ORIGINAL ARTICLE

Evaluation of facial attractiveness from

end-of-treatment facial photographs
Roxanne Shafiee,a Edward L. Korn,b Helmer Pearson,c Robert L. Boyd,d and Sheldon Baumrinde
San Francisco, Calif, Rockville, Md, and Newark, NJ

Introduction: Orthodontists typically make judgments of facial attractiveness by examining groupings of

profile, full-face, and smiling photographs considered together as a “triplet.” The primary objective of this
study was to determine the relative contributions of the 3 photographs— each considered separately—to the
overall judgment a clinician forms by examining the combination of the 3. Methods: End-of-treatment triplet
orthodontic photographs of 45 randomly selected orthodontic patients were duplicated. Copies of the profile,
full-face, and smiling images were generated, and the images were separated and then pooled by image type
for all subjects. Ten judges ranked the 45 photographs of each image type for facial attractiveness in groups
of 9 to 12, from “most attractive” to “least attractive.” Each judge also ranked the triplet groupings for the
same 45 subjects. The mean attractiveness rankings for each type of photograph were then correlated with
the mean rankings of each other and the triplets. Results: The rankings of the 3 image types correlated highly
with each other and the rankings of the triplets (P !.0001). The rankings of the smiling photographs were
most predictive of the rankings of the triplets (r " 0.93); those of the profile photographs were the least
predictive (r " 0.76). The difference between these correlations was highly statistically significant (P " .0003).
It was also possible to test the extent to which the judges’ rankings were influenced by sex, original Angle
classification, and extraction status of each patient. No statistically significant preferences were found for sex
or Angle classification, and only 1 marginally significant preference was found for extraction pattern.
Conclusions: Clinician judges demonstrated a high level of agreement in ranking the facial attractiveness of
profile, full-face, and smiling photographs of a group of orthodontically treated patients whose actual differences
in physical dimensions were relatively small. The judges’ rankings of the smiling photographs were significantly
better predictors of their rankings of the triplet of each patient than were their rankings of the profile photographs.
(Am J Orthod Dentofacial Orthop 2008;133:500-8)

