Documente Academic
Documente Profesional
Documente Cultură
By
Ida Bagus Subanada
SPECIFIC LEARNING OUTCOME
Note: 1
P ( P1 P2 )
2
Example:
Journal A reported that the recovery rate of
drug 1 for a disease was 50% and drug 2
was 80%. If we shall conduct a replicative
study, it is wrong if we use 30% (80%-50%)
as the value of effect size. There are two
reason for these:
1. Conceptual reason
2. Technical reason
Example of no effect
X2 = 4.467; df = 1; P = 0.035
P-value interpretation:
If the effectiveness of “X” drug = standard drug, there is
0.035 (3.5%) of probability to get 15% or more of recovery
difference.
OR
Sample
No reject H0 beta (1– alpha)
………HT
1. Non-parametric:
* Fisher exact test, Chi-square, Mc Nemar
* Mann-Whitney, Kruskal-Wallis, Kendall’s r,
Wilcoxon, Friedman.
2. Parametric @:
* t test, Anova, Manova, Hotelling’s T2,
Correlasi Pearson
2. The degree of CI
99% CI >> 95% CI
Ex.: Mean difference of blood pressure was 5 (95% CI 3 to 7) mmHg
Mean difference of blood pressure was 5 (99% CI -0.5 to10.5)
mmHg
………CI
CI formula
CI = P ± (Zα x SE)
CI = confidence interval
P = point estimate = effect size
(e.g. mean difference,
difference of proportion, etc.)
Zα = normal standard deviate,
e.g. 95% CI ~ α = 0.05
Zα = 1.96
99% CI ~ α = 0.01
Zα = 2.58
SE = standard error
………CI
CALCULATION OF CI
Example:
• Mean difference of birth weight between
rural and urban of newborn babies was 500
gram, 95% CI 300 to 700, P= 0.002
statistically significant.
Example:
• OR to suffer from bronchiolitis acute for
babies who exposure to tobacco smoke is
1.29, 95% CI 0.82 to 2.01, P=0.302
not statistically significant.
Hypothesis Estimation
testing (p-value) (confidence
interval)
Population
How to present data with effect size, confidence
interval, and P-value ?
Example: