Biostatistics POPH90013

Enes Makalic
 Staff – Coordinators

Dr Enes Makalic

High Dimensional Analytics (Unit Head)

Room 309, Level 3, 207 Bouverie Street
Centre for Epidemiology and Biostatistics
Melbourne School of Population and Global Health

Phone: +61 (3) 8344 0860

emakalic@unimelb.edu.au
 Suggested Reading

Essential Medical Statistics

Kirkwood BR, Sterne JAC.
2nd Edition, 2003
Wiley-Blackwell
 Statistics Software

• To purchase a Stata license:

– https://surveydesign.com.au/buystudent.html

• If you are only doing Biostatistics and will NOT be enrolling into other statistics subjects
– Stata/IC 16, 6 Month license (~$82)

• Biostatistics subjects in the future

– Stata/IC 16, Perpetual license (~$382)

• Stat/Transfer; Stata SE not required

• Stata Book: “Getting Started with Stata for Windows”

• Free, online version: http://www.stata.com/bookstore/getting-started-windows/
 Calculators

• The only University approved calculator is

– Casio FX82 (any suffix)

• Other models of calculators will NOT be permitted!

[Note – cannot use calculators on mobile phones in exam]

 Sessions (1 – 6)

Teaching
Lecture Chapter Practicals Date
Week
1 Introduction, datasets & types of variables 2 None March 4th

2 Displaying, presenting & graphing data 3 Tutorial March 11th

Measures of central tendency and spread, Tutorial March 18th

3 4
sampling variation & standard error Stata Practical* Thu/Fri 19th/20th

4 The normal distribution 5&6 Tutorial March 25th

Confidence interval for a mean, April 1st
5 7 Tutorial
comparison of two means Thu/Fri 2nd/3rd
Tutorial
6 Interpreting results of statistical analyses 8 April 8th
Stata Practical*

Teaching Break (Friday, 10th April - Sunday, 19th April)

 Sessions (7 – 12)

Teaching
Lecture Chapter Practicals Date
Week
7 Proportions & confidence intervals 15 & 16 Tutorial April 22nd

8 Measures of association 16 & 37 Tutorial April 29th

Controlling for confounding: stratification & Tutorial May 6th

9 18, 19 & 20
regression Stata Practical* Thu/Fri 7th/8th
10 Design, sample size & statistical power 35 Tutorial May 13th
Tutorial May 20th
11 Simple linear regression 10
Stata Practical* Thu/Fri 21st/22nd
12 Revision Tutorial May 27th
 Assessment

Date & Time Due Content

Assignment 1 Monday 20th April, 1pm All material covered in the
30% of the total mark lectures, tutorials and Stata
sessions in weeks 1 to 6.

Assignment 2 Monday 18th May, 1pm All material covered in the

40% of the total mark lectures, tutorials and Stata
sessions in weeks 7 to 10.

Examination June All material covered in

(2.0 hours, 15 minutes reading time) 30% of the total mark weeks 1 to 11.
Open book, basic calculator required.
 Extension requests

• Students may apply for an extension if…

– unforeseen circumstances arise that impair a student’s
ability to submit an assessment

• Extension
– Should be submitted before the assignment due date
– Maximum extension is 10 working days

• Policy and how to apply: Canvas, Assessment section

 Special consideration requests

• Students whose studies are genuinely and significantly

impacted by:
– Exceptional or extenuating circumstances outside of
their control
– Events or circumstances of national or state
significance

• Detailed information: Canvas, Assessment section

 Late submission

• A reduction of 5% for each day the work is late within the first
week

• Eight to fourteen days after the due date, 50% will be

deducted

• After 14 days no mark will be awarded

• If you have applied for an extension or special consideration

these penalties do not apply
 Submitting assignments in Canvas

• Submitting the correct assignment is your responsibility

• Submission of wrong file:

• Late assignment penalties apply
• Or treated as non-submission (0%) if the error is not found
prior to return of the assignments
 Academic Honesty & Plagiarism

• Information on definitions, advice & penalties

https://academichonesty.unimelb.edu.au/
 Biostatistics

• Statistical methods may seem very difficult

– But only from the first look

• In practice, they are…logical, intuitive and fun!

Lecture 1

Introduction, datasets and types of variables

 Introduction, datasets and types of variables

• To introduce using samples to

make inferences about populations
• Datasets, variables and
observations
• Types of variables
– Data types
– Outcomes and exposures

[See Kirkwood & Sterne: Chapter 2 – Defining the data]

 Inferential statistics

• Population
– Complete set of similar ‘items’

• Sample
– Subset (subgroup) drawn from the population

• Statistics
– Uses the sample to make inferences about the population
 Statistical Inference – Example 1 (Height)

All students enrolled in POPH90013

10 students enrolled in
POPH90013 attending
Lecture 1

POPULATION SAMPLE

Avg. Height = 173cm

INFERENCE
 Statistical Inference – Example 2 (Mobile phones)

UK Women aged 50 and over

791,710 women aged 50
and over in the UK who
participated in the Million
Women Study

POPULATION SAMPLE

During 7 years
follow-up, 51,680
(6.5%) incidence of
INFERENCE invasive cancers

Source: Benson V.S. et al. Int. Jrnl. of Epi. 2013;42:792-802

 Data sets, observations and variables

• We study samples by gathering data

– Collectively called a data set

• Data set is a table

– Records about individual members of the sample (rows)
• A variable is any observation that is measured (columns)
 Example data set

• To study the effects of vivax malaria infection during pregnancy

Source: Nosten F et al. Lancet 1999; 354:546-549

 Pregnant women & malaria dataset

ID Maternal age Birthweight Vivax malaria

(years) of baby ever during
(gms) pregnancy

1 17 2900 No
2 16 2600 No
3 19 2950 Yes
4 18 3200 No
5 19 2500 No
6 35 2900 No
 Types of variables

• Numerical
– Continuous
– Discrete

• Categorical
– Nominal
– Ordinal
 Numerical variables

• Continuous
– Can take any value
– Measured on a continuous scale
– Examples: Height, body weight, blood pressure,
haemoglobin level
 Numerical variables

• Discrete
– Can only take certain values
– Usually whole numbers
– Examples: Parity, number of visits to hospital, number
of people in a household
 Categorical variables

• Nominal
– No ‘natural’ ordering
– Binary variable – two values only
– Examples: male or female; alive or dead;
experimental groups; blood group O, A, B, or AB;
place of birth
 Categorical variables

• Ordinal
– Have a ‘natural ordering’ (ranking)
– Examples: social class, grade of breast cancer
• How do you rate the MSPGH statistics course?
– poor / fair / good / very good / excellent
 Derived variables

• To simplify display or analysis

– Often need to classify or transform variables

• Examples:
– Calculate or categorise body mass index, age groups

– Categorise by threshold values

• Low birthweight (<2.5 kg)
• Hypertension (Yes / No; diastolic blood pressure > 90mmHg)
 Outcome and exposure variables

• Outcomes are the variables we want to know more about

• Exposures are the variables we think might explain the

variation in outcomes

• Statistics quantifies the association between outcomes

and exposures
 Why is it important to know the variable type?

• Absolutely critical to choosing the appropriate form of

statistical analysis

• Type of the outcome variable

– Dictates the chosen statistical model
 Summary

• Population, sample

• Variables
– Numerical
• Continuous, discrete
– Categorical
• Nominal, ordinal
– Derived

• Outcome vs. Exposure