F
acial attractiveness has interested artists, philos- have been relatively few studies of how orthodontists
ophers, and psychologists at least since the record and measure the relative attractiveness of their
heyday of Greek civilization.1-6 What makes patients.
faces attractive was profoundly interesting to orthodon- Many cues that orthodontic clinicians use to eval-
tists in Angle’s time7-9 and has since assumed an even uate the facial attractiveness of patients during actual
more important role in our specialty.10-12 Yet there office visits are transient. These include motion, ani-
mation, and facial expression. The importance of infor-
a
Assistant professor, Department of Restorative Dentistry, School of Dentistry,
University of the Pacific, San Francisco, Calif.
mation of these kinds has been recognized, and some
b
Mathematical Statistician, Biometric Research Branch, National Cancer Insti- attempts have been made to capture these data.13-15 But
tute, NIH, Rockville, Md.
c
thus far, such information seems to be too difficult for
Professor and clinic director, Department of Orthodontics, New Jersey Dental
School, University of Medicine and Dentistry of New Jersey, Newark, NJ.
orthodontists to capture and evaluate routinely. Instead,
d
Professor and chair, Department of Orthodontics, School of Dentistry, the predominant convention in orthodontic treatment
University of the Pacific, San Francisco, Calif. for some years has been to examine semistandardized
e
Professor, Department of Orthodontics, and director, Craniofacial Research
Instrumentation Laboratory, School of Dentistry, University of the Pacific, San static representations of the face, which usually consist
Francisco, Calif. of sets of 2-dimensional facial photographs, each typ-
Supported by NIH-NIDR grants DE07332 and DE08713, the American
ically including profile, full-face, and smiling images.
Association of Orthodontists Foundation, and the University of California
Orthodontic Alumni. In this article, these sets of 3 images are referred to
Reprint requests to: Sheldon Baumrind, Craniofacial Research Instrumentation collectively as “triplets.”
Laboratory, (CRIL) Room 617, Arthur A. Dugoni School of Dentistry,
University of the Pacific, 2115 Webster St, San Francisco, CA, 94115; e-mail,
The use of photographic triplets has developed by
sbaumrind@pacific.edu. consensus over a number of years. However, little
Submitted, February 2006; revised and accepted, April 2006. rigorous investigation has been conducted concerning
0889-5406/$34.00
Copyright © 2008 by the American Association of Orthodontists. the contribution that each of the 3 different kinds of
doi:10.1016/j.ajodo.2006.04.048 image makes to the overall evaluation of facial attrac-
500
American Journal of Orthodontics and Dentofacial Orthopedics
Volume 133, Number 4
Fig 1. Representative triplet: profile, full-face, and smiling photographs.
tiveness when the triplet is examined as a whole. More
recently, it was proposed that additional photographs
including at least a three-quarters view should be added
to the standard record set.16,17 Instrumentation is now
becoming available for the generation of fully 3-dimen-
sional digital images of the face that are viewable,
maneuverable, and measurable on a computer moni-
tor.18 Here again, however, no rigorous tests of the
clinical usefulness of the additional images are avail-
able.
This study was directed toward increasing our
understanding of the use of the currently standard facial
photographic triplets in clinical practice. The intent was
to examine and measure the contribution of each
photographic view to the overall evaluation of facial
attractiveness when clinicians examine the combined
group of 3. We were interested primarily in evaluations
made by clinicians with training and experience in
dentistry rather than those of laypeople. We believe that
this investigation should be of special interest to orth-
odontists, particularly because it has been asserted that
members of our specialty focus more heavily on views
of the face from the lateral aspect than do most other
viewers.19-21
MATERIAL AND METHODS
The general plan of this study was to collect data
from a group of judges with training in dentistry, and
each would independently evaluate a sample of stan-
dard photographic records generated during routine
orthodontic treatment. A representative set of 3 images
is shown in Figure 1.
As part of an ongoing study of the usual course and
outcome of orthodontic treatment by experienced cli-
nicians, a random sample of 135 treated patients with
complete clinical records was identified in the practice
of an experienced clinician (H.P.). Maintaining random
order (Fig 2), a random subset of 48 patients containing
equal numbers of Class I and Class II extraction and
Shafiee et al 501
nonextraction subjects was selected. For each of these
48 subjects, the written treatment records and all
