View
820
Download
5
Embed Size (px)
Citation preview
Learning outcomesAt the end of this workshop, you should be able toDescribe about population & sample, causation,
level of measurement, distribution of dataSummarise categorical & numerical dataUse appropriate statistical test for bi-variable
analyses using SPSS
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
2
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
3
We observe, we believe.What we observe might not be the
truth
Population vs. Sample 26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
4
Parameter vs. StatisticsParameter
characteristic of the whole population Statistics
characteristic of a sample, presumably measurable.
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
5
Statistics estimate parametersRepresentativeSampling errorDifferent samples yield different estimatesStatistics = Parameter if sampling done properlyHow to prove?
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
6
N = 35Red = 18/35 = 51.4%
N = 6Red = 3/6 = 50%
N = 6Red = 2/6 = 33.3%
N = 6Red = 4/6 = 66.7%
Average % Red = 50% + 33.3% + 66.7%3
= 50%
50% ≈ 51.4%Parameter
Statistics
Statistics Parameter
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
7
Variable & its roleA value and whose associated value may be
changed
DependentIndependent
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
8
CausationRelation of events (cause and effect)But correlation (between two events) does not
(always) imply causationRooster's crow does not cause the sun to riseSwitch does not cause the bulb to light
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
9
Time & causation
Time
Exposure Exposure
Exposure
Outcome
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
10
Time & causation (example)
Time
AgeExposure to
silica
Smoking
Lung Cancer
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
11
Causal Model
Slide 12
OutcomeExposure Exposure Exposure
Exposure
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
13
Type of data(Level of measurement)
Categorical
Nominal Ordinal
Numerical
Discrete Continuous
e.g. Gender, Race e.g. Cancer staging, Severity of CXR for PTB
e.g. Parity, Gravida
e.g. Hb, RBS, cholesterol.
Distribution (shape) of dataApplicable to numerical valueDiscrete or ContinuousDiscrete ~ Binomial, Poisson, Negative Binomial,
Hypergeometry, Multinomial etc.Continuous ~ Normal, t, chi-square, F etc.
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
14
Central limit theorem “Given a distribution with a mean μ and variance
σ², the sampling distribution of the mean approaches a normal distribution with a mean (μ) and a variance σ²/N as N, the sample size, increases” (David M. Lane)
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
15
Normal Distribution 26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
16
𝑓𝑓 𝑥𝑥; 𝜇𝜇,𝜎𝜎2 =1
𝜎𝜎 2𝜋𝜋𝑒𝑒−
12(𝑥𝑥−𝜇𝜇𝜎𝜎 )2)
Why Normal?- Because many biological & psychological variables are distributed normally
CharacteristicsBell shaped curveSymmetricalUnimodal
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
17
Test of Normality Anderson–Darling Test Corrected Kolmogorov–Smirnov Test (Lilliefors Test) Cramér–von-Mises Criterion D'agostino's K-squared Test Jarque–Bera Test Pearson's Chi-square Test Shapiro–Francia Shapiro–Wilk Test
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
18
Use Normality test with cautionSmall samples almost always pass a normality
test. Normality tests have little power to tell whether or not a small sample of data comes from a Gaussian distribution.
With large samples, minor deviations from normality may be flagged as statistically significant, even though small deviations from a normal distribution won’t affect the results of a t test or ANOVA.
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
19
Why run statistical test? 26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
20
1. Determine presence of difference (or similarity)2. Determine degree of difference3. Determine the direction of changes (trend)
Predict changes (outcomes)
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
21
A B C
Is there any difference between A & B?
How big is the difference between A & B?
Which one is taller? A or B?
Is C different from A & B?
Is there any pattern now?
If there will be D, can you predict how tall is D?
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
22
Statistical analysis
AnalyticalDescriptive
e.g. Describe socio-demographic characteristics -Age, Sex, Race etc.
e.g. Prevalence of hypertension.
e.g. Compare demographic characteristics between two population – Compare age between male & female
e.g. Distribution of gender by hypertension status
e.g. How demographic characteristics (more than one factors) explain hypertension
Univariable Bivariable Multivariable
IV DV IV DV IV DV IV
Descriptive StatisticsExplain one variable at one timeMethod based on type of measureCategorical
Frequency (Percentage)Numerical
Central measures (e.g. mean, median) & Dispersion (e.g. variance, standard deviation, range, min-max, interquartile range)
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
24
Data
CategoricalFrequency (count) &
Percentage
Numerical
Normal Mean (SD)
Not Normal Median (Range/IQR)
How to describe a data 26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
25
Comparing differenceA
B C
Which of the following shows true difference between two populations?
