European Journal of Orthodontics 33 (2011) 434–440

© The Author 2010. Published by Oxford University Press on behalf of the European Orthodontic Society.
doi:10.1093/ejo/cjq098 All rights reserved. For permissions, please email: journals.permissions@oup.com
Advance Access Publication 13 November 2010

Knowledge of European orthodontic postgraduate students on

biostatistics
Argy Polychronopoulou*, Theodore Eliades**, Konstantina Taoufik*,
Moschos A. Papadopoulos** and Athanasios E. Athanasiou**
*Department of Preventive and Community Dentistry, School of Dentistry, University of Athens and **Department
of Orthodontics, School of Dentistry, Aristotle University of Thessaloniki, Greece

Correspondence to: Theodore Eliades, 57 Agnoston Hiroon, Nea Ionia 14231, Greece. E-mail teliades@ath.forthnet.gr

Downloaded from https://academic.oup.com/ejo/article-abstract/33/4/434/397550 by guest on 12 May 2019

SUMMARY The purpose of this study was to explore the level of knowledge in biostatistics of orthodontic
postgraduate students. A four-section questionnaire, which included a knowledge test/quiz on biostatistics
and epidemiology, was developed. This questionnaire was distributed to postgraduate programme
directors of European universities to be delivered to students for completion under mock examination
conditions (in-class session). The frequency distributions of demographic characteristics were examined,
the percentages of participants who agreed or strongly agreed with each attitudinal statement were
calculated, and the percentages of participants who felt fairly to highly confident for each statement
were determined. Knowledge scores were calculated by the percentage of correct answers; missing
values were counted as incorrect answers. The Student’s t-test or one-way analysis of variance, where
appropriate, was utilized to determine the participants’ characteristics associated with mean knowledge
scores. Data were further analysed with multiple linear regression modelling to determine the adjusted/
unconfounded effect of possible knowledge score predictors. A two-tailed P-value of 0.05 was considered
statistically significant with a 95 per cent confidence interval (CI).
One hundred and twenty seven from a total of 129 orthodontic students who replied completed the
questionnaire. The mean correct answers of the participants were 43.8 per cent with a 95 per cent CI
of 40.2–47.3 per cent. This score was not influenced by gender, years elapsed from graduation, other
advanced degree, or year of study; the sole parameter, which seemed to influence this score was
attendance at a biostatistics/epidemiology course (51.9 versus 39.5 per cent score of participants who
had previously taken a course versus those who had not, P < 0.001). A surprising finding was the inability
of the responders to identify the appropriate use of the chi-square test (11.8 per cent, 95 per cent CI: 6.1–
17.5 per cent). The knowledge on biostatistics of orthodontic postgraduate students in Europe is only
influenced by previous relevant education.

Introduction
1995, 1999), there is still a considerable amount of research,
With the current emphasis placed on evidence-based care, the which lists inappropriate analyses, lacks descriptive data
clinician is faced with a plethora of articles, which address (Golditz and Emerson, 1985; Kay and Locker, 1996; Varnell
several clinical issues, assessing treatment modalities and et al., 2004), or includes erroneous data interpretation
exploring the predictive value of various factors on (Weiss and Samet, 1980; Wulff et al., 1987). Because of the
orthodontic therapeutic outcome. This process necessitates a introduction of complex statistical data elaboration and
substantial level of expertise of the reader to appraise the the importance given to clinical trials, basic biostatistical
design, methodology, data analysis, and interpretation of knowledge must be enhanced with knowledge of advanced
findings of relevant studies to arrive at conclusions. The methods frequently encountered in clinical research (Windish
foregoing requirements, together with the progressively et al., 2007).
increasing complexity of statistical methods reported in the To overcome these deficiencies, surveys have been
literature, make information processing a complex task. conducted to record the knowledge of health care professionals
In health care sciences, understanding biostatistics may on biostatistics and epidemiology, as well as to identify
have important implications in modulating clinical practice influencing factors. These surveys have mainly targeted
as it possesses a large effect on evidence-based diagnostic specializing physicians and have shown that even those who
and treatment applications. Similarly, in academics, sufficient are more familiar with the literature and research principles
knowledge of epidemiological principles is required to have limited biostatistical knowledge and an impaired
successfully conduct a study and correctly analyse data capacity to understand a number of epidemiological basics
derived from clinical investigations. Although check lists (Berwick et al., 1981; Altman and Bland, 1991; O’Donell,
have been developed to assess study quality (Moher et al., 2004; Estellat et al., 2006; Windish et al., 2007). A recent
KNOWLEDGE ON BIOSTATISTICS OF ORTHODONTIC STUDENTS 435

