Biostatistics in Orthodontics

APPLICATION
OF
BIOSTATISTICS
IN
ORTHODONTICS
www.indiandentalacademy.com
INDIAN DENTAL ACADEMY
Leader in continuing dental education
STATISTICS
STATISTICS AS A SINGULAR NOUN IS “A

SCIENCE OF FIGURES”
WHERE AS PLURAL NOUN IT MEANS “FIGURES”

OR NUMERICAL DATA OR INFORMATION.
BIOSTATISTICS
BIOSTATISTICS CAN BE DEFINED AS ART AND
SCIENCE OF COLLECTION, COMPILATION,
PRESENTATION, ANALYSIS AND LOGICAL
INTERPRETATION OF BIOLOGICAL DATA
AFFECTED BY MULTIPLICITY OF FACTORS
“An ounce of truth produces tons of statistics”

STATISTICS
THE WORD STATISTIK IS DERIVED FROM AN

ITALIAN WORD STATISTA MEANING
STATESMAN.
GOTTFRED CHENWALL, A PROFESSOR AT
MARLBOROUGH USED THIS WORD FOR THE
FIRST TIME.
ZIMMERMAN INTRODUCED THE WORD
STATISTICS INTO ENGLAND.
HISTORY OF STATISTICS
DURING THE OUTBREAK OF PLAGUE IN ENGLAND,
IN 1532 THEY STARTED PUBLISHING THE
WEEKLY DEATH STATISTICS.THIS PRACTICE
CONTINUED AND BY 1632, THESE BILLS OF
MORTALITY, LISTED BIRTHS AND DEATHS BY SEX
HISTORY OF STATISTICS..
IN 1662, CAPT.JOHN GRAUNT USED 30

YEARS OF THESE BILLS TO MAKE
PREDICTIONS ABOUT THE NUMBER OF
PEOPLE WHO WOULD DIE FROM
VARIOUS DISEASES AND PROPORTIONS
AF MALE AND FEMALE BIRTHS THAT
COULD BE EXPECTED.
KNOWLEDGE OF STATISTICAL METHODS
1. ENABLES US TO MAKE INTELLIGENT USE OF THE

CURRENT LITERATURE.
2. OPENS UP NEW PATHS OF EXPERIMENTAL
PROCEDURES
3. ENABLES A RESEARCH WORKER TO COLLECT,
ANALYZE AND PRESENT HIS DATA IN THE MOST
MEANINGFUL AND EXPEDITIOUS MANNER.
4. ALLOWS A BIOINFORMATICS PROFESSIONAL USE
STATISTICAL SOFTWARES IN A MEANINGFUL
MANNER www.indiandentalacademy.com
LIMITATIONS
STATISTIC LAWS ARE NOT EXACT LAWS LIKE

MATHEMATICAL OR CHEMICAL LAWS BUT ARE
ONLY TRUE IN MAJORITY OF CASES.
EX: WHEN WE SAY THAT THE AVERAGE HEIGHT

OF AN ADULT INDIAN IS 5’ 6’’ , IT INDICATES
THE HEIGHT NOT OF INDIVIDUAL BUT OF A
GROUP OF INDIVIDUALS.
SUBDIVISIONS OF STATISTICS
THEY CAN BE SEPERATED INTO TWO BROAD
CATEGORIES:
1. DESCRIPTIVE STATISTICS
2. INFERENTIAL STATISTICS
DESCRIPTIVE STATISTICS
95% C I for Mean

Norm Sample Mean Std. Max
Std. Error Lower Upper Min
size Deviation
bound bound
LED 40 sec
10 9.659 0.615891 0.19476168 9.218418476 10.099581 8.34 10.7
LED 20 sec
10 7.596 0.816921 0.25833312 7.011609886 8.1803901 6.36 8.95
Argon Laser 10 sec
10 7.568 1.741518 0.5507163 6.322193174 8.8138068 3.6 9.47
Argon Laser 5 sec
10 5.824 1.636773 0.51759315 4.653122953 6.9948770 4.37 8.93
Halogen Light 40 sec
10 10.374 1.688939 0.53408946 9.165805693 11.582194 8.21 12.97
DATA
WHENEVER AN OBSERVATION IS MADE, IT WILL

BE RECORDED AND A COLLECTIVE RECORDING
OF THESE OBSERVATIONS, EITHER NUMERICAL
OR OTHERWISE, IS CALLED A DATA.
EX: RECORDING THE SEX OF A PERSON IN A

GROUP OF PERSONS
VARIABLE
IN EACH OF CASES A CERTAIN OBSERVATION IS
MADE FOR A CHARACTERISTIC AND THIS
CHARACTERISTICS VARIES FROM ONE
OBSERVATION TO OTHER OBSERVATION AND IS
CALLED A VARIABLE
TYPES OF DATA
I. QUALITATIVE / QUANTITATIVE
II. DISCRETE / CONTINUOUS
III.GROUPED / UNGROUPED
IV.PRIMARY / SECONDARY
V. NOMINAL / ORDINAL
TYPES OF CLINICAL DATA THAT
CAN BE SUPPORTED BY STATISTICS
STATISTICS CAN BE USED TO HELP THE READER

