Transcript Document

Chapter 10
Copyright Kaplan University 2009
The drawing of
conclusions by the use
of quantitative or
qualitative information

Inductive

Finding valid
answers from
examination of the
data

Deductive

Finding valid
answers using
mathematic
applications against
the data

Inductive

Studying relationship
between two types of
data

Deductive

Proving or
disproving of a
hypothesis
 Ex: The rate of
hypertension among
smokers
 Ex: Smoking causes
high blood pressure


An equation may be written using the same
formula and have different applications in
math or statistics
Consider the equation: y = mx + b


Mathematically: Formula for defining a straight line
in geometry
Statistically: Formula for simple regression analysis

Null Hypothesis

States there is no
difference between the
means of the two
compared groups
being studied

Alternative Hypothesis

States that there is a
true difference
between the means of
the two compared
groups being studied

This list simplifies the steps to testing a null
hypothesis for Statistical Significance


Generally p (false positive) = .05 or 5%
Leads to a 95% confidence interval of arriving
at the right hypothesis

Obtain a “p” value for the data

Example:
 Standard Deviation
 Confidence Intervals
 Mean, Mode and Median
 Student t-test (1 or 2 tailed)


Compare the p values (answers) to the alpha
level
Does this answer satisfy the null or alternative
hypothesis?

H0: Men who eat pizza three times a week will
gain ten pounds over the three week period
(Null Hypothesis)

H1: Men who eat pizza three times a week will
not gain ten pounds over the three week period
(Alternative Hypothesis)












Subjects
1
2
3
4
5
6
7
8
9
10
Mean
PRE Wt
128
100
110
145
201
200
198
157
300
194
173
POST WT
138
110
120
41
215
201
196
157
289
195
176
Difference
10
10
10
-4
4
1
-2
0
-11
1


Do you accept or reject the null hypothesis
Do you accept or reject the alternative
hypothesis



Most commonly used statistical test in
medicine
Compares means of the variables of two
research samples (groups)
May be used in research groups which differ


i.e. Male/Female; dogs/cats
May be 1 or 2 tailed (which affects
interpretation)

In regards to the “t” test and the use of “p”

If “t” is large (means of samples) then “p” is small
(percentage of error) and the difference is regarded
as real (i.e. believable)

If “p” is large (larger than 5%) then the difference is
not real (unrealistic or unbelievable)
INTRODUCTION TO
PREVENTATIVE MEDICINE
Promotes general health
 Prevention of disease
Application of epidemiological concepts




Aid in prevention
Aid in promotion

A state of complete physical, mental and social
well-being, not merely the absence of disease or
infirmity
-World Health Organization

Good

Known as Eustress

Bad

Known as distress
 Exercise
 Mal-adaption
 Infant stimulation
 Environmental



Mortality Data
Life Expectancy
Quality of Adjusted Life Years (QALY)

Latent:



Symptomatic:


Also known as: “hidden”
Offers a window of opportunity for early detection
Produces clinical manifestations that are measurable
Tertiary:

Disease progression in the absence of intervention

Primary:



Secondary:



Eliminate the cause of disease
Example: Vaccinations
Interrupt the disease process prior to symptoms
occuring
Example: Medication or Surgical intervention
Tertiary:


Limiting physical and social consequences of
symptomatic disease
Example: Rehabilitation/therapy

Nutritional Factors – How important is this
factor?
How can nutritional issues be addressed within
the scope of preventive medicine.
What is the difference between Environmental
and Occupational health promotion?
Explore routes of exposure to environmental
hazards. How dangerous are these?
What are some sources?

Behavioral factors: How do we promote
prevention here. Someone cite an example and
let us discuss briefly?
HS 310: Epidemiology and Statistics
Copyright Kaplan University 2009

Sample Size:
Used to determine time and amount of funding
needed for research
 Directly affects presence of statistical significance
 Defines the realism of the proposed research







Need for paired data
Will there be large/small variance in variables
of interest?
Consideration of Beta and/or Alpha Errors
Acceptance of 95% Confidence Interval/5%
Error
1 sided or 2-sided t-test
Degree of difference desired

Calculation of Paired t-test w/Alpha Error only
Formula: N = (zx)2 . (s)2
(d)2
Plug in the #: N = (1.96)2 . (15) 2
(10) 2
Work from Center: N = (3.84) (225)
100
Can you solve from here, what is the answer?
8.64 = “9”

Consider the differences in the equation
N = (zx)2 . 2 .(s)2
(d)2
Again work from center:
N = (1.96)2 . (2) . (15)2
(10)2
Can you solve for “N”
17.28 or 18

Utilizing Page 200 again, Box 12-2 for the
numbers
N = (zx + zb) 2 . (2) . (s)2
(d) 2
N = (1.96 + 0.84) 2 . (2) . (15) 2
(10) 2
Solve for “N”
35.28 (nope)
70.56 (NOPE)
72
(remember you have to have the same number in
both categories so even though 70.56 is
numerically correct you cannot divide 71 into 2
even groups.)
Also 36(2) = 72

A method of assigning subjects to the control or
experimental group in such a way that the
choice is in no way influenced.

Example of ways to randomize:
 Can you think of some?

Simple Random
Allocation

Use of random
numbers table

Randomization into
groups of “2”

Used to get 2 groups
with same number of
participants

Systematic Allocation

Assign 1st person to
group 1, second
person to group 2
and so on.

Stratified Allocation

Used in clinical
research whereas
patients are assigned
to certain groups
according to severity
of their condition



Independence Rule: One probability is not
influenced by the outcome of another
probability
Product Rule: determination that the
probability of two things being true
Addition Rule: Determination that the
probability of one thing being true under all
possibilities.

Multivariable Statistics


Involves more than one variable
These variables are called “multivariable models”




Determination of interactions between
variables
To develop prediction models in clinical
settings
Adjust inter-group differences
Useful in propensity matching and scoring

ANOVA – Analysis of Variance


Definition: Use to analyze results of experimental
studies or categorical independent variables
Two types: 1-way ANOVA (aka F-Test)
 Comparison of more than two means simultaneously
 Involves estimating the independent variance in one of
two ways



Type I = Between groups
Type II = within groups
N-way ANOVA


Aka 2-way ANOVA
Testing of two or more independent variables

ANCOVA: Analysis of Covariable


Definition: Method of analyzing continuous
dependent variables
MLR: Multiple Linear Regression



Definition: Method of analyzing dependent variables
and all independent variables which are continous
Most common is the “stepwise linear regression”
Not used much in clinical medicine