pretreatment and end-of-treatment physical records
were duplicated electronically and used for further
study.
The end-of-treatment photographs of these 48 pa-
tients are the material for this study. The profile,
full-face, and smiling photographs for each patient
were reproduced individually and assembled into sep-
arate categories by image type (Fig 3). The 48 images
of each type were arranged in 4 groups of 12 subjects
each. Each group of 12 contained 3 subjects originally
classified as Angle Class I nonexaction, 3 as Angle
Class I extraction, 3 as Angle Class II nonextraction,
and 3 as Angle Class II extraction. However, before the
images were presented to the judges, 3 subjects were
dropped from a group of 12 because of poor quality of
at least 1 image. The final sample included images of 45
subjects, arranged in 3 groups of 12 and 1 group of 9.
Ten judges participated in the evaluation of the
images. All were dental students at the Arthur A.
Dugoni School of Dentistry at the University of the
Pacific, Stockton, Calif, who had satisfactorily com-
pleted the predoctoral didactic program in orthodontics.
At the start of the study, each judge had had approxi-
mately 1 year of clinical experience. The instructions to
the judges before they examined the 45 images of each
type were as follows.
Photographs of four groups of patients will be pre-
sented to you sequentially. The task is to sort each
group in the order of decreasing “Facial Attractive-
ness.” We solicit your personal opinion; there is no
“right answer.” The time required to sort each group
should not exceed 4 or 5 minutes. Be aware that when
two or more subjects appear equally matched, small
differences in their order will not affect the experi-
mental outcome.
Each judge, working independently, ranked all 4
502 Shafiee et al
Fig 2. Selection of the original sample: sequence of steps.
groups of photographs of the same image type sequen-
tially at the same session in order of facial attractive-
ness, which was defined individually by each judge
according to his or her own concepts. The judges
evaluated the profile images first, the full-face images a
minimum of 1 week later, and the smiling photographs
at least 1 week after that. Then, at least 2 weeks later,
each judge ranked the attractiveness of the triplet
images, each of which contained all 3 of the previously
examined images of each patient mounted together.
In each image type, the judges ranked the 3 groups
containing 12 subjects each from “most attractive” to
“least attractive.” Later, for purposes of computation, a
numerical rank was assigned to each image on a
descending scale from 12 for “most attractive” to 1 for
“least attractive.” For the 1 group with only 9 subjects,
the analogous judge rankings from 9 to 1 were each
multiplied by 1.3 to yield values ranging from 11.7 to
1.3. These transformed values for this group had the
American Journal of Orthodontics and Dentofacial Orthopedics
April 2008
same mean (6.5) and approximately the same standard
deviation as each of the other 3 groups, facilitating the
merging of data from all 4 groups for each image type.
For each image of each type, the rankings of the 10
judges were averaged. These averages make it possible
to compare the rankings for facial attractiveness of the
different image types across patients. Additional avail-
able information for each patient included data on sex,
extraction status, and original Angle class.
For the statistical analysis of this study, we used
Pearson correlations, Student t tests, and stepwise regres-
sions, all performed with the SAS statistical package
(version 9.1; SAS Institute, Cary, NC). The statistical
significance of comparisons between correlations was
calculated by using the method of Meng et al.22
RESULTS
The key question asked in this study was how much
the rankings of each image type separately contributed
American Journal of Orthodontics and Dentofacial Orthopedics
Volume 133, Number 4
Fig 3. Preparation of the photographic sample: sequence of steps.
on average to the ranking of the triplet. Answers to this
question are illustrated graphically in Figure 4. In each
scatter plot, the average ranking of the 10 judges for a
patient’s image of any one type (profile, full-face, or
smiling) is correlated with the average ranking of the
triplet of the same patient by the same judges.
The relationships between the average triplet rankings
and the average rankings of each image type are further
Shafiee et al 503
explored in Table I, which shows the correlations of the
triplets with each of the 3 image types considered sepa-
rately.
Table I shows that the ranking of each image type
by itself predicted in a highly significant way the
judges’ rankings of the triplet images. The strongest
predictors were the smiling images (r " 0.93), with the
full-face images the next strongest (r " 0.86), and the
504 Shafiee et al American Journal of Orthodontics and Dentofacial Orthopedics
April 2008