MAKE A CRITICAL EVALUATION OF VIRTUALLY
ANY QUANTITATIVE DATA.
IT IS IMPORTANT THAT THE STATISTICAL

TECHNIQUES USED ARE APPROPRIATE FOR THE
GIVEN EXPERIMENTAL DESIGN.
NEED FOR ORGANISING THE DATA
DATA ARE NOT NECESSARILY INFORMATION,

AND HAVING MORE DATA DOES NOT
NECESSARILY PRODUCE BETTER DECISIONS.
THE GOAL IS TO SUMMARISE AND PRESENT

DATA IN USEFUL WAYS TO SUPPORT PROMPT
AND EFFECTIVE DECISIONS.
METHODS OF PRESENTATION OF DATA
•TABULATION
•CHARTS AND DIAGRAMS
GUIDELINES PRESENTATION OF TABLES
1. TABLE MUST BE NUMBERED
2. TITLE-BRIEF AND SELF EXPLANATORY –

SHOULD BE GIVEN
3. THE HEADINGS OF COLUMNS AND ROWS MUST

BE CLEAR, SUFFICIENT, CONCISE AND FULLY
DEFINED
GUIDELINES PRESENTATION OF TABLES..
4. THE DATA MUST BE PRESENTED ACCORDING TO

SIZE OF IMPORTANCE - CHRONOLOGICALLY,
ALPHABETICALLY OR GEOGRAPHICALLY
5. FULL DETAILS OF DELIBERATE EXCLUSIONS IN
COLLECTED SERIES MUST BE GIVEN.
6. IF DATA INCLUDES RATE OR PROPORTION
MENTION THE DENOMINATOR I.E. NUMBER OF
OBSERVATIONS FROM WHICH THEY ARE DERIVED.
6. TABLE SHOULD NOT BE TOO LARGE.

8. FIGURES NEEDING COMPARISON SHOULD BE
PLACED AS CLOSE AS POSSIBLE
9. ARRANGEMENT SHOULD BE VERTICAL.
10. FOOT NOTES SHOULD BE GIVEN WHEREVER
NECESSARY.
Table-11Descriptive Statistics of Shear bond strength

Norm 95% C I for
Sample Mean Mean Max
SD S.E. Min
size Lower Upper
bound bound
LED 40sec 10 9.659 0.6158 0.1947 9.2184 10.09 8.34 10.7
PRESENTATION THROUGH
CHART / DIAGRAM / GRAPH
LINE CHART
BAR DIAGRAM
40
35
30
25
20
4th Qtr
15
10
3rd Qtr
5
0
1st Qtr 2nd Qtr 3rd Qtr 4th Qtr
2nd Qtr
1st Qtr
0 20 40 60
BAR DIAGRAM…
90
MULTIPLE BAR
80
70
60 COMPONENT BAR
50 East
40 West
30 North
100%
20 90%
10 80%
70%
0
1st Qtr 2nd Qtr 3rd Qtr 4th Qtr 60%
50%
40%
30%
20%
10%
0%
1st 2nd 3rd 4th
Qtr
www.indiandentalacademy.com Qtr Qtr Qtr
HISTOGRAM
FREQUENCY POLYGON
PIE DIAGRAM
1st Q
1st Qtr 1st Qtr
2nd
2nd Qtr 2nd Qtr 3rd Q
3rd Qtr 3rd Qtr 4th Q
4th Qtr 4th Qtr
1st Q
2nd Q
3rd Q
4th Q
1st Qtr
2nd Qtr
3rd Qtr 1st Q
4th Qtr 2nd Q
3rd Q
4th Q
SCATTER DIAGRAMS
100
90
80
70
60 East
50 West
40 North
30
20
10
0
0 2 4 6
BOX PLOT
VENN DIAGRAM
PICTORGRAM
SHADED MAPS / SPOT MAPS / DOT MAPS
STEPS IN STATISTICAL METHODS
1. COLLECTION OF DATA
2. CLASSIFICATION
3. TABULATION
4. PRESENTATION BY GRAPHS
5. DESCRIPTIVE STATISTICS
6. ESTABLISHMENT OF RELATIONSHIP
7. INTERPRETATION
TYPES OF STUDIES
DESCRIPTIVE ANALYTICAL
•OBSERVATIONAL
•CORRELATIONAL
- CASE CONTROL
•CASE STUDIES
- COHORT
-CASE REPORTS
•INTERVENTIONAL
-CASE SERIES -CLINICAL TRIALS
•CROSS SECTIONAL SURVEYS -ANIMAL EXPERIMENTS
RESEARCH DESIGNS
EXPLORATIVE
DESCRIPTIVE
DIAGNOSTIC
EXPERIMENTAL
DESIGN OF THE INVESTIGATION
1. RETROSPECTIVE SURVEYS
2. PROSPECTIVE SURVEYS
3. FOLLOW UP STUDIES
4. CROSS SECTIONAL SURVEYS
5. PROPHYLACTIC TRIALS
6. THERAPEUTIC TRIALS
COHORT STUDY
SUBJECTS ARE DIVIDED INTO GROUPS

DEPENDING ON PRESENCE OR ABSENCE OF A
RISK FACTOR AND THEN FOLLOWED UP FOR A
PERIOD OF TIME TO FIND OUT WHETHER THEY
DEVELOP THE DISEASE OR NOT. THIS IS
PROSPECTIVE RESEARCH.
TROHOC STUDY
THE STUDY IS DESIGNED TO INVESTIGATE THE
ASSOCIATION BETWEEN A FACTOR AND A
DISEASE.THESE STUDIES ARE KNOWN AS
TROHOC STUDY. SINCE THESE FORM A
RETROSPECTIVE INVESTIGATION i.e. OPPOSITE
OF A COHORT STUDY.
INTERVENTIONAL STUDIES
THESE ARE ALSO KNOWN AS EXPERIMENTAL