survey (West and Ficalora, 2007) demonstrated that only a All students participating in this study had to complete the
small percentage (17.6 per cent) of medical students, internal document individually, under the conditions of a formal
medicine residents, and internal medicine teaching faculties examination (i.e. no collaboration, discussion with colleagues,
believe that they have adequate training in biostatistics, while searching the Internet or books). After completion of the
even fewer (14.6 per cent) feel adequately prepared to questionnaire, the programme directors were asked to return
conduct a statistical analysis. Furthermore, in that survey, the questionnaires, using the self-addressed envelope provided.
just one-quarter of participants reported that they could The frequency distributions of demographic characteristics
identify if correct statistical methods had been applied, of participants were examined and the percentage of
whereas almost 9 out of 10 believed that they would benefit participants who agreed or strongly agreed with each
from further biostatistical training. attitudinal statement were calculated; percentages of
A review of the available literature revealed a sole participants who felt fairly to highly confident for each

Downloaded from https://academic.oup.com/ejo/article-abstract/33/4/434/397550 by guest on 12 May 2019

source of evidence on this subject as it relates to dentistry confidence statement were also determined. Knowledge
(El Tantawi, 2009). Because of its unique characteristics, scores were calculated by the percentage of correct answers;
orthodontic research deals with outcome assessment and missing values were counted as incorrect answers.
comparative analysis of population variables, materials, As values of mean knowledge scores obtained were found
and basic sciences. As such, it includes a wide range of to be approximately normally distributed, the Student’s t-test
analyses and, therefore, requires a thorough understanding or one-way analysis of variance, where appropriate, were
of biostatistics, which would better equip the practicing utilized to determine the participants’ characteristics associated
clinician to critically review the literature and formulate with mean knowledge scores. Data were further analysed with
an informed decision on the validity of novel diagnostic multiple linear regression modelling to determine the adjusted/
methods and the effectiveness of treatment modalities. unconfounded effect of possible knowledge score predictors.
This is the reason that courses on biostatistics and A two-tailed P-value of 0.05 was considered statistically
epidemiology have been included in all guidelines of significant with a 95 per cent confidence interval (95% CI).
contemporary postgraduate orthodontic education aiming All analyses were performed with the Stata version 10.0
to provide relevant familiarity and competence (van der software (Stata Corporation, College Station, Texas, USA).
Linden, 1992; Commission on Dental Accreditation, 2006;
Athanasiou et al., 2009).
Results
The purpose of this survey was to explore the level
of postgraduate students’ knowledge on biostatistics. The The demographics of responders are shown in Table 1.
overall objective was to identify areas where emphasis Completed questionnaires were returned from 127 participants
should be placed in postgraduate orthodontic curricula. from 21 postgraduate programmes in 10 European countries;
two questionnaires from those distributed were not completed.
The distribution of universities per country was as follows:
Subjects and methods
Bulgaria 1, Finland 3, Germany 8, Greece 2, Italy 2,
A questionnaire (Appendix) was structured to include four Netherlands 1, Norway 1, Switzerland 1, Sweden 1, and
basic sections: the demographics of participants, attitude UK 1. The profile of the students who responded to the
towards statistics, self-reported confidence on biostatistics, questionnaire was as follows: between 27 and 32 years of
and a knowledge section comprising 13 questions. The first age, 3–6 years from graduation from dental school, and no
section included personal data (age and gender), current previous training in biostatistics.
education level, and past statistical education of the participants. Table 2 depicts the attitude and confidence of participants
The second and third parts rated, on a five-point scale, from towards statistics. Of the 106 participants who felt confident
strongly disagree—no confidence to strongly agree— about P-value interpretation, only 37.7 per cent gave the
confidently, the self-perceived knowledge of responders. The correct answer to the corresponding question listed in the
last section included questions on various statistical subjects knowledge section. The overall correct percentage score is
in the form of cases or research in the orthodontic field. shown in Table 3; the mean correct answers was 43.8 per cent
The questionnaire, accompanied by a covering letter with with a 95% CI of 40.2–47.3 per cent. The participants achieved
directions on the correct set and mode of responding, was the highest score in recognizing the purpose of clinical trial
given in person to all orthodontic programme directors who double blinding (77.9 per cent, 95% CI: 70.6–85.2 per cent)
participated in the founding meeting of the Network of and the lowest score in recognizing a case–control investigation
Erasmus Based European Orthodontic Programmes in (3.1 per cent, 95% CI: 0.0–6.2 per cent). It is surprising that
Geneva, Switzerland, in 2008. Directors who were invited responders achieved a very low score in identifying the use of
to the meeting but were not present received the same chi-square test (11.8 per cent, 95% CI: 6.1–17.5 per cent).
material by post. The intention was that these persons would Table 4 lists the results of univariate analysis used to explore
distribute the questionnaire to the students in their participants’ demographic characteristics possibly related to
orthodontic postgraduate programmes. knowledge scores. There was no effect of gender, the number
436 A. POLYCHRONOPOULOU ET AL.