Fig 4. Scatterplots comparing the associations between pairs of image types. For each scatterplot,
n " 45 and P !.0001. A, Profile images vs triplets (r " 0.76); B, full-face images vs triplets (r " 0.86);
C, smiling images vs triplets (r " 0.93).

Table I. Pearson correlations (r) and R2 values Table II. Regression analysis
Profile Full-face Smiling P value to
Variable entered
Pearson r Step entered Partial R2 Cumulative R2 variable
Full-face 0.68
Smiling 0.68 0.75 1 Smiling 0.8524 0.8524 !.0001
Triplet 0.76 0.86 0.93 2 Full-face $ 0.0618 0.9142 !.0001
3 Profile $ 0.0109 0.9252 0.0188
R2 values
Triplet 0.58 0.74 0.86 Results of a stepwise regression to assess the relative contributions of
the rankings of each of the 3 individual image types to the ranking of
Measures of association between the mean values of the different the triplet images.
types of image for the same patient.

profile images the weakest (r " 0.76). (One might ask
whether any of the 3 individual image types was a
significantly stronger predictor than the others. Further
analysis showed that the difference between the corre-
lations of the smiling images with the triplet images and
of the profile images with the triplet images was highly
statistically significant (P " .0003). The analogous
difference between the correlations of the profile and
full-face images with the triplets had a probability of P
" .07, and the difference between the correlations of
the full-face and smiling images with the triplets had a
significance value of P " .054.)
The R2 values (the squares of the r values in Table
I) tell us that, when the image types were considered
separately, the ranking of the smiling images accounted
for 86% of the variability in the ranking of the triplet
images. Similarly, when they were considered sepa-
rately, the full-face image rankings accounted for 74%,
and the profile image rankings accounted for only 58% of
the variability in ranking the triplets. To understand how
the 3 views complemented each other when considered as
a triplet, a stepwise regression analysis was performed
with the following model.
[triplet rank] ! " # $L ! [profile rank]
# $F ! [full-face rank]
# $S ! [smiling rank]
# error
The results of this analysis are shown in Table II.
In this stepwise regression, the smiling variable,
entered first, accounted for an R2 of 0.85. The full-face
variable, entered next, adding 0.06, and the profile
variable, entered last, adding a small but still statisti-
cally significant increment of 0.01, yielding a total R2
of 0.93. On the basis of this analysis, we can assert that,
in this sample, 93% of the variability in the mean triplet
rankings could be accounted for by the profile, full-face,
and smiling rankings taken together—91% of which
could be accounted for by the smiling and full-face images
together, without including the profile images.
We next asked, for each kind of image, how much the
rankings of each judge correlated with the average rank-
ings of all judges. For the profile images, the average
correlation of each judge with the mean of all judges was
0.79 # 0.07 (mean # SD). The highest correlation of any
American Journal of Orthodontics and Dentofacial Orthopedics
Volume 133, Number 4
Fig 5. Scatterplot comparing profile images vs smiling
images (n " 45, r " 0.66, P !.0001). Even though the
correlations were highly significant, there were outliers,
some of which were analyzed further.
judge with the mean of the 10 judges was 0.87, and the
lowest was 0.63. For the full-face images, the average
correlation was 0.75 # 0.11; the highest correlation with
the mean was 0.86, and the lowest was 0.49. For the
smiling images, the average correlation was 0.78 # 0.06;
the highest was 0.87, and the lowest was 0.70. For the
triplets, the average correlation was 0.75 # 0.08; the
highest was 0.85. and the lowest was 0.55.
In general, patients whose profile photographs were
judged to be attractive tended to have smiling and
full-face images that were also considered attractive.
For example, the rankings for the profile images corre-
lated strongly with those of the smiling images. This
relationship is shown in Figure 5, a scatterplot of the
correlation between those variables for all 45 subjects.
It can also be seen in the plot that, although the
association between the 2 sets of values was highly
statistically significant and fairly strong, the values for
some subjects departed distinctly from the general rule.
For example, Case 603 had an average judge score of
9.8 for the smiling photo but only 2.6 for the profile
photo. On the other hand, Case 681 had an average
judge score of 8.4 on the profile photo but only 3.4 for
the smiling photo. (Ranking the images in each group
from 12 to 1 automatically resulted in an average value
of 6.5 for each type of image.)
To test the correspondence of these rankings with
the judgments of other clinicians who had not partici-
pated in the original study, a small additional empirical
test was conducted. The triplet images for the 4
outlying subjects were presented for separate review to
16 advanced orthodontic residents and 10 members of
the clinical faculty at the University of the Pacific with
the following instruction.
End-of-treatment Facial Photos for four orthodontic
Shafiee et al 505
patients are displayed on this table. These sets of
photos have been drawn from a larger sample of
forty-five that were ranked for “facial attractiveness”
by ten judges with dental backgrounds. Each set of
photos includes a lateral, a frontal, and a smiling
view. In general, the lateral photos and the smiling
photos for the same case tended to be ranked simi-
larly. But in these four cases, the rankings of the
lateral photos tended to be quite different from the
rankings of the smiling photographs. Please examine
the photos for each patient carefully. Then, within
each case, please indicate whether the smiling image
or the lateral image is the more attractive.
The triplet images for these 4 subjects are shown in
Figure 6. These images were modified slightly to
protect the subjects’ privacy.
Table III reports the responses of these 26 more
experienced clinicians and compares their preferences
with the rankings of the 10 predoctoral judges in the
original study.
For these 4 subjects examined, the predominant
views of the more experienced judges were similar to
the average rankings of the predoctoral judges.
Additional information from the subjects’ treatment
records enabled us to test for differences in judge
preference with respect to sex, pretreatment Angle
classification, and extraction status for each image type.
Our original expectation was that girls would be ranked
more attractive than boys, and that patients who were
originally Angle Class I would be considered more
attractive than those who were originally Class II.23 For
extraction vs nonextraction preferences, we had no
a-priori hypotheses.
The results of t tests for these variables are shown
in Tables IV, V, and VI. In these tables, when the 2
groups being compared for any variable are judged to
be equally attractive in any image type, the mean value
for each will be exactly 6.5. If one group is deemed
more attractive than the other, its mean value will be
higher than 6.5; when it is deemed less attractive, its
mean value will be less than 6.5.
We infer from the findings of Table IV and V that
the judges had no strong preferences with respect to
differences in sex, original Angle classification, and
extraction status among the images they evaluated for
facial attractiveness. The nearest approach to a statisti-
cal significance concerned extraction status; there
seemed to be a slight tendency to prefer the photo-
graphs of the nonextraction subjects to those of the
extraction subjects. This tendency was slight and
reached mild statistical significance only for the profile
images. It differed somewhat from the findings of
Bowman and Johnston.24
506 Shafiee et al American Journal of Orthodontics and Dentofacial Orthopedics
April 2008

