More Related Content
More from Azmi Mohd Tamil (20)
How to run Simple Linear Regression on SPSS
- 5. ©drtamil@gmail.com 2016
What test to use?
Pearson’s Correlation done earlier showed
that there is a significant positive and fair
correlation between MOTHERS’ BODY
MASS INDEX (in kg) and BIRTH WEIGHT
(in kg)
Now model the relationship between
MOTHERS’ BODY MASS INDEX (in kg) and
BIRTH WEIGHT (in kg) by fitting a linear
equation to observed data.
y = a + bx
- 7. ©drtamil@gmail.com 2016
What test to use?
Birth weight – interval/continuous data.
Body Mass Index – interval/continuous
data.
The aim here to model the relationship
between MOTHERS’ BODY MASS INDEX
(in kg) and BIRTH WEIGHT (in kg) by
fitting a linear equation to observed data.
Assuming both BMI & birth weight are
normally distributed, most suitable test is
Simple Linear Regression.
- 11. ©drtamil@gmail.com 2016
Results
y = a + bx
Birth weight = 1.524 + 0.053 mBMI
For every increase of 1 unit of
mother’s BMI, the baby’s birth
weight increases 53 grams.
Equation valid since p<0.05
- 12. ©drtamil@gmail.com 2016
Conclusion
Birth weight = 1.524 + 0.053 mBMI
For every increase of 1 unit of mother’s
BMI, the baby’s birth weight increases 53
grams.
So if the mother’s BMI is 20, then the
birth weight would be around:
1.524 + 0.053 x 20 = 2.584 kg
If the mother’s BMI is 40, then the birth
weight would be around:
1.524 + 0.053 x 40 = 3.644 kg
r2 = 0.186, therefore 18.6% of the birth
weight variability is contributed by mBMI.
- 13. ©drtamil@gmail.com 2016
Exercise
Repeat the same statistical test between;
◦ Mothers’ Weight and Birth Weight
◦ You can also swap Weight for Age or Height but
take note of the p-values for a & b, some won’t
be significant, indicating that the linear
equation is not valid.