Table 1  Demographic characteristic of the 127 postgraduate Table 3  Percentages of correct answers for the knowledge
students who participated in the survey. questions.

Characteristic Category N % Correct answer

Question
Country Bulgaria 2 1.6 number Knowledge objective % 95% confidence
Finland 14 11.0 (Appendix) interval (95% CI)
Germany 35 27.6
Greece 11 8.7
7a Identify a continuous variable 44.8 36.1–53.6
Italy 19 14.9
7b Identify a nominal variable 44.0 35.3–52.8
Netherlands 12 9.5
7c Identify a dichotomous variable 54.3 45.5–63.1
Norway 9 7.0
8 Recognize a case–control study 3.2 0.0–6.2
Switzerland 5 5.5
9 Recognize a meta-analysis 70.0 62.0–78.1

Downloaded from https://academic.oup.com/ejo/article-abstract/33/4/434/397550 by guest on 12 May 2019

Sweden 8 6.3
10 Recognize the purpose of 25.2 17.5–32.8
UK 10 7.9
  randomization
Gender (%) Male 48 37.8
11 Recognize bias definition 70.8 62.8–78.8
Female 79 62.2
12 Interpret standard deviation 37.8 29.2–46.3
Age range (years) ≤26 13 10.3
13 Recognize purpose of double 77.9 70.6–85.2
27–29 43 33.8
  blinding
30–32 42 33.0
14 Interpret null hypothesis 70.0 62.0–78.1
≥33 29 22.9
15 Recognize parametric methods 62.2 53.6–70.7
Other advanced degrees Doctorate 21 16.5
16 Interpret P-value 33.0 24.7–41.3
Master of Science 36 28.3
17 Identify Cox proportional 14.1 8.0–20.3
Other 12 9.5
  hazard regression
None 58 45.7
18a Identify analysis of variance 42.5 33.8–51.2
Time since Dental School ≤2 23 18.2
18b Identify chi-square test 11.8 6.1–17.5
  graduation (years) 3–4 34 26.7
18c Identify t-test 40.1 31.5–48.8
5–6 34 26.7
19 Interpret odds ratio and 95% CI 46.4 37.6–55.2
≥7 36 28.4
20 Recognize reliability measures 40.1 31.5–48.8
Current level of training, First 47 37.0
Overall 43.8 40.2–47.3
  academic year Second 23 18.2
Third 43 33.8
Fourth 11 8.7 attendance at a biostatistics/epidemiology course where
Other/fellow 3 2.3
Previous training/coursework in No 83 65.3 participants who had previously taken this course scored
  epidemiology/biostatistics (%) Yes 44 34.7 higher than their peers (51.9 versus 39.5 per cent, P <
0.001). Even after adjustment for participants’ demographic
Table 2  Attitudes towards biostatistics and confidence of 127 characteristics, a previous statistical course attendance was
participants. found to be associated with a significant net increase in
correct answers, which reached 11.8 per cent (95% CI: 4.3–
19.4 per cent, P = 0.002; Table 5).
Attitude statement Agree or strongly
agree
Discussion
N %
The results of this survey showed that orthodontic
  I would like to gain more knowledge on 80 63.0 postgraduate students’ knowledge on biostatistics in Europe