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
27
3 methods to compare values1. P-value2. Confidence interval3. Effect size
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s
28
Hypothesis testingTruth
Ho True Ho FalseTe
st
Do notreject Ho Correct Type II error
(β)
Reject Ho Type I error (α) Correct
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s
29Please revise the concept on your own
P valueP-value is ‘likely’ or ‘unlikely’ that Ho is true Taking 0.05 as the cut-off point (α), if P ≤ 0.05, it is
then ‘unlikely’ Ho is true, therefore reject Ho
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s
30
One-tailed vs. two-tailed Is there a difference between Hb 14 g% vs. Hb 12
g% in male & female respectively?Ho: HbM – HbF = 0H1: HbM = HbFH2: HbM > HbFH3: HbM < HbM
Note: Should be determined a priori
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
31
One-tailed vs. two-tailed
Two-sided Right-sidedLeft-sided
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
32
Hypothesis testingTruth
Ho True Ho False
Test
Do notreject Ho Correct Type II error
(β)
Reject Ho Type I error (α) Correct
P-value is the probability to make Type I error (based on frequentist inference)
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
33
The truth about P value Measures effectiveness (even by US FDA) < P means statistically significant, NOT clinical
significant But, be careful when interpreting P value P is affected BOTH by effectiveness AND sample size P can be < 0.05 even though the effectiveness is
marginal when sample size is huge Compare Ps between studies only appropriate if the
sampling & sample size is the same
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
35
P < 0.05Why 5%?Cut-off point proposed by Sir Ronald A. Fisher
1925 to reject or not to reject a hypothesis If P < 0.05 = Probability to make Type I error is
less than 5% If P > 0.05, > 5% of the difference occurred by
chance & not due to the TRUE difference
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
36
Hypothesis Testing using bivariable analysis Try to prove that Exposure causes the Disease
e.g. Smoking causing Lung CancerHo: No difference of risk to get Lung Cancer
between smoker and non-smoker
37
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
Lung Cancer No Lung Cancer
Smoking 20 (18.2%) 90 (81.8%)
Not Smoking 5 (4.5%) 105 (95.5%)
The occurrence of lung cancer is significantly higher (18.2%) among smokers compared to non-smokers (4.5%) (χ2 (df=1)= 10.15, P =0.001, OR = 4.7 (CI95% 1.7 – 13.0))
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
38
26 O
ctob
er 2
016
Confidence IntervalRange of plausible valuesNarrow interval high precision
Wide interval poor precisionHow narrow is narrow? And how wide is wide?
Base on your clinical judgment
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
40
26 O
ctob
er 2
016
Interpret single CI Compare with the null value
i.e. can be 0 for % or 1 for risk Compare with practical significance or the clinical
significance/indifference
41
Null
Null
A
B
Source: http://www.childrens-mercy.org/stats/journal/confidence.asp
Null
Null
C
D
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
Comparing multiple CIs 26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
42
A B C
1
2
1 12
2
Effect size The measure of effect irrespective of sample sizeCohen (1988) classify ES into Low (<0.3)Medium (0.3-0.7)Large (> 0.7)
Manual calculation or web based calculation
43
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
Statistical TestBivariable (univariate) ~ One dependent & one
independentMultivariate ~ Multiple dependent & multiple
independent variable
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
45
What test to use?Variable 1 Variable 2 Test
Categorical Categorical Chi-square
Categorical (2 pop) Numerical (Normal) Independent sample t-test
Categorical (2 pop) Numerical (Not Normal) Mann-Whitney U test
Categorical (> 2 pop) Numerical (Normal) One-way ANOVA
Categorical (> 2 pop) Numerical (Not Normal) Kruskal-Wallis test
Numerical (Normal) Numerical (Normal) Pearson Correlation Coefficient Test
Numerical (Normal/ Not Normal)
Numerical (Not Normal) Spearman Correlation Coefficient Test
Numerical (Normal) Numerical (Normal) –Paired
Paired t-test
Numerical (Not Normal) Numerical (Not Normal) –Paired
Wilcoxon Signed Rank Test
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
46
Bivariable Analyses Compare meansIndependent sample t-test (Unpaired t-test) ~ Two unrelated
meansPaired t-test ~ Two related meansOne-way ANOVA ~ More than 2 means
χ2 Test ~ Between categorical variables Non-parametric tests (Kruskall-Wallis, Man-Whitney U
tests) ~ If data is not normally distributed
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
47
Writing plan for statistical analysis #1
26 October 2016Basic Biostatistics (C) Jamalludin Ab Rahman 2015 48
