1
Validity
: A valid study has no logic, sampling, or
measurement errors.
◦
Logic
The question that will be asked of the data.
◦
Selection or sampling
Determine whether to use a sample or a census.
◦
Measurement
How to measure the characteristic of interest.
Variable type: categorical (qualitative) or numeric
(quantitative)
Why are the data needed?
What will the data be used for?
What questions are going to be asked of the data?
Are the patterns of the past going to be repeated in the
future?
Census versus sample
Nonrandom methods
Simple random sampling
Stratified sampling
Systematic or sequential sampling
Cluster or area sampling
Sample size
•
Precision

How precise should the measurements be?
Accuracy
◦
Does the measurement measure what we want it
to measure (i.e., say = do)?
Reliability
◦
Would the measurement be the same if we
repeated it?
This preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
2
Mapping
Visual representations of data
Histograms and Pareto charts
St
l t
d t
l t
Stem plots, dot plots
Box (and whisker) plots
Normal probability plots
Length of Hospital Stay
Diagnosis Category
10
12
14
8
10
12
ncy
0
2
4
6
8
12
34
56
78
910
1112 1314 1516 1718
Length of Hospital Stay (days)
Frequency
0
2
4
6
Heart Disease
Delivery
Pnuemonia
Malignant Neoplasms
Psychoses
Fractures
Diagnosis
Frequen
Length of Hospital Stay
Days
18
15
12
9
6
3
Produced with Minitab® Statistical Software
Percentage of diabetic Medicare enrollees receiving eye exams
among 306 hospital referral regions (2001)
Source
: Wennberg, J. E. 2005. Data from the Dartmouth Atlas Project
Length of Hospital Stay
e Probability
1.00
.75
Observed Cumulative Probability
1.00
.75
.50
.25
0.00
Expected Cumulative
.50
.25
0.00
Produced with SPSS for Windows
Strong Negative Correlation
X
Y
r = 0 86
Strong Positive Correlation
X
Y
Microsoft Excel® screen shots
r = 0.86
r = 0.91
Positive Correlation
X
Y
r = 0.70
No Correlation
X
Y
r = 0.06