   biostatistics is only influenced by previous relevant education.
  I understand all the statistical terms seen 19 19.9
   in journal articles As expected from other surveys, advanced statistical tests,
  I often use statistical information to 28 22.0 such as Cox proportional hazard regression, were difficult to
   formulate decisions in orthodontic correctly identify by most of the participants (Horton and
   treatment
Switzer, 2005; Windish et al., 2007); the present study
Confidence statement Fairly to highly showed only 17 per cent correctly answered this question.
confident
However, the finding that almost 9 out of 10 students could
not correctly utilize a chi-square test, and even less were
N %
able to identify a case–control study, emphasizes the
  I can interpret the P-value for a given test 106 83.5 importance of the need for substantial comprehensive
  I can assess the soundness of a statistical 83 65.3
   method used in research
education on biostatistics in postgraduate orthodontic
  I can interpret the results of a statistical 100 78.7 education. Although the above finding is in agreement with
   analysis reported in journal articles the participants’ self-assessment outcome, which revealed a
lack of understanding of biostatistical terms as well as their
of years elapsed since graduation, other advanced degree, or desire to acquire further knowledge on the subject, it is
the year of study on the mean correct score of participants. still surprising because of the widespread use of chi-square
The sole factor that seemed to influence this score was test in both applied and clinical orthodontic research.
KNOWLEDGE ON BIOSTATISTICS OF ORTHODONTIC STUDENTS 437

Table 4  Knowledge scores by selected participants’ characteristics. to emphasize the findings in a clear and concise manner in
the Results and Conclusions sections.
Characteristic Category Mean correct P-value Questionnaires were not directly sent to students but
were distributed to the participating directors at a meeting,
% 95% confidence and no follow-up took place. Of the 61 directors, 21
interval distributed the questionnaire (cooperation rate: 34.4 per
cent) to 129 students. After receiving the questionnaires
Gender (%) Male 40.8 34.8–46.8 NS* from the directors that participated in the survey, only 2 of the
Female 45.6 41.2–50.0
Age range (years) ≤26 52.1 40.7–63.5 NS** 129 that received the questionnaire failed to participate
27–29 42.1 36.6–47.6 (response rate: 98.4 per cent). Selection bias is a possibility
30–32 45.3 38.5–52.2 whenever correlates of the outcome capable of influencing
≥33 40.4 32.4–48.3

Downloaded from https://academic.oup.com/ejo/article-abstract/33/4/434/397550 by guest on 12 May 2019

