Yulia Sofiatin
Research scenario :
An investigator wanted to study
the effect of treatment A and B on BP
reduction in patients with
Hypertension
The investigator asked 4 GPs to collect
the data
Normal BP
Treatment A
High BP
Normal BP
Treatment B
High BP
Result
60 70 80 90 100
Within individual:
1 observer
2 observers
between visits
Among patients
Variability
is a natural characteristic of the
system that produces the population
of interest
is inherent in the population
cant be reduced through study or
measurement
What are the possible errors that
could occur during the study ?
- Manometer
Variability
- The GPs
What are the consequences of
variability?
random error X PRECISION
25%
40%
0%
SAMPLING VARIABILITY
Hypothetical study
Diet MI +
High cholesterol
population
Drug MI -
200 subjects
B
B
A
A
200 subjects
2000 subjects
SAMPLING VARIABILITY
Reducing random error
60 70 80 90 100
Measurement variability
How do we react to different pace?