Scribd
Upload
12 views6 pages

Assignment Afework

Uploaded by

Afework Ketema
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
12 views6 pages

Assignment Afework

Uploaded by

Afework Ketema
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 6

Arba MinchUniversity

Analytical Epidemiology Assignment


Name: - Afework Ketema
I.D:- PRAS/O11/11
Part 1.
Purpose: to investigate the effect of various variables to predict milk yield expressed as milk120
from various factors (hoping that prediction will be valid, or meaningful, for wider population of
lactations and cows). The variables used are twin birth (designated twin), vaginal discharge
(designated vag-disch) and parity. Answer the following questions based on the analytic out put
given

. regress milk120 i.twin i.vag_disch parity

Source SS df MS Number of obs = 6610


F( 3, 6606) = 441.95
Model 591182245 3 197060748 Prob > F = 0.0000
Residual 2.9455e+09 6606 445888.49 R-squared = 0.1672
Adj R-squared = 0.1668
Total 3.5367e+09 6609 535137.178 Root MSE = 667.75

milk120 Coef. Std. Err. t P>|t| [95% Conf. Interval]

twin
yes -4.057658 57.65939 -0.07 0.944 -117.0887 108.9734

vag_disch
yes 97.81619 52.46494 1.86 0.062 -5.032036 200.6644
parity 196.1924 5.399211 36.34 0.000 185.6082 206.7766
_cons 2474.822 17.27616 143.25 0.000 2440.955 2508.689

Question 1. Which analytic method was used?


The outcome (milk yield) is continuous variable and there is more than one independent
variables (twin birth, vaginal discharge and parity) the analytic method used here is multiple
linear Regression.
Question 2. What is the sample size?
Suppose there is no missing value in the model the sample size (number of observation) is 6610
Question 3. What is the coefficient of determination or R2?
The coefficient of determination R² is 0.1672 x 100 = 16.72% of the proportion variations in the
dependent variable (milk yield) have been explained by the independent variables (twin birth,
vaginal discharge and parity) leaving 83.38% unexplained.
Question 4. Does twin birth have impact on milk yield? Explain how?
It has no impact because its p-value is p > 0.05 and the confidence interval includes zero. So
there is no enough evidence to reject the null hypothesis. The regression coefficient seen is due
to chance. That is, the presence of twin birth is not related or not an important factor in
predicting to changes in the milk yield.
Question 5. Does the presence of vaginal discharge have impact on the milk yield? Explain how?
Still vaginal discharge has no impact on milk yield because its p-value is p > 0.05 and the
confidence interval include zero. So there is no enough evidence to reject the null hypothesis.
The observed 97.8 units increase of milk yield 120 in presence of vaginal discharge was occurred
due to chance
Question 6. How do you explain the effect of parity on the milk yield?
It has significant effect on outcome i.e. milk yield because the p-value p < 0.05 so there is
sufficient evidence to reject the null hypothesis. The average value of milk yield (outcome)
increases by 196.7 units if there is a unit increase in parity holding other predictors (vaginal
discharge and twin birth) in the model constant.
Part 2
Objectives: to investigate the effect of infection with Manhemia hemolytica expressed as
seroconversion to its cytotoxins (designated as phcysc) on the occurrence of bovine respiratory
disease designated brd.

First the effect of phcysc on the brd was analyzed using appropriate statistical methods and the
result is given below.
Logistic regression Number of obs = 588
LR chi2(1) = 5.39
Prob > chi2 = 0.0202
Log likelihood = -372.26257 Pseudo R2 = 0.0072

brd Coef. Std. Err. z P>|z| [95% Conf. Interval]

1.phcysc .5238734 .2316111 2.26 0.024 .069924 .9778229


_cons -1.109662 .2105287 -5.27 0.000 -1.522291 -.6970334

7. Which analytic method was used?

The analytical method used was Binary logistic regression

8. What is the sample size?

Suppose there was no missing variables, sample size is usually the number of observations so it
was 588

9. What is the proportion of animals having brd in those which did not seroconvert to phcysc?

For proportion of animals having brd in those which did not seroconvert to phcysc.

Suppose convert =1 and do not=0

Log (odd of brd) = -1.109662 + 0.5238734(phcysc)

= -1.109662 + 0.5238734 (0)

= -1.109662

Odd = =

Proportion = ⁄

P = = = 24.8%
10.What is the proportion of animals having brd in those which seroconverted to phcysc?

The proportions of animals that have brd and did convert to cytotoxins are፡-

Suppose convert = 1 and do not = 0

Log (odd of brd) = -1.109662 + 0.5238734(phcysc)

= -1.109662 + 0.5238734 (1)

= -1.109662 + 0.5238734

0dd = = = 0.5566670131

P= = = = = 35.8%

Now infection with infectious bovine rihenotrheitis designated ibrsc is added to the model
containing phcysc and the result depicted here.

Logistic regression Number of obs = 588


LR chi2(2) = 25.19
Prob > chi2 = 0.0000
Log likelihood = -362.36504 Pseudo R2 = 0.0336

brd Coef. Std. Err. z P>|z| [95% Conf. Interval]

1.phcysc .6956164 .2391906 2.91 0.004 .2268115 1.164421


1.ibrsc .8076476 .1827452 4.42 0.000 .4494736 1.165822
_cons -1.591266 .2441081 -6.52 0.000 -2.069709 -1.112823

11.Is there evidence of confounding? Explain

Yes there is evidence for the presence of confounding. We have to check for the presence of
confounding if

Compound effect = ≥ 25%


Compound effect = = 33% which is the evidence for the presence of

confounding.

When we look at the data in the presence of ibrsc the effect of the Mh is changed that

shows its relationship to exposure. The seroconversion of Mahnemia hemolyticam to its


cytotoxins in the presence of ibrsc for the occurrence of bovine respiratory disease increased by
coefficient of 0.5238 to 0.6956 units. The association remains statistically significant when ibrsc
is added to the model.

The confounding variable is causes of the outcome independent of the exposure but also is
related to the exposure of interest, and might be treated as a confounding variable. When we see
ibrsc to brd because of belief that ibrsc could enhance the respiratory pathogenicity of other
unmeasured agents, besides Mh, and hence cause brd. To see the effect of exposure we have to
control the confounding.

We adjust and block the both path ways and leave the path way from exposure to outcome.

12. Draw the causal diagrams representing the association between the exposure of interest and
the candidate confounder in this case

Mh brd

Ibrsc
13. Which of the two variables is the predictor of brd? Explain

The ibrsc is the predictor to brd. From the data it appears that the strength of the association is
slightly increased in the presence of ibrsc. In addition, as we know respiratory disease caused by
ibrsc is associated with high morbidity but low mortality in susceptible animals but it is fatal in
the presence of especially M. haemolytica, or concurrent viral infection. The ibrsc could enhance
the respiratory pathogenicity of other unmeasured agents, besides Mh, and hence cause brd.

Controlling ibrsc will impact on the significance of coefficient depending on the strength of the
ibrsc effect on Mh and on brd. The R2 of the model should increase and this shows the the
variation in outcome is more explained by the model.

You might also like