Other advanced Doctorate 34.4 25.4–43.3 NS** study participation exist in some individuals at the beginning
  degrees Master of 44.7 38.3–51.1 of the study. The common element of such bias is that the
  Science relationship between exposure and outcome is different for
Other 47.2 35.4–59.0
None 45.9 40.4–51.4 those who participate and those who would be theoretically
Years since Dental ≤2 43.9 34.0–53.8 NS** eligible but do not eventually participate. In the present
  School graduation 3–4 46.5 39.8–53.2 investigation, it is unlikely that issues related to the questions
5–6 43.7 37.1–50.4
≥7 41.2 34.1–48.2 under investigation led programme directors not to distribute
Current level of First 41.1 35.0–47.2 NS** the questionnaire. Moreover, it could be implied that the
  training/year (%) Second 41.3 31.7–50.8 questionnaires were distributed in programmes with extended
Third 49.2 43.7–54.6
Fourth 44.4 29.9–58.9 student exposure to biostatistical/epidemiological issues
Other/fellow 25.9 −2.8 to 54.6 (volunteer bias and subtype of selection bias). In this scenario,
Previous training/ No 39.5 35.1–43.9 <0.001 the results may underestimate the true level of statistical
  coursework in Yes 51.9 46.5–57.2
  epidemiology/ knowledge, thus presenting a more conservative approach.
  biostatistics Previous studies have shown that biostatistics are considered
important but not to such an extent as other components of
*Analysis by t-test. the orthodontic curriculum (West and Ficalora, 2007). This
**Analysis by one-way analysis of variance. could be explained by the fact that biostatistics is most often
taught as an independent undergraduate course (West and
Table 5  Multiple regression-derived estimate of knowledge Ficalora, 2007), which may preclude the appropriate
score difference related to previous statistical training and
corresponding 95 per cent confi dence interval (95% CI) among emphasis being placed on the clinical application of terms.
127 participants. One of the advantages of the present survey is that it
included a broad range of training backgrounds, as the
results derive from 21 universities and 10 European countries.
Predictor Category Adjusted score 95% CI P-value
difference (%)* However, it should be noted that the invitation to participate in
this survey was extended to 61 universities in 17 countries.
Previous training/ No Baseline Despite the variation in postgraduate curricula among the
  coursework in Yes 11.8 4.3–19.4 0.002 countries included in this survey, the inability to identify a
  epidemiology/
  biostatistics
chi-square test or a case–control study showed decreased
variance with a 95% CI of 11–17 and 0–6 per cent,
*Score difference adjusted for gender, age, training level, other advanced
respectively. This finding implies a need to modify the current
degree, and time since graduation. curriculum format on biostatistics in orthodontics and to
include topics that would focus on research design of
It may be worth noting that the literature is replete with methods most frequently encountered in orthodontic
studies reporting that medical and dental students and research (materials and applied research, clinical trials, and
instructors do not understand basic statistics and usually basic research). This idea is further supported by the finding
provide wrong interpretations of statistical statements, despite that no differences were found between doctoral and master
the growth in statistical use (Weiss and Samet, 1980; Berwick level students. Knowledge of the English language may not
et al., 1981; Wulff et al., 1987; Altman and Bland, 1991; constitute a factor influencing the responses since there
O’Donell 2004; Estellat et al., 2006; Windish et al., 2007; was an agreement in the false answers in this specific question
El Tantawi, 2009). This lack of understanding may lead to regardless of origin. The argument that the example
both an erroneous interpretation of research findings and a reported in the question on chi-square might have been
lack of ability to critically review the evidence presented in affected by the assumption of the variable ‘flossing’ as
relevant articles. This raises a question on the applicability quantitative is not valid because firstly, for dental care
of clinical research in practice and the necessity for authors professionals, the term flossing implies a correct exercise
438 A. POLYCHRONOPOULOU ET AL.

of this hygiene effect and secondly, even if that was P. Cozza, R. Fuhrmann, G. R. Göz, K. Hansen, W. Harzer,
considered as a quantitative variable, the appropriate A. Hohoff, A. Jäger, I. Jonas, P.-G. Jost-Brinkmann,
response should have been different, that is analysis of S. Kiliaridis, J. Lisson, C. Paganelli, P. Pirttiniemi, D. Rice,
variance, from that recorded by the majority of participants. J. R. Sandy, C. Strahm, V. Vandevska-Radunovich, H. van
This survey was addressed exclusively to orthodontic Beek and J. Varrela.
postgraduate students and the questionnaire content was
derived from research published in orthodontic journals. References
Nonetheless, the results might be generalized to dental Altman D G, Bland J M 1991 Improving doctors’ understanding of
specialty training, since biostatistics at the graduate level is statistics. Journal of the Royal Statistical Society 154: 223–267
usually taught within the core course programme, and as Ambrosius W T, Manatunga A K 2002 Intensive short courses in biostatistics
such is directed to postgraduate students in general. The for fellows and physicians. Statistics in Medicine 21: 2739–2756