STUDIES OR CLINICAL TRIALS. IN THESE
STUDIES THE INVESTIGATOR DECIDES WHICH
SUBJECT GETS EXPOSED TO A PARTICULAR
TREATMENT (OR PLACEBO). THESE STUDIES MAY
BE COHORT OR CASE-CONTROL.
EX-ANIMAL EXPERIMENTS,ISOLATED TISSUE

EXPERIMENTS,IN VITRO EXPERIMENTS.
INTERVENTIONAL STUDIES
•RANDOMIZED CONTROLLED TRIALS/CLINICAL

TRIALS-WITH PATIENTS AS UNIT OF STUDY
•FIELD TRIALS/COMMUNITY INTERVENTION

STUDIES-WITH HEALTHY PEOPLE AS UNIT OF
STUDY
•COMMUNITY TRIALS-WITH COMMUNITIES AS

UNIT OF STUDY
STUDY DESIGNS
1. CASE REPORT
2. CASE SERIES REPORT
3. INCIDENCE PREVALENCE STUDIES
4. TROHOC STUDY
5. COHORT STUDY
6. RANDOMIZED CONTROLLED TRIALS
7. META ANALYSIS
SAMPLING
SAMPLING IS THE SELECTION OF THE PART OF AN

AGGREGATE TO REPRESENT THE WHOLE
SAMPLE A FINITE SUBSET OF STATISTICAL

INDIVIDUALS IN A POPULATION
SAMPLE SIZE THE NUMBER OF INDIVIDUALS IN A

STUDY
SAMPLE SELECTION-GUIDELINES
1.WELL CHOSEN
2.SUFFICIENTLY LARGE (TO MINIMIZE SAMPLING ERROR)
3.ADEQUATE COVERAGE
METHODS OF SAMPLING
1. NON RANDOM SAMPLING
2. PROBABILITY SAMPLING
PROBABILITY SAMPLING
1. SIMPLE RANDOM SAMPLING- WITH OR WITHOUT REPLACEMENT
2. SYSTEMATIC SAMPLING
3. STRATIFIED SAMPLING
4. CLUSTER SAMPLING
5. SUB SAMPLING/ MULTISTAGE SAMPLING
6. MULTIFACE SAMPLING
FACTORS INFLUENCING SAMPLE SIZE
1. DIFFERENCE EXPECTED
2. POSITIVE CHARACTER
3. DEGREE OF VARIATION AMONG SUBJECTS
4. LEVEL OF SIGNIFICANCE DESIRED- p VALUE
5. POWER OF THE STUDY DESIRED
6. DROP OUT RATE

DETERMINATION OF SAMPLE SIZE
QUANTITATIVE DATA
4 SD2 SD= STANDARD DEVIATION

N=
L2 L = ALLOWABLE ERROR
QUALITATIVE DATA
P = POSITIVE CHARACTER
4 pq
L = ALLOWABLE ERROR
N=
L2 Q = 1- p
THE SAMPLE SIZE WAS DETERMINED FROM THE

PARAMETER OF ARCH LENGTH WITH THE LIKELY
CHANGE IN ARCH LENGTH BEING HALF OF THE
DECIDUOUS INCISORS(3MM) WITH A SD OF 2.8MMS,
A POWER OF .85 WITH SIGNIFICANCE AT THE LEVEL
OF .05 WOULD REQUIRE A SAMPLE SIZE OF 35
Journal of orthodontics Vol 31:2004,107-114
PRECISION
INDIVIDUAL BIOLOGICAL VARIATION, SAMPLING

ERRORS AND MEASUREMENT ERRORS LEAD TO
RANDOM ERRORS LEAD TO LACK OF PRECISION
IN THE MEASUREMENT. THIS ERROR CAN NEVER
BE ELIMINATED BUT CAN BE REDUCED BY
INCREASING THE SIZE OF THE SAMPLE
PRECISION
PRECISION= square root of sample size
standarad deviation
STANDARD DEVIATION REMAINING THE SAME,

INCREASING THE SAMPLE SIZE INCREASES
THE PRECISION OF THE STUDY.
STRATEGIES TO ELIMINATE ERRORS
1. CONTROLS
2. RANDOMIZATION OR RANDOM ALLOCATION
3. CROSS OVER DESIGN
4. PLACEBO
5. BLINDING TECHNIQUE -SINGLE/ DOUBLE BLINDING
EXPERIMENTAL VARIABILITY
ERROR/ DIFFERENCE / VARIATION
THERE ARE THREE TYPES
1. OBSERVER-subjective / objective
2. INSTRUMENTAL
3. SAMPLING DEFECTS OR ERROR OF BIAS
BIAS IN THE SAMPLE
THIS IS ALSO CALLED AS SYSTEMATIC

ERROR. THIS OCCURS WHEN THERE IS A
TENDENCY TO PRODUCE RESULTS THAT
DIFFER IN A SYSTEMATIC MANNER FROM
THE TRUE VALUES. A STUDY WITH SMALL
SYSTEMATIC ERROR IS SAID TO HAVE
HIGH ACCURACY.ACCURACY IS NOT
AFFECTED BY THE SAMPLE SIZE.
BIAS IN THE SAMPLE..
ACCURACY IS NOT AFFECTED BY THE SAMPLE

