Home 
Contact 
Profile 
Videos 
Presented in the Biostatistics module of the Clinical Research Coordinators Course on July 9, 2020 13.00-15.00 by Professor Omar Hasan Kasule MB ChB (MUK), MPH (Harvard), DrPH (Harvard) Professor of Epidemiology and Bioethics King Fahad Medical City
1.0 TYPES OF STATISTICAL ANALYSIS
• Univariate: compare a single mean or proportion against a fixed number
• Bivariate: compare two variables
• Multivariate: compare more than 2 variables
2.0 PURPOSE OF STATISTICAL ANALYSIS
• Testing the null hypothesis (H0): eg there is no difference in height between males and females
• Use the appropriate test statistic: Student T-test, F test, chi-square test
• Get a p-value and decide
• If p<0.05 reject the null hypothesis
• If p>0.05 do not reject the null hypothesis
3.0 STUDENT T-TEST (compare means in 2 groups)
• Dependent variable: continuous
• Independent variable: categorical
• Test statistic: Student T-test
• example: height (cm) by gender (male and female)
4.0 ANALYSIS OF VARIANCE, ANOVA (compare means in 3 groups or more)
• Dependent variable: continuous
• Independent variable: categorial
• Test statistic: F statistic
• Example: Weight (Kg) by color preference (blue, red, green, yellow)
5.0 CHISQUARE TEST (compare proportions in 2 groups)
• Dependent: categorical
• Independent: categorical
• Test statistic: Chi-square statistics
• Example: wearing glasses (yes, no) by gender (male, female)
6.0 2 x 2 Contingency table rows (male, female) columns (glasses+ and glasses)
7.0 CHISQUARE TEST (compare proportions in 3 or more)
• Dependent: categorical
• Independent: categorical
• Test statistic: Chi-square statistic.
• Example: wearing glasses (yes, no) by region of birth (East, North, West, Central, West)
• This test tells us there is some relationship among the variables but cannot tell what is related to what?
8.0 2 x 5 contingency table rows (glasses+, glasses-) columns (East, North, West, Central, West)
9.0 LINEAR REGRESSION (modeling data)
• Dependent (y): continuous
• Independent (x): continuous
• Test statistic: linear regression coefficient and t-test
• Example: weight by height and shoe size
10.0 LINEAR REGRESSION EQUATION
Weight = a + b (height)
a = intercept
b = regression coefficient
11.0 LOGISTIC REGRESSION (modeling data)
• Dependent(y): dichotomous
• Independent(x): all categorical, a mixture of categorical and continuous,
• Test statistic: logistic regression coefficient and t-test
• Example: glasses by gender and type of primary school and color preference
12.0 LOGISTIC REGRESSION EQUATION
Equation: Weight = a + b1 (height) + b2 (school) + b3 (color)
a = intercept
b1, b2 = regression coefficient
13.0 COX REGRESSION
• Dependent(y): dichotomous (varies with time)
• Independent(x): categorical and continuous
• Cox regression is used to analyze survival in cancer trials
• Example: death (yes/no) by type of time at death(day), treatment(new/old), gender(male/female), hemoglobin level (gm/cc)
14.0 GRAPH OF COX REGRESSION SHOWING SURVIVAL BYB TIME AND TREATMENT
15.0 LOG-LINEAR ANALYSIS
• Dependent: nominal
• Independent: categorical
• Example: gender (male/female) by wearing glasses (yes/no) and color preference (blue, red, green, yellow).