Downloaded from https://academic.oup.com/ejo/article-abstract/33/4/434/397550 by guest on 12 May 2019

validity of this hypothesis is confirmed by the results of Athanasiou A E et al. 2009 World Federation of Orthodontists (WFO)
guidelines for postgraduate orthodontic education. World Journal of
a recent survey (El Tantawi, 2009), where the statistical Orthodontics 10: 153–166
knowledge of dental postgraduate students was not influenced Berwick D M, Fineberg H V, Weinstein M C 1981 When doctors meet
by the curriculum of a specific dental specialty. numbers. American Journal of Medicine 71: 991–998
The present study identified the attendance of a Cheatham M L 2000 A structured curriculum for improved resident
biostatistics/epidemiology course as the sole factor that education in statistics. American Surgeon 66: 585–588
influenced the mean correct score of participants. The Commission on Dental Accreditation 2006 Accreditation standards for
advance specialty education programs in orthodontics and dentofacial
association of statistical knowledge and previous training in orthopedics. American Dental Association, Chicago
this field is a common finding in related research (West and El Tantawi M M 2009 Factors affecting postgraduate dental students’
Ficalora, 2007; Windish et al., 2007). Interestingly, in medicine, performance in a biostatistics and research design course. Journal of
this advantage in students who have received training in Dental Education 73: 614–623
statistics tends to diminish with the increasing number of years Estellat C, Faisy C, Colombet I, Chatellier G, Burnand B, Durieux P 2006
French academic physicians had a poor knowledge of terms used in
following graduation from the statistical course, a finding, clinical epidemiology. Journal of Clinical Epidemiology 59: 1009–1014
which was not observed in the present study. This might be Godwin M, Seguin R 2003 Critical appraisal skills of family physicians in
explained by the fact that specialty training in medicine exceeds Ontario, Canada. BMC Medical Education 9: 3–10
the duration of the average orthodontic postgraduate programme Golditz G A, Emerson J D 1985 The statistical content of published
in Europe. Similarly, this study showed no effect of gender, medical research: some implications for biomedical education. Medical
Education 19: 248–255
years since graduation, other advanced degree, or year of study
Horton N J, Switzer S S 2005 Statistical methods in the journal. New
on knowledge mean score. Nonetheless, an effect of gender England Journal of Medicine 353: 1977–1979
on the statistical knowledge is an isolated finding in some Kay E J, Locker D 1996 Is dental health education effective? A systematic
studies (Godwin and Seguin, 2003; Windish et al., 2007), review of current evidence. Community Dentistry and Oral Epidemiology
whereas many studies lack gender comparison (Weiss and 24: 231–235
Samet, 1980; Berwick et al., 1981; Wulff et al., 1987; Cheatham, Moher D, Jadad A R, Nichol G, Penman M, Tugwell P, Walsh S 1995 Assessing
the quality of randomized controlled trials: an annotated bibliography
2000; Ambrosius and Manatunga, 2002; Estellat et al., 2006). of scales and checklists. Controlled Clinical Trials 16: 62–73
Moher D, Cook D J, Eastwood S, Olkin I, Rennie D, Stroup D F 1999
Conclusions Improving the quality of reports of meta-analyses of randomised
controlled trials: the QUOROM statement. Quality of reporting of meta-
The mean correct answers of postgraduate orthodontic analyses. Lancet 354: 1896–1900
students to a biostatistics questionnaire was 43.8 per cent O’Donell C A 2004 Attitudes and knowledge of primary care professionals
towards evidence-based practice: a postal survey. Journal of Evaluation
This score was not influenced by gender, years elapsed from in Clinical Practice 10: 197–205
graduation, other advanced degree, or year of study; the van der Linden F P G M 1992 Three year postgraduate programme in
sole parameter, which seemed to influence this score was orthodontics: the final report of the Erasmus Project. European Journal
attendance at a biostatistics/epidemiology course (51.9 of Orthodontics 14: 85–94
versus 39.5 per cent score of participants who had previously Varnell P S, Murray M D, Janega B J, Blitstein J L 2004 Design and
analysis of group-randomized trials: a review of recent practices.
taken a course versus those who had not, P < 0.001). A American Journal of Public Health 94: 393–399
surprising finding was the inability of the responders to Weiss S T, Samet J M 1980 An assessment of physician knowledge
identify the appropriate use of the chi-square test (11.8 per of epidemiology and biostatistics. Journal of Medical Education 55:
cent, 95 per cent CI: 6.1–17.5 per cent). 692–697
West C P, Ficalora R D 2007 Clinicians attitudes towards biostatistics.
Mayo Clinic Proceedings 82: 939–943
Acknowledgements Windish D M, Huot S J, Green M L 2007 Medicine residents’ understanding
of the biostatistics and results in the medical literature. Journal of the
The authors acknowledge the contribution of the following American Medical Association 298: 1010–1022
orthodontic programme directors/representatives participated Wulff H R, Andersen B, Brandenhoff P, Guttler F 1987 What do doctors
in this survey: A.E. Athanasiou, L. Andreeva, K. Birkeland, know about statistics? Statistics in Medicine 6: 3–10
KNOWLEDGE ON BIOSTATISTICS OF ORTHODONTIC STUDENTS 439