SIZE. THERE ARE AS MANY AS 45 TYPES OF
BIASES, HOWEVER THE IMPORTANT ONES
ARE:
1. SELECTION BIAS
2. MEASUREMENT BIAS
3. CONFOUNDING BIAS
ERRORS IN SAMPLING
SAMPLING ERRORS NON SAMPLING ERRORS
Faulty sampling design Coverage error

-due to non response or non
cooperation of the informant
Small size of the sample Observational error
-due to interviewers bias,imperfect
exptl. design,or interaction
Processing error
-due to errors in statistical analysis
DAHLBERG’S FORMULA
DAHLBERG IN 1940 USED THIS FORMULA TO

CALCULATE THE METHOD ERROR
Method error=√d2
2n
WHERE d=DIFFERENCE BETWEEN TWO MEASUREMENTS OF
A PAIR
n = NUMBER OF SUBJECTS
DISTRIBUTIONS
WHEN YOU HAVE A COLLECTION OF POINTS
YOU BEGIN THE INITIAL ANALYSIS BY
PLOTTING THEM ON A GRAPH TO SEE HOW
THEY ARE DISTRIBUTED
DISTRIBUTION-TYPES
1. NORMAL-GAUSSIAN
2. BINOMIAL
3. POISSON
4. RECTANGULAR OR UNIFORM
5. SKEWED
6. LOG NORMAL
7. GEOMETRIC
DISTRIBUTION-TYPES..
UNIFORM OR RECTANGULAR
BIMODAL
NORMAL OR GAUSSIAN DISTRIBUTION
CHARACTERISTICS OF NORMAL DISTRIBUTION
1. THE CURVE HAS A SINGLE PEAK, THUS IT IS

UNI MODAL
2. IT HAS A BELL SHAPE
3. MEAN, MEDIAN AND MODE ARE THE SAME
VALUES.
4. TWO TAILS EXTEND INDEFINITELY AND NEVER
TOUCH THE HORIZONTAL AXIS (THIS MEANS THAT
INFINITE NUMBER OF VALUES ARE POSSIBLE)
CONFIDENCE LIMITS
POPULATION MEAN+1 SE LIMITS INCLUDE

68.27% OF THE SAMPLE MEAN VALUES
POPULATION MEAN+1.96 SE LIMITS INCLUDE
95% OF THE SAMPLE MEAN VALUES
99% OF THE SAMPLE MEAN VALUES
CONFIDENCE LIMITS

99.9% OF THE SAMPLE MEAN VALUES
THESES LIMITS ARE CALLED CONFIDENCE

LIMITS AND THE RANGE BETWEEN THE TWO
IS CALLED THE CONFIDENCE INTERVAL
NORMAL DISTRIBUTIONS WITH
SAME MEAN AND VARIED STANDARD DEVIATION
BINOMIAL DISTRIBUTION
THE BINOMIAL DISTRIBUTION IS USED FOR

DESCRIBING DISCRETE NOT THE CONTINUOUS
DATA. THESE VALUES ARE AS A RESULT OF AN
EXPERIMENT KNOWN AS BERNOULLI’S
PROCESS.THEY ARE USED TO DESCRIBE
1. ONE WITH CERTAIN CHARACTERISTIC
2. REST WITHOUT THIS CHARACTERISTIC
THE DISTRIBUTION OF THE OCCURRENCE OF THE
CHARACTRERISTIC IN THE POPULATION IS
DEFINED BYTHE BINOMIAL DISTRIBUTION.
THE POISSON DISTRIBUTION
IF IN A BINOMIAL DISTRIBUTION THE VALUE OF

PROBABILITY OF SUCCESS AND FAILURE OF AN
EVENT BECOMES INDEFINITELY SMALL AND THE
NUMBER OF OBSERVATION BECOMES VERY LARGE,
THEN BINOMIAL DISTRIBUTION TENDS TO
POISSON DISTRIBUTION.
THIS IS USED TO DESCRIBE THE OCCURRENCE OF

RARE EVENTS IN A LARGE POPULATION.
DISPERSION ?
DATA OBSERVATIONS TOTAL .MEAN

SET
I 00 10 20 25 70 125 25
II 23 24 25 26 27 125 25
IT IS NECESSARY TO STUDY THE VARIATION. THIS

VARIATION IS ALSO KNOWN AS DISPERSION.IT
GIVES US INFORMATION, HOW INDIVIDUAL
OBSERVATIONS ARE SCATTERED OR DISPERSED
FROM THE MEAN OFwww.indiandentalacademy.com
LARGE SERIES.
DIFFERENT MEASURES OF DISPERSION
1. RANGE
2. QUARTILE DEVIATION
3. COEFFICIENT OF QUARTILE DEVIATION
4. MEAN DEVIATION
5. STANDARD DEVIATION
6. VARIANCE
7. COEFFICIENT OF VARIATION
STANDARD DEVIATION
1. STANDARD DEVIATION INDICATES HOW CLOSE

