SAS Experience

Question 1

Research project

Answer 1

FluDB: collect flu seq
MAFFT: align seq
BEAST+Tracer: BSP
GFT: Flu estimates
BSP+GFT: linear interpolation 
SAS: proc arema to perform cross-correlaton to see at which time lag correlation highest.... hoping to see diversity increases before flu outbreaks

Question 2

STATS Homework T-test

Answer 2

Comparing vitamin users to non-users
Differences in mean age? Difference in mean cholesterol consumption?
Difference in mean fiber consumption?

o Recode variable to appropriate categories
o Use PROC UNIVARIATE to construct histogram to see if safe to assume normality
o Use PROC TTEST to conduct a two tailed t-test

Question 3

SAS final DATA

Answer 3

Data: Data from a study with 581 patients who had either acute viral or acute bacterial meningitis. Goal to determine whether pre-diagnostic measurements can aid in differential diagnosis (determining if patient has bacterial/viral meningitis)
o Variables: age, race, sex, WBC count, blood glucose, CSF protein, different kinds of white blood cells

Question 4

SAS final question

Answer 4

Create model that can be used to predict whether patient has bacterial/viral meningitis

Question 5

SAS Final steps

Answer 5

1) look at raw data
2) recode male and female and races into numeric variables
3) use PROC MI for multiple imputation of missing values
4) PROC CORR for summary stats and look at correlation matrix (multicolinearity = bad)
5) Logistic regression w/stepwise selection (binary outcome)
6) Interpreted odds ratios
7) Devised formula using parameter estimates that can predict types of meningitis

Question 6

SAS final results - how to interpret

Answer 6

if odds obtained from formula >1 then one can conclude bacterial meningitis
ABM was set at either 0 or 1 that is why we interpret in terms of ABM

Question 7

SAS final formula includes

Answer 7

WBC count, blood glucose, CSF glucose, polymorphonuclear leukocyte percentage of WBC in blood