Fig 6. Triplet images for the 4 atypical subjects shown in the scatterplots of Fig 5 and examined
further in Table III. A, Case 603; B, Case 681; C, Case 690; D, Case 691.

DISCUSSION of profile, full-face, and smiling images of the same

This study is a first attempt to identify the relative
strengths of profile, full-face and smiling photographs
individually in a standard orthodontic cluster in pre-
dicting a clinician’s conclusion about a patient’s facial
attractiveness when these photographs are considered
as a triplet.
The average rankings of 10 judges for separate sets
patients correlated highly with each other and with the
average rankings of the triplets. Although the values for
all 3 image types correlated highly significantly with
the average values of the triplets (P !.0001), the
average values for the smiling photographs correlated
with the average values of the triplets significantly
better than did the average values for the profile views
American Journal of Orthodontics and Dentofacial Orthopedics Shafiee et al 507
Volume 133, Number 4

Table III. Retest of relative attractiveness—smiling vs profile photos

Predoctoral
students’ mean
rankings Preferences of 16 orthodontic residents Preferences of 10 orthodontic faculty members
Case
number Profile Smiling Profile (n) Smiling (n) No preference (n) Profile (n) Smiling (n) No preference (n)

603 2.6 9.8 1 15 0 1 8 1

681 8.4 3.4 14 1 1 8 2 0
690 3.0 9.2 0 16 0 0 10 0
691 5.4 9.7 4 9 3 3 5 2

Results of a retest to confirm the unusual relationship between the relative attractiveness of the Profile and Smiling photographs in 4 unusual cases.
As in the original rankings, the Profile image was strongly preferred to the Smiling image in Case 681, while in the other 3 cases, the Smiling
image was generally strongly preferred to the Profile image (pattern indicated in bold). The images can be seen in Fig 5.

Table IV. Mean ranks for different diagnostic categories, boys vs girls
Boys (n " 10) Girls (n " 35) Difference

Image type Mean SD Mean SD Mean SD P value

Profile 5.63 2.96 6.75 2.66 %1.12 2.73 .30 (NS)

Full-face 6.00 2.75 6.64 2.59 %0.65 2.63 .51 (NS)
Smiling 6.56 3.08 6.48 2.65 0.08 2.74 .94 (NS)
Triplet 6.45 2.88 6.51 2.58 %0.06 2.64 .96 (NS)

Note. Tables IV-VI. There were no discernable differences in judge preference as between boys and girls. There was no discernable difference
between patients who had originally presented with Class I vs Class III malocclusions. There appeared to be a slight tendency for the judges to
prefer non-extraction cases vs extraction cases, but this tendency was too small overall for us to be sure it was real.
NS, Not significant.