THE INDIVIDUAL READINGS TO THE MEAN.
2. THE SMALLER THE STANDARD DEVIATION, THE

MORE HOMOGENEOUS IS THE SAMPLE.
3. A LARGER SD IMPLIES THAT THE INDIVIDUAL

SUBJECTS MEASUREMENTS VARY WIDELY.
4. THE SD TENDS TO GET SMALLER AS THE

SAMPLE SIZE INCREASES.
COEFFICIENT OF VARIATION
WHEN YOU WANT TO COMPARE TWO OR
MORE SERIES OF DATA WITH EITHER
DIFFERENT UNITS OF MEASUREMENTS OR
EITHER MARKED DIFFERENCE IN MEAN, A
RELATIVE MEASURE OF DISPERSION,
COEFFIENT OF VARIATION IS USED.
C.V. = ( S X100)
X
STANDARD ERROR OF THE Mean
STANDARD ERROR OF THE MEAN= STANDARD DEVIATION
A LARGE STANDARD ERROR IMPLIES THAT WE
SQUARE ROOT OF NUMBER OF SUBJECTS
CANNOT BE VERY CONFIDENT THAT OUR SAMPLE
STATISTICS ARE REALLY GOOD ESTIMATES OF
POPULATION PARAMETERS
A SMALL STANDARD ERROR ALLOWS US TO FEEL

MORE CONFIDENT THAT OUR SAMPLE STATISTICS
ARE REPRESENTATIVE OF POPULATION
PARAMETERS.
Population means are best used as bases for comparison,not as treatment goals.
“P” VALUE- SIGNIFICANCE
IT REPRESENTS THE PROBABILITY.

TO DETERMINE IF THE TREATMENT GROUP IS
DIFFERENT FROM CONTROL GROUP
IF IT IS LESS THAN .05, IT MEANS THERE ARE
FEWER THAN 5 CHANCES OUT OF 100 THAT THE
DIFFERENCE WE OBSERVE ARE DUE TO RANDOM
CHANCE ALONE.
LESS THAN .01
LESS THAN .001 www.indiandentalacademy.com
CRITICAL RATIO, Z SCORE
It indicates how much an observation is bigger or

smaller than mean in units of SD
Z ratio = Observation – Mean
Standard Deviation
The Z score is the number of SDs that the simple
mean depart from the population mean.
As the critical ratio increases the probability of
accepting null hypothesis decreases.
VARIANCE RATIO OR FISCHER “F” TEST
FOR COMPARISON OF VARIANCE (SD2 ) BETWEEN

THE GROUPS (OR SAMPLES SD12 AND SD22 )
VARIANCE RATIO TEST IS UTILISED. THIS TEST
INVOLVES A DISTRIBUTION KNOWN AS “F”
DISTRIBUTION. THIS WAS DEVELOPED BY
FISHER AND SNEDECOR WITH DEGREES OF
FREEDOM OF N1-1 AND N2-1
IF THE CALCULATED F VALUES ARE GREATER

THAN THE VALUE TABULATED F VALUE AT
0.05% OR AT 1% LEVEL THAN THE VARIANCES
ARE SIGNIFICANTLY DIFFERENT FROM EACH
OTHER. IF THE F VALUE CALCULATED IS LOWER
THAN THE TABULATED THAN THE VARIANCES
BY BOTH SAMPLES ARE SAME AND ARE NOT
SIGNIFICANT
LEVENE’S TEST FOR EQUALITY
F Significance
SB with LED 40sec
SB with Halogen40sec 10.35895 0.004764
NULL HYPOTHESIS
IT IS A HYPOTHESIS WHICH ASSUMES THAT THERE

IS NO DIFFERENCE BETWEEN TWO VALUES SUCH
AS POPULATION MEANS OR POPULATION
PROPORTIONS.
WHEN YOU ARE SUBJECTING TO NULL HYPOTHESIS
CERTAIN TERMINOLOGIES SHOULD BE CLEAR.
NULL HYPOTHESIS…..
1. ALTERNATE HYPOTHESIS
2. TEST STATISTIC
3. DEGREES OF FREEDOM
4. SAMPLING ERRORS
5. LEVEL OF SIGNIFICANCE
6. POWER OF THE TEST
7. REGIONS OF ACCEPTANCE AND REJECTION
8. ONE TAILED / TWO TAILED TEST
PROCEDURE FOR TESTING
THE HYPOTHESIS
STEP-1 SET UP THE NULL HYPOTHESIS
STEP-2 SET UP THE ALTERNATE HYPOTHESIS
STEP-3 CHOOSE THE APPROPRIATE LEVEL OF

SIGNIFICANCE
STEP-4 COMPUTE THE VALUE OF TEST STATISTIC

Z VALUE = OBSERVED DIFFERENCE
STANDARD ERROR
PROCEDURE FOR TESTING
THE HYPOTHESIS…
STEP-5 OBTAIN THE TABLE VALUE AT THE
GIVEN LEVEL OF SIGNIFICANCE
STEP-6 COMPARE THE VALUE OF Z WITH THAT
OF TABLE VALUE
STEP-7 DRAW THE CONCLUSION
NULL HYPOTHESIS…..
CONCLUSION BASED ON SAMPLE

POPULATION
NULL HYPOTHESIS NULL HYPOTHESIS
REJECTED ACCEPTED
NULL HYPOTHESIS TYPE I ERROR CORRECT

TRUE DECISION
NULL HYPOTHESIS CORRECT DECISION TYPE II ERROR

FALSE
AREA OF ACCEPTANCE, REJECTION
TESTS OF SIGNIFICANCE
Parametric Non Parametric