Appendix

The questionnaire used in the survey.

BIOSTATISTICAL KNOWLEDGE SURVEY IN ORTHODONTIC EDUCATION

Country For questions 7-20 please choose the best answer

Downloaded from https://academic.oup.com/ejo/article-abstract/33/4/434/397550 by guest on 12 May 2019

University 7. A study wishes to assess dental characteristics in an
orthodontic clinic population. Which of the following
variables describes the appropriate measurement scale
Gender Male Female or type?
Fill in each blank below with your answer. Use each letter
Age years as many times as is appropriate
A. discrete
Advanced Degrees MS Doctorate Other B. dichotomous
C. ordinal
Specify other D. nominal
E. continuous
a. _____ Age at first visit in years
Years since graduation from Dental school years b. _____ Type of orthodontic problem classified as
Angle class I, class II, and class III
c. _____ Orthodontic treatment need classified as present
Current level of training1st year 2nd year 3rd year or absent

4th year Other/fellow

8. To determine if eruption sequencing is associated with
Ever taken a postgraduate course in lower incisor crowding, data from 100 orthodontic
epidemiology/biostatistics? No Yes patients with lower arch crowding were collected. The
dental charts of these patients were then reviewed to
determine whether unusual eruption patterns were
For questions 1-6 please circle the number, which best observed. This study type is known as:
describes your preference/opinion a. Cross-sectional
b. Case-control
1. I would like to gain more knowledge on biostatistics c. Retrospective cohort
1 2 3 4 5 d. Randomized clinical trial
Strongly agree Neutral Strongly disagree

2. I understand all the statistical terms seen in journal 9. The results of several clinical studies are combined
articles into a summary comparison of the association between
1 2 3 4 5 type of bonding agent and orthodontic bracket
Strongly agree Neutral Strongly disagree adherence. This summary is best described as:
a. Decision analysis
3. I often use statistical information to formulate b. Correlation analysis
decisions in orthodontic treatment c. Matched analysis
1 2 3 4 5 d. Meta-analysis
Strongly agree Neutral Strongly disagree

4. I can interpret the p value for a given result 10. A randomized clinical trial was designed to compare
1 2 3 4 5 two different treatment approaches for palatal
No Fairly Confidently expansion. The purpose of randomization in this study
was to:
5. I can assess the soundness of a statistical method a. Select a representative study sample
used in research b. Decrease the likelihood that observed outcome
1 2 3 4 5 differences are due to chance
No Fairly Confidently
c. Obtain treatment groups of equal size
d. Obtain treatment groups with comparable baseline
6. I can interpret the results of a statistical analysis
prognosis
reported in journal articles
1 2 3 4 5
No Fairly Confidently
440 A. POLYCHRONOPOULOU ET AL.