Table V. Mean ranks for different diagnostic categories, Angle Class I vs Angle Class II
Class I (n " 21) Class II (n " 24) Difference

Image type Mean SD Mean SD Mean SD P value

Profile 6.30 2.66 6.68 2.84 %0.38 2.76 .65 (NS)

Full-face 6.02 2.56 6.92 2.63 %0.90 2.60 .25 (NS)
Smiling 5.92 2.65 7.01 2.72 %1.09 2.69 .18 (NS)
Triplet 6.01 2.69 6.93 2.52 %0.93 2.60 .24 (NS)

NS, Not significant.

Table VI. Mean ranks for different diagnostic categories, extraction vs nonextraction
Nonextraction (n " 24) Extraction (n " 21) Difference

Image type Mean SD Mean SD Mean SD P value

Profile 7.32 2.87 5.56 2.29 1.77 2.61 .03*

Full-face 6.76 2.73 6.20 2.50 0.56 2.63 .48 (NS)
Smiling 7.14 2.97 5.77 2.23 1.36 2.65 .09 (NS)
Triplet 7.14 2.83 5.76 2.18 1.38 2.55 .08 (NS)

*P !.03; NS, not significant.

(P " .0003). This finding tends to reinforce the
previously held impression of some clinicians that the
profile view is not the optimum perspective from which
to evaluate facial attractiveness.
The patients whose images were evaluated in this
study were a fairly homogeneous group. All had re-
ceived orthodontic treatment from the same expert
clinician, and all were well dressed, reasonably attrac-
tive, and similar in age, ethnicity, and socioeconomic
status. Therefore it seems noticeable that it was still
508 Shafiee et al
possible to identify strong patterns of agreement among
the judges in each type of image.
Tables IV, V, and VI show that differences in sex,
original Angle classification, and extraction vs nonex-
traction treatment had surprisingly little impact on the
judges’ preferences for attractiveness of any type of
image. We had expected that girls would be considered
more attractive than boys, but the study produced no
evidence to support that belief. We were also mildly
surprised that no differences in preference were iden-
tified between patients who had been originally Angle
Class I vs Angle Class II. In the tests of preference
between patients who had received extraction and
nonextraction treatment, a mildly statistically signifi-
cant finding was recorded for the profile images of the
nonextraction subjects. However, when the profile im-
ages were evaluated as part of the triplet group, this
modest difference became less apparent, and no statis-
tically significant differences were noted.
As with any study that asks a new question, some
caution is necessary in generalizing from our find-
ings. Since the original sample contained no Class III
subjects, our findings are not generalizable to Class
III patients. In addition, the study might have been
more representative of actual clinical practice if we
had used experienced orthodontists rather than pre-
doctoral students for the primary rankings. However,
because the ratings of the predoctoral judges, the
orthodontic residents, and the orthodontic faculty
members substantially agree for the 4 extreme sub-
jects in Table III, this increases our confidence that
the general thrust of our findings is appropriate.
CONCLUSIONS
Clinician judges showed high consistency in rank-
ing the facial attractiveness of profile, full-face, and
smiling photographs of a group of orthodontically
treated patients whose actual differences in physical
dimensions were relatively small. The rankings of the
smiling photographs were significantly better predic-
tors of their rankings of groupings of all 3 photographs
for each patient than were the rankings of the profile
photographs (P " .0003). The rankings of the full-face
photographs appeared to be intermediate between the
profile and the smiling photographs. In general, patients