1 Student paired T test 1 Wilcoxan signed rank test
2 Student unpaired T test 2 Wilcoxan rank sum test

3 One way Anova 3 Kruskal wallis one way anova
4 Two way Anova 4 Friedman one way anova
5 Correlation coefficient 5 Spearman’s rank correlation
6 Regression analysis 6 Chi-square test
STUDENT’S ‘t’ TEST
THIS TEST IS A PARAMETRIC TEST DESCRIBED

BY W.S.GOSSETT WHOSE PEN NAME WAS
“STUDENT”. IT IS USED FOR SMALL SAMPLES,
I.E. LESS THAN 30.
T Test can be:
Paired t test
Unpaired t test
STUDENT’S ‘t’ TEST
PAIRED ‘T’ TEST IS USED FOR A GROUP WHICH IS

ITS OWN CONTROL
Ex Effect of bionator on mandibular length
UNPAIRED ‘T’ TEST FOR COMPARING TWO

DIFFERENT GROUPS, ONE OF WHICH MAY BE
CONTROLLED AND THE OTHER TEST GROUP.
Ex:Assessment of arch width of maxilla in thumbsuckers and
normal subjects
ANALYSIS OF VARIANCE (ANOVA)
THIS TEST IS USED TO COMPARE THE MEANS

OF THREE OR MORE GROUPS TOGETHER.
THIS IS USED WHEN-
•SUBGROUPS TO BE COMPARED ARE DEFINED

BY JUST ONE FACTOR
•SUBGROUPS ARE BASED ON TWO FACTORS.
•DATA ARE NORMALLY DISTRIBUTED.

ANALYSIS OF VARIANCE (ANOVA)…
THE SHEAR BOND STRENGTH OF ADHESIVE

CURED USING FOUR DIFFERENT LIGHT CURING
UNITS ARE TO BE COMPARED.
SBS BELONGING TO THE FOUR LIGHT CURING
UNITS ARE TAKEN AND MEAN SBS FOR EACH
CURING LIGHT IS DETERMINED.
THESE MEANS ARE COMPARED TOGETHER TO
ASCERTAIN ANY DIFFERENCE BETWEEN THEM.
ANOVA and POST HOC TEST-
MULTIPLE TEST OF BONFERRONI
Source of variation Sum of Mean
Squares df Square F Sig.
Between groups 132.6448 4 33.1612 17.2515 <0.00000012
Within groups
86.4999 45 1.92222
CONTROL OTHER GROUPS SIGNIFICANCE
LED 20 seconds 0.01754

LED 40 seconds Argon Laser 10 seconds 0.01540
Argon Laser 5 seconds 1.6575
Conventional Halogen 40 1
seconds
The mean difference is significant at the .05 levels
RESULTS OF ANOVA
IF F1>F0.05 >F0.01
THEN THE PROBABILITY OF SIGNIFICANCE IS
P<0.05 P<0.01 RESPECTIVELY
F1<F0.05
THEN THE PROBABILITY OF SIGNIFICANCE IS
P>0.05(not significant)
TWO WAY ANALYSIS OF VARIANCE
TWO WAY ANALYSIS CAN BE USED IN THE ABOVE

SITUATION IF THE INFLUENCE OF TIME APART
FROM THE CURING LIGHT IS ALSO TO BE TAKEN
INTO CONSIDERATION.
IN THIS CASE THE DATA ARE CLASSIFIED BY
TWO FACTORS I.E. CURING LIGHT AND TIME.
MANOVA
VARIABLE Before appliance
insertion
End of active
expansion
Immediately after
removal of appliance
36.325± 3.169 42.754± 3.030 42.302± 2.926
Molar cusp width
29.119± 2.446 Not measured 35.063± 2.230
Molar gingival width
29.725± 2.886 32.943± 2.913 32.759± 2.476

Canine cusp width
23.411± 3.247 26.637± 3.200 26.526± 2.914
Canine gingival width
0.719± 0.814 3.095± 1.447 Not measured
Diastema width
73.256± 4.133 77.137± 4.224 76.157± 4.759

Maxillary perimeter
Not measured 5.790± 1.141 Not measured
Screw separation
Not measured 4.046± 1.115 Not measured

Anterior suture expansion
Comparison of skeletal and dental changes between 2 point and 4 point rapid
Not measured palatal 1.000
1.837± expanders AJO:2003 Not
123;321-328
measured
Posterior suture expansion www.indiandentalacademy.com
DETERMINATION OF “r” VALUE
WHEN THE DEGREE OF LINEAR (STRAIGHT LINE)

ASSOCIATION BETWEEN TWO VARIABLES IS
REQUIRED, CORRELATION COEFFICIENT IS
CALCULATED.
Ex: MEASURE THE CHANGES IN FMA AND THE
CHANGES THAT OCCURRED IN POGONION POSITION
AND PLOT THE DETERMINED VALUES ON GRAPH
PAPER.
CORRELATION COEFFICIENT (r)…
A LINE OF BEST FIT IS THEN MADE TO CONNECT

