasantness scores using VAS for each grade on the Likert scale, which identifies if there is an equivalence of scores given by the evaluators on different scales. It was observed that when the score on the Likert scale increased, the value for VAS also increased; however, there was no significant difference in VAS values between the Likert scores 4 and 5 (ie, pleasant and very pleasant). In addition, the mean and median values of VAS scores were within the equivalent range on the Likert scale in all the assessments except for the very pleasant score, in which VAS scores were lower than the comparable Likert scores.