Data were analyzed using the complex sample function of SPSS (version 13.0). Sampling errors were estimated using the primary sampling units and strata provided in the data set. Sampling weights were used to adjust for nonresponse bias and the oversampling of blacks, Mexican Americans, and the elderly in NHANES. The prevalence of hypertension, as well as the awareness, treatment, and control rates, were age adjusted by direct standardization to the US 2000 standard population.10 To analyze differences over time, the 2003–2004 data were compared with the 1999–2000 data. Estimates with a coefficient of variation >0.3 were considered unreliable. A 2-tailed P value <0.05 was considered statistically significant. (Ong et al. 2009)
Writing plan for statistical analysis #2
26 October 2016Basic Biostatistics (C) Jamalludin Ab Rahman 2015 49
To assess the effect of the selection process on the characteristics of the cases, we compared cases included in the final analysis to the rest of the cases. Since controls included in the present analysis were different from the rest of the diabetes free participants by design, no similar comparisons were performed for that group. To compare baseline characteristics of cases and controls appropriate univariate statistics were used. Similar binary logistic and multiple linear regression models were built with incident diabetes or HbA1c as respective outcomes and additive block entry of adiponectin and potential confounders. For linear regression CRP and triglycerides were log transformed. Since HbA1c could be modified by drug treatment, we ran a sensitivity analysis excluding all participants on antidiabetic medication. A p-value of <0.05 was considered significant. Analyses were performed with SPSS 14.0 for Windows.
Reporting analysis (example) 26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
50
51
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
26 O
ctob
er 2
016
Reporting analysis (example)
Summary1. Identify & define variables2. Type – independent vs. dependent3. Level of measurements – nominal, ordinal or
continuous4. Check distribution – Normal vs. Not Normal5. Decide what to do - descriptive vs. analytical
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
53
PracticalDownload data from
http://bit.ly/1RC6Zte
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
54
Chapter 2Introduction to SPSSIBM SPSS Statistics v21 for WindowsJamalludin Ab Rahman MD MPHDepartment of Community MedicineKulliyyahof Medicine
IBM SPSS Statistics
IBM Corporation Software Group Route 100 Somers, NY 10589Produced in the United States of AmericaMay 2012
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
56
Layout 26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
58
Main menu
Toolbar
Variables
Data editor 26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
59
Rows = variablesRows = each data
Define & describe your variables here
Enter your data here
Viewer 26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
60
The output of analyses will be displayed here.
Output is separated from
data
Syntax 26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
61
We can compile all the steps of the analyses here. Extend the programming
function in SPSS. Ability to perform complex steps e.g.
“looping”
Before even you start SPSS! 26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
63
You must identify & define relevant variables Define means
1. Name – preferably short single name, begins with alphabet, no special character, no space
2. Type of data – e.g. Numeric, Date, String3. Width & Decimal Places (if numeric)4. Label – description for the Name (will be displayed in Viewer)
5. Values – labels for value e.g. 1=Male, 2=Female6. Missing – define missing value e.g. 999 for N/A
Define your variables 26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
64
Variable Type 26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
66
Decide the suitable
variable type
For numeric, determine Width
& Decimal. Decimal < Width
For String, no option for
Decimal Place
New option for Numerics with leading zeros
Missing Value 26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
68
This question is Not Applicable
to male
e.g. Assign 999 to represent N/A value
& this won’t be included in any
analysis
Chapter 3Descriptive StatisticsIBM SPSS Statistics v21 for WindowsJamalludin Ab Rahman MD MPHDepartment of Community MedicineKulliyyahof Medicine
Exercise data 26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
70
HypotheticalStudy to describe factors related to HbA1c &
Homocystein (HCY)N=301 13 variables
Retrieve file information 26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
71
Task 1 26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
73
1. Describe socio-demographic characteristics of the respondent (age, gender & race)
2. Describe the explanatory variables1. Exercise2. smoking status3. BMI status &4. BP status
3. Describe HbA1c (taking cut-off for Poor HbA1c ≥ 6.5%) & HCY
Results 26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
77
Check for Normality.Is Age data distributed Normally?
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
78
Is this Normaldistribution?