THE MAJORITY OF THE PLOTTED VALUES.
ONE HAS TO LOOK AT A SCATTER PLOT OF THE
DATA BEFORE PLACING ANY IMPORTANCE ON THE
MAGNITUDE OF CORRELATION.
Height in cms Weight in Kg
1 182.1 79.5
2 172.5 61.5
3 175.7 68.2
4 172.8 66.4
5 160.3 52.6
6 165 .5 54.3
7 172.8 61.1
8 162.4 52.8
POSITIVE CORRELATION NEGATIVE CORRELATION
PARTIAL
POSITIVE CORRELATION
PARTIAL
NEGATIVE CORRELATION
ABSOLUTELY
www.indiandentalacademy.com NO CORRELATION
LINEAR REGRESSION ANALYSIS
LINEAR REGRESSION IS RELATED TO CORRELATION

ANALYSIS.
THIS SEEKS TO QUANTIFY THE LINEAR

RELATIONSHIP THAT MAY EXIST BETWEEN AN
INDEPENDENT VARIABLE “x” AND A DEPENDENT
VARIABLE “y”
Y=a+bx
LINEAR REGRESSION ANALYSIS
COMPARABLE PARAMETRIC and
NON PARAMETRIC TESTS
use parametric Non parametric
To compare two paired Paired ‘t” test Wilcoxan signed rank

samples for equality of means test
To compare two independent Unpaired ‘t” test Mann Whitney test
samples for equality of means
To compare more than two ANOVA Kruskal-Wallis

samples for equality of means Chi square test
ADHESIVE REMNANT INDEX
ARI Value Shear Bond strength

Group I Group Group Group Group
II A1 II A2 III III B2
B1
0
No adhesive left on the tooth 2 3 1 0 2
surface
1
Less than half of the adhesive left 3 1 4 2 1
on the tooth surface
2
More than half of the adhesive 1 1 2 1 3
left on the tooth surface
3
Entire adhesive left on the tooth 4 5 3 7 4
surface www.indiandentalacademy.com
WILCOXAN RANK TEST
(SIGNED RANK AND RANK SUM)
THESE TESTS ARE NON-PARAMETRIC
EQUIVALENT OF STUDENT “t” TESTS. WILCOXAN
SIGNED RANK IS USED FOR PAIRED DATA AND
WILCOXAN RANK SUM IS USED IN CASE OF
UNPAIRED DATA.
KRUSKAL-WALLIS AND FRIEDMAN
THESE ARE SIMILAR TO PARAMETRIC ANOVA
TESTS. KRUSKAL-WALLIS IS USED FOR ONE
WAY ANALYSIS OF VARIANCE AND
FRIEDMAN IS FOR TWO WAY ANALYSIS OF
VARIANCE.
SPEARMAN’S RANK CORRELATION
SPEARMAN’S RANK CORRELATION AND
KENDALL’S RANK CORRELATION ARE THE
NON-PARAMETRIC EQUIVALENTS OF
CORRELATION COEFFICIENT TEST.
CHI SQUARE TEST (2 TEST)
THIS TEST IS A “ GOODNESS OF FIT” TEST, USED

TO FIND OUT THE ASSOCIATION BETWEEN
VARIABLES.THIS TEST IS USEFUL IN VARIOUS
SITUATIONS WHERE PROPORTIONS OR
PERCENTAGES OF TWO GROUPS ARE COMPARED
e.g. PROPORTIONS OF DIED AND SURVIVED IN
TREATED AND UNTREATED CHILDREN WITH
DIARRHOEA CAN BE COMPARED.
DISCRIMINANT FUNCTION ANALYSIS
IT IS USED TO CLASSIFY CASES INTO THE

VALUES OF A CATEGORICAL DEPENDENT,
USUALLY A DICHOTOMY.IF DISCRIMINANT
FUNCTION ANALYSIS IS EFFECTIVE FOR A SET
OF DATA, THE CLASSIFICATION TABLE OF
CORRECT AND INCORRECT ESTIMATES WILL
YIELD A HIGH PERCENTAGE CORRECT.
META ANALYSIS
GENE GLASS(1976) COINED THE TERM ‘META

ANALYSIS’.
THE TECHNIQUE OF META ANALYSIS INVOLVES

REVIEWING AND COMBINING THE RESULTS OF
VARIOUS PREVIOUS STUDIES. PROVIDEDTHE
STUDIES INVOLVED SIMILAR TREATMENTS, SIMILAR
SAMPLES, AND MEASURED SIMILAR OUTCOMES,
THIS CAN BE A USEFUL APPROACH.
CONTROLLED/UNCONTROLLED TRIALS
CLINICAL RESEARCH CAN INDEED HAVE CONTROLS.

PROVIDED THAT STUDIES ARE CONDUCTED ON A
PROSPECTIVE BASIS, CONTROLLED CLINICAL
STUDIES CAN BE QUITE POWERFUL.
UNCONTROLLED CLINICAL STUDIES ARE OF

QUESTIONABLE VALIDITY, WHETHER OR NOT THEY
ARE SUBJECTED TO STATISTICAL ANALYSIS.
SENSITIVITY, SPECIFICITY AND ROC
The sensitivity of a test is the probability that the

test is positive for those subjects who actually have
the disease. A perfect test will have a sensitivity of
100%. The sensitivity is also called the true positive
rate.
The specificity of a test is the probability that the
test is negative for those in whom the disease is
absent. A perfect test will have a specificity of I
100%. The specificity is also called the true negitive
rate.
SENSITIVITY, SPECIFICITY AND ROC…
TRUE DISEASE STATUS OR