11. Any systematic error in the design, conduct or 17. In the same randomized clinical trial the researchers
analysis of a study that results in a mistaken estimate of wished to assess further if there were any differences
an exposure’s effect on the risk of disease is called: between groups over time with respect to the duration of
a. Interaction pain while controlling for other potential confounders.
b. Bias What analytic method would be the most appropriate in
c. Misclassification assessing their question?
d. Stratification a. Chi-square test
b. Kaplan-Meier analysis
c. Cox proportional hazard regression
12. In a clinical study, the age of the orthodontic patients d. Linear regression
was 11 ± 2 years (mean ± standard deviation). Which of
the following is the most correct?
a. Approximately 95% of the patients were aged between 7 18. A prospective study investigated new caries

Downloaded from https://academic.oup.com/ejo/article-abstract/33/4/434/397550 by guest on 12 May 2019

and 15 years development, dietary habits and oral hygiene habits
b. Most of the patients were aged 11 years; the remainder among orthodontic adolescents. Match the appropriate
was aged between 9 and 13 years analytic method for each of the following hypotheses.
c. It is 95% certain that the true mean lies within the interval Fill in each blank below with your answer. Use each letter
of 7-15 years as many times as is appropriate
d. No patients were younger than 7 or older than 15 years A. T-test
B. Chi-square test
C. ANOVA
13. The purpose of a double-blind orthodontic trial is to: D. Logistic regression
a. Reduce the effect of sampling variation a. _______ Mean daily sugar intake does not vary across 3
b. Achieve comparability of untreated and treated groups of decay (i.e. absent, moderate, severe)
participants b._______ Use of dental floss does not vary across 3 groups
c. Avoid observer and subject bias of decay (i.e. absent, moderate, severe)
d- Avoid observer bias and sampling variation c._______ Mean daily sugar intake is the same for those
with decay development as compared to those with no decay

14. Characteristics of the null hypothesis include:

a. Is a statement of no difference between/among groups 19. A case control study investigated the possible
b. Is a statement of difference between/among groups relationship between thumb sucking and orthodontic
c. If the null hypothesis is rejected, there is no difference treatment need. If the estimated odds ratio for treatment
d. If the null hypothesis is accepted, there is difference was 1.5 in the presence of habit compared to the
absence, with a 95% confidence interval of (1.1, 2.2), it
can be inferred that there is:
15. A case control investigation explores risk factors for a. A biological plausible relationship
lower incisor crowding. To employ parametric methods b. A clinically important finding
of statistical analysis the data must be: c. A statistically significant result
a. Age-matched d. A statistically non significant result
b. Normally distributed
c. Nominal
d. Linear 20. In the previous case control study, information on
sucking habit was based on child’s self-report. If
exposure information could also be obtained from an
16. In a randomized clinical trial of the use of aspirin independent source (such as dental records, or reports
and placebo to prevent mild post banding pain, 47% of from parents), then the agreement between these two
the patients receiving aspirin and 48% of those receiving methods could be compared. Which of the following
placebo reported pain. In reporting this overall finding measures would be most appropriate to quantify the
the authors stated that p>.05. This means: reliability between the two methods?
a. The chance is 95% that the study is correct
b. The chances are greater than 1 in 20 that a difference a. A kappa coefficient
would be found again if the study were repeated b. Correlation coefficient of reproducibility
d. The probability is less that 1 in 20 that a difference this c. Intraclass correlation coefficient
large could occur by chance alone d. Product-moment correlation
d. The probability is greater than 1 in 20 that a difference
large could occur by chance alone