Describe age 26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
79
Normal The subjects distributed between 23-67 years old
with the average of 34 (SD=8) years.If not Normal The subjects distributed between 23-67 years old
with the median of 33 (IQR=11) years
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
86
weight / ((height / 100) ** 2)
26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
88
Normal < 23Overweight 23 to < 27.5Obese >= 27.5
Chapter 4Bivariable analysesIBM SPSS Statistics v21 for WindowsJamalludin Ab Rahman MD MPHDepartment of Community MedicineKulliyyahof Medicine
To check association of two variables? 26
Oct
ober
201
6Ba
sic
Bios
tatis
tics
(C) J
amal
ludi
n Ab
Rah
man
201
5
91
HbA1cAge
The steps 26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
92
1. Determine which is dependant & which is independent
2. Determine level of measurements3. Determine Normality of the numerical
measurement4. Determine the suitable statistical test
What are the tests? 26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
93
Variable 1 Variable 2 Test
Categorical Categorical Chi-square
Categorical (2 pop) Numerical (Normal) Independent sample t-test
Categorical (2 pop) Numerical (Not Normal) Mann-Whitney U test
Categorical (> 2 pop) Numerical (Normal) One-way ANOVA
Categorical (> 2 pop) Numerical (Not Normal) Kruskal-Wallis test
Numerical (Normal) Numerical (Normal) Pearson Correlation Coefficient Test
Numerical (Normal/ Not Normal)
Numerical (Not Normal) Spearman Correlation Coefficient Test
Numerical (Normal) Numerical (Normal) – Paired Paired t-test
Numerical (Not Normal) Numerical (Not Normal) –Paired
Friedman test
Tasks 2 26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
94
1. Determine association between socio-demographic characteristics & all the risk factors with HbA1c
2. Determine association between socio-demographic characteristics & all the risk factors with HCY
Note: It would be good if you could construct dummy table for the answers even before the analyses started
Independent sample t-TestComparing Two Means
26 O
ctob
er 2
016
Basic Biostatistics (C) Jamalludin Ab Rahman 2015 96
Results 26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
99
Levene’s test check equality between variances. Ho: There is no difference of variances. So if P is significant, we reject Ho, and therefore equal variances assumed.
The original t-test (Student’s t-test) assumes equal variances for equal sample sizes. However if the variances are equal, it is robust for different sizes.
Welch's correction
Table – Distribution of age by blood pressure status 26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
100
N Mean SD Statistics df P
Normal BP 156 33.9 7.9 t=0.431 299 0.667
High BP 145 34.4 8.9
Chi-squared TestDifference of Two Proportions
26 O
ctob
er 2
016
Basic Biostatistics (C) Jamalludin Ab Rahman 2015 101
Results 26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
104
Describe this table first. What is your impression? 49% women vs. 47% men with high BP
Some books may suggest the use of Continuity Correction at ALL time, but recent simulations showed that CC (or Yate’s correction) is OVERCONSERVATIVE. Hence, use Pearson χ2 when < 20% of cells have expected count < 5
When ≥ 20% of cells have EC < 5, use Fisher’s Exact Test
This is given because we code the variables using numbers. Can be used to measure P-trend
One-way ANOVAComparing More than Two Means
26 O
ctob
er 2
016
Basic Biostatistics (C) Jamalludin Ab Rahman 2015 105
Results 26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
108
Describe these results first. What is your impression? HbA1c between races? 6.4 (SD 2.1) vs. 6.7 (SD 2.2) vs. 6.5 (SD 2.2)
The F test shows that there is no single significant difference between any two groups
Results – BMI status vs. HbA1c 26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
109
The F test shows that at least there is ONE pair with significant different. Either N vs. OW, N vs. OB or OW vs. OB
We need to run Post-hoc test to determine which of the PAIR is significant.
To decide which Post-hoc test to choose, we have to test for equality of variances i.e. Homogeneity of variances (Levene’s test)
Results – Post hoc 26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
111
The significant difference is only for Normal vs. Obese (P=0.002)
Report 26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
112
There is a significant association between BMI Status and HbA1c (F(2,298)=13.129, P<0.001). Post-hoc test showed that Obese subjects have significantly higher HbA1c compared to Normal and Overweight subjects (P=0.001 and P < 0.001 respectively).
Mann-whitney UNon Parametric Tests
26 O
ctob
er 2
016
Basic Biostatistics (C) Jamalludin Ab Rahman 2015 113
Results 26 O
ctob
er 2
016
Basi
c Bi
osta
tistic
s (C
) Jam
allu
din
Ab R
ahm
an 2
015
115
This ranks table is not to be cited in the research paper. Instead, describe their MEDIAN
KRUSKALL WALLISNon-Parametric TESTS
26 O
ctob
er 2
016
Basic Biostatistics (C) Jamalludin Ab Rahman 2015 116
Correlation TestRELATIONSHIP OF TWO NUMERICAL DATA
26 O
ctob
er 2
016
Basic Biostatistics (C) Jamalludin Ab Rahman 2015 119