for whom any single type of image was ranked highly
also tended to have high rankings for their images of
other types.
REFERENCES
1. Arnheim R. The face and the mind behind it. In: McNamara JA
Jr, Carlson DS, Ferrara A, editors. Esthetics and the treatment of
facial form. Volume 28. Craniofacial Growth Series. Ann Arbor:
American Journal of Orthodontics and Dentofacial Orthopedics
April 2008
Center for Human Growth and Development; University of
Michigan; 1993.
2. Olds C. Facial beauty in western art. In: McNamara JA Jr,
Carlson DS, Ferrara A, editors. Esthetics and the treatment of
facial form. Volume 28. Craniofacial Growth Series. Ann Arbor:
Center for Human Growth and Development; University of
Michigan; 1993.
3. Moyers RE. Evolution of concepts of the face. In: McNamara JA Jr,
Carlson DS, Ferrara A, editors. Esthetics and the treatment of facial
form. Volume 28. Craniofacial Growth Series. Ann Arbor: Center
for Human Growth and Development; University of Michigan;
1993.
4. Rhodes G, Zebrowitz LA, editors. Facial attractiveness: evolu-
tionary, cognitive, asocial perspectives [reviewed by Donovan
JM. Hum Nature Rev 2003;3:226-8].
5. Panofsky E. The art and life of Albrecht Durer. 4th ed. Princeton,
NJ: Princeton University Press; 1955.
6. Hogarth W. The analysis of beauty. New Haven, Conn: Yale
University Press; 1997.
7. Angle EH. The treatment of malocclusion of the teeth and fractures
of the maxillae. 6th ed. 1900. S. S. White; Philadelphia: p. 15-23.
8. Wuerpel EH. On facial balance and harmony. Angle Orthod
1937;7:81-9.
9. Case CS. The development of esthetic facial contours. In: Kirk
EC, editor. The American text-book of operative dentistry.
Philadelphia: Lea Brothers; 1905. p. 849-85.
10. Reidel RA. An analysis of dentofacial relationships. Am J
Orthod 1957;43:103-19.
11. Riedel RA. Esthetics and its relation to orthodontic therapy.
Angle Orthod 1950;20:168-78.
12. Peck H, Peck S. A concept of facial esthetics. Angle Orthod
1970;40:284-317.
13. Rubenstein AJ. Variation in perceived attractiveness: differences
between dynamic and static faces. Psychol Sci 2005;16:759-62.
14. Ackerman MB, Brensinger C, Landis JR. An evaluation of
dynamic lip-tooth characteristics during speech and smile in
adolescents. Angle Orthod 2004;74:43-50.
15. Sarver DM. Videoimaging: he pros and cons. Angle Orthod
1993;63:167-70.
16. Mackley RJ. An evaluation of smiles before and after orthodontic
treatment. Angle Orthod 1993;63:183-9.
17. Peck S. Commentary: an evaluation of smiles. Angle Orthod
1993;63:190.
18. Boehnen C, Flynn P. Accuracy of 3D scanning technologies in a
face scanning context. Proceedings of the 5th International
Conference on 3-D Digital Imaging and Modeling; June 13-16,
2005; Ottawa, Ontario, Canada.
19. Matoula S, Pancherz H. Skeletofacial morphology of attractive
and non-attractive faces. Angle Orthod 2006;76:204-10.
20. Lundstrom A, Popovich F, Woodside DG. Panel assessments of
the facial frontal view as related to mandibular growth direction.
Eur J Orthod 1989;11:290-7.
21. Vargo J, Gladwin M, Ngan P. Association between ratings of
facial attractiveness and patients’ motivation for orthognathic
surgery. Orthod Craniofacial Res 2002;6:63-71.
22. Meng XL, Rosenthal R, Rubin DB. Comparing correlated
correlation coefficients. Psychol Bull 1992;111:172-5.
23. Donovan JM, Hill E, Jankowiak WR. Gender, sexual orientation,
and truth-of-consensus in studies of physical attractiveness. J Sex
Res 1989;26:264-71.
24. Bowman SJ, Johnston LE. The esthetic impact of extraction and
nonextraction treatments on Caucasian patients. Angle Orthod
2000;70:3-10.