Types and Relationships of Causality
Translated by
ScribdTranslationsTypes and Relationships of Causality
Translated by
ScribdTranslationsINFERENCE
CAUSAL
TYPES OF CAUSALITY
SUFFICIENT CAUSE
When the factor is present,
the ending appears
The cause always precedes the
appearance of the outcome
Factor (+) Disease (+)
Genetic anomalies
homozygous (sickle cell anemia,
Tay-Sachs disease.
NECESSARY CAUSE
When the factor is absent, no
the outcome appears.
The cause must be present to
let the outcome occur
The present cause does not determine the
appearance of the outcome
Factor (–) Disease (-)
•e.g. in the absence of Mycobacterium
tuberculosis is not a disease, but
her mere presence does not determine that the
disease appears.
RISK FACTOR
Characteristic whose active presence in a
exposed group increases the probability of
observe the outcome in comparison with the
unexposed group
Factor (+) Disease Risk
Increased
•E.g. smoking habit and lung cancer.
Not all smokers have the
disease (is not sufficient cause) and the
disease appears in non-smokers (no
is a necessary cause), but the disease itself
is more commonly seen in smokers.
The association
between variables
does not imply that
a sea cause
from the other
TYPES OF RELATIONSHIPS
CAUSALS
DIRECTThe factor exerts its effect on
absence or without the participation of
other factors.
Factor Effect
•E.g. a strong impact on organs
internals cause injury without necessity
from other concurrent factors
TYPES OF RELATIONS
CAUSALITIES
[Link] or more factors
they have an effect on other factors
Factor 1 Efecto Factor 2
•Ej. la mejor educación se asocia
indirectly with better health,
promoting knowledge,
motivation to preserve it and
possibility of obtaining resources
TYPES OF RELATIONS
CAUSALS
NO [Link] relationship between
factors (variables) exist
statistically but not
casually. It may be due to:
chance
bias
incorrect temporal relation
influence of a confounding factor
CAUSAL RELATIONS
Relatives
• a. Based on random error
•b. Based on systematic error
•2. REALS
• a. Effect - cause
•b. Effect - effect
•[Link] - effect
The five situations that can occur when an observation is made in a sample
association between two variables (coffee consumption and myocardial infarction)
EXPLANATION TYPE OF ASSOCIATION BASEFORTHE REALITY IN THE CAUSAL MODEL
ASSOCIATION POPULATION
CHANCE
1. RANDOMERROR THECONSUMPTIONOFCOFFEEAND
APPARENT
THEHEARTATTACKNOSE
RELATE
SYSTEMATIC ERROR THECONSUMPTIONOFCOFFEEAND
[Link] APPARENT
THEHEARTATTACKNOSE
RELATED
[Link]-CAUSE REAL EFFECT BEFORE THE THEINFARCTIONCAUSE
Heart attack
CAUSE COFFEE CONSUMPTION
COFFEE CONSUMPTION
FACTOR X
4-EFFECT-EFFECT REAL DISTRACTORORVARIABLECOFFEECONSUMPTIONAND
INTERVENOR CONCURRENT INFARCT [Link] HEART ATTACK
FOR ANOTHER FACTOR
COFFEE CONSUMPTION
[Link]-EFFECT REAL CAUSE AND EFFECT THECONSUMPTIONOFCOFFEEIS
UNACAUSE OF HEART ATTACK
HEART ATTACK
HOW TO DISCARD AN ASSOCIATION
APPARENT CAUSAL (SPURIOUS)
Based on chance
Increase accuracy in the
measurements
Increase the sample size
Use probability samples
Write and apply a manual of
operations
Train and certify the observers
Consider the significance values
statistics
HOW TO DISPOSE OF AN ASSOCIATION
APPARENT CAUSAL (SPURIOUS)
Based on the bias
– Establish inclusion criteria and
exclusion
Calibrate the measuring instruments
Analyze the results in the context of
the previous evidence
-Take blind measurements
HOW TO DISMISS AN ASSOCIATION
CAUSAL REAL EFFECT – CAUSE
(Due to a timing misjudgment)
In the design/planning phase
Conduct longitudinal study (cohorts)
Use measured variables with
anticipation
In analysis phase
Evaluate the association in the context of
the prior evidence
Analyze the biological plausibility
COMO DESCARTAR UNA ASOCIACION
CAUSAL REAL EFFECT–EFFECT
(Due to a confounding factor)
•a. In design phase(measures
early and irreversible)
-i-Specify
-ii-Paired sampling
•b. In the analysis phase(measures
reversible and flexible
to stratify
–ii-Make statistical adjustment
–Analysis by scores
ADVANTAGES AND DISADVANTAGES OF THE
CONTROL MEASURES
OF CONFUSION VARIABLES
Measure a Phase Advantages Disadvantages
apply
Specification Planning-Easy to -Limita:
to understand the generalization
Limit the selection the sample size
of subjects regarding
from the PI
Advantages and disadvantages of measures
CONTROL OF CONFUSION VARIABLES
MatingPlanning-Remove the It's a slow, expensive process,
influence of less effective than increasing
factors: N
constitutional Early decision,
difficult to measure irreversible that can
Balance cases and change the analysis, and that
controls in each conditions the relationship
stratum variable
Facilitates the selection of independent/district contractor.
controls -Remove option a
evaluation of the factor
shown as a variable
Demand paired analysis
There is a risk of
overlapping
ADVANTAGES AND DISADVANTAGES OF MEASURES
CONTROL OF CONFUSION VARIABLES
Stratification Analysis - Easy to Limited by N for
to understand each stratum
-Flexible and *weathervariables
reversible fewstrata
Lascovariables
they must have been
measures
in advance
ADVANTAGES AND DISADVANTAGES OF CONTROL MEASURES
CONFUSION VARIABLES
Adjust Analysis-They can Themodelof
statistician control several interaction may not
factors to be appropriate
simultaneously estimate
Flexible and inaccurate of the force
reversible of the effect
Allvariables control
continues are incomplete of the
tools for adjusting factors of
confusion
The results are
moredifficultto
interpret
Lascovariables
they must have been
measures
in advance
CHARACTERISTICS THAT INCREASE THE
PROBABILITY OF CAUSALITY.
Principles of Political Economy (John Stuart Mill, 1856)
Strengthen the association
Consistency of the association
Specificity of the association
Biological plausibility of the
association
Dose-response relationship in the
association
CAusality in Epidemiology
(Bradford Hill, 1937)
Consistency(repeated observation in different
circumstances
Specificity(observation limited to certain
characteristics
Temporal relationshipthe cause precedes the occurrence of
effect
Biological gradient(dose–response relationship)
– Biological plausibility (varies with the state of the
knowledge
Coherence of the evidence(free from contradictions)