TEST CHARACTERISTIC
RESULT DISEASE DISEASE TOTAL
PRESENT ABSENT
POSITIVE a ( 8) b (10) a +b=(18)
(+)
NEGATIVE c (20) d ( 62) c+d = (82)
(-)
TOTAL a +c = (28) b +d (72) N =100

SENSITIVITY, SPECIFICITY AND ROC…
YANCEY’S 10 RULES
-Evaluating Scientific literature
1. BE SKEPTICAL
2. LOOK FOR THE DATA
3. IDENTIFY THE TYPE OF STUDY
4. IDENTIFY THE POPULATION SAMPLED
5. DIFFERENTIATE BETWEEN DESCRIPTIVE AND

INFERENTIAL STATISTICS
JCO May 1997,307-314
YANCEY’S 10 RULES
-Evaluating Scientific literature
6. QUESTION THE VALIDITY OF DESCRIPTIVE

STATISTICS
7. QUESTION THE VALIDITY OF INFERENTIAL
STATISTICS
8. BE WEARY OF CORRELATION AND REGRESSION
ANALYSES
9. LOOK FOR THE INDICES OF PROBABLE MAGNITUDE
OF TREATMENT EFFECTS
10.DRAW YOUR OWN CONCLUSIONS.
JCO May 1997,307-314
SOFTWARES-STATISTICAL PACKAGES
SPSS
MINITAB
EPIINFO
MICROSOFT EXCEL
THANKYOU

Biostatistics in Orthodontics

Uploaded by

Copyright:

Available Formats

Biostatistics in Orthodontics

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Biostatistics in Orthodontics

Uploaded by

Copyright:

Available Formats

APPLICATION

STATISTICS AS A SINGULAR NOUN IS “A

WHERE AS PLURAL NOUN IT MEANS “FIGURES”

BIOSTATISTICS CAN BE DEFINED AS ART AND

SCIENCE OF COLLECTION, COMPILATION,

PRESENTATION, ANALYSIS AND LOGICAL

INTERPRETATION OF BIOLOGICAL DATA

AFFECTED BY MULTIPLICITY OF FACTORS

“An ounce of truth produces tons of statistics”

THE WORD STATISTIK IS DERIVED FROM AN

DURING THE OUTBREAK OF PLAGUE IN ENGLAND,

IN 1532 THEY STARTED PUBLISHING THE

WEEKLY DEATH STATISTICS.THIS PRACTICE

CONTINUED AND BY 1632, THESE BILLS OF

MORTALITY, LISTED BIRTHS AND DEATHS BY SEX

IN 1662, CAPT.JOHN GRAUNT USED 30

1. ENABLES US TO MAKE INTELLIGENT USE OF THE

STATISTIC LAWS ARE NOT EXACT LAWS LIKE

EX: WHEN WE SAY THAT THE AVERAGE HEIGHT

THEY CAN BE SEPERATED INTO TWO BROAD

95% C I for Mean

WHENEVER AN OBSERVATION IS MADE, IT WILL

EX: RECORDING THE SEX OF A PERSON IN A

IN EACH OF CASES A CERTAIN OBSERVATION IS

MADE FOR A CHARACTERISTIC AND THIS

CHARACTERISTICS VARIES FROM ONE

OBSERVATION TO OTHER OBSERVATION AND IS

II. DISCRETE / CONTINUOUS

STATISTICS CAN BE USED TO HELP THE READER

IT IS IMPORTANT THAT THE STATISTICAL

DATA ARE NOT NECESSARILY INFORMATION,

THE GOAL IS TO SUMMARISE AND PRESENT

•CHARTS AND DIAGRAMS

1. TABLE MUST BE NUMBERED

2. TITLE-BRIEF AND SELF EXPLANATORY –

3. THE HEADINGS OF COLUMNS AND ROWS MUST

4. THE DATA MUST BE PRESENTED ACCORDING TO

6. TABLE SHOULD NOT BE TOO LARGE.

Table-11Descriptive Statistics of Shear bond strength

LED 40sec 10 9.659 0.6158 0.1947 9.2184 10.09 8.34 10.7

4. CROSS SECTIONAL SURVEYS

SUBJECTS ARE DIVIDED INTO GROUPS

THE STUDY IS DESIGNED TO INVESTIGATE THE

ASSOCIATION BETWEEN A FACTOR AND A

DISEASE.THESE STUDIES ARE KNOWN AS

TROHOC STUDY. SINCE THESE FORM A

RETROSPECTIVE INVESTIGATION i.e. OPPOSITE

THESE ARE ALSO KNOWN AS EXPERIMENTAL

EX-ANIMAL EXPERIMENTS,ISOLATED TISSUE

•RANDOMIZED CONTROLLED TRIALS/CLINICAL

•FIELD TRIALS/COMMUNITY INTERVENTION

•COMMUNITY TRIALS-WITH COMMUNITIES AS

SAMPLING IS THE SELECTION OF THE PART OF AN

SAMPLE A FINITE SUBSET OF STATISTICAL

SAMPLE SIZE THE NUMBER OF INDIVIDUALS IN A

2.SUFFICIENTLY LARGE (TO MINIMIZE SAMPLING ERROR)

1. NON RANDOM SAMPLING

1. SIMPLE RANDOM SAMPLING- WITH OR WITHOUT REPLACEMENT

3. DEGREE OF VARIATION AMONG SUBJECTS