Correlation Flashcards
Where do associations exist?
Between:
Categorical variables.
Numerical variables.
Categorical and Numerical.
What can be an association between categorical variables?
Age and smoking.
What can be an association between numerical values?
Height and max expired lung volume.
What can be an association between categorical and Numerical values?
DBP and Gender.
How is the gender and DBP characterised in the association between them?
DBP: Response.
Gender: Factor.
What can a factor be?
Grouping:
Gender.
Smoking.
Something controllable in an experiment:
pH.
Drug concentration.
Temperature.
How is the factor called alternatively?
An independent variable.
What is the response?
The observation we make.
By what can the response be/not influenced?
The factor.
How is the response known alternatively?
The depended variable.
How can association be formally tested?
With Chi-Square tests.
Which id the factor and which the response?:
Are DBP, SBP different for females and males?
Is Height influenced by Gender?
Does smoking have an effect on Exreg, Exmax?
F: Gender R: DBP, SBP.
F: Gender R: Height.
F: Smoking R: Exreg, Exmax.
How do we evaluate an association?
By comparing histograms. Boxplots. Descriptive statistics. Formal methods. Correlation Coefficients.
How can we examine if smoking has an effect on Exmax?
Using Histograms:
1: No smokers –> Values on right = bigger lung volume.
2: Yes smokers –> Values on left = smaller lung volume.
X axis: lung volume.
Y axis: Frequency of smoking.
= Smoking is related with lung volume.
Using Boxplots:
Compare median values.
Using Descriptive Statistics: Report Smoker status. Number of smokers/not. Mean. Median. Range. Std. Deviation.
= Decrease in Exreg values of smokers.
T-test for means.
Mann-Whitney test for medians.
How can we find the association between 2 numerical values?
Making a table of data of variables.
Plot one variable against the other in a scatter plot.
Is there any relation between Height and Exmax?
Do a scatterplot.
= As height increases, Exmax increases
= Indicates a relationship.
What is association the same with?
Relationship.
Correlation.
Link.
What does a correlation describe?
How 2 variables vary together.
numerical vs numerical.
numerical vs categorical.
categorical vs categorical.
Where do we use scatterplots?
To view numerical vs numerical correlations.
How is the correlation of a graph with increased X and increased Y values characterised?
Positive.
How is the correlation of increased X values but decreased Y values on a graph characterised?
Negative.
How is the correlation of mixed variables random on a graph characterised?
No relationship.
Why can 2 variables vary?
Due to:
Changes in X which cause changes in Y.
A third variable independently influences X and Y.
A coincidence.
What is not a correlation?
A causation.
Why is a correlation not a causation?
Because it supports the argument for the causation but it does not prove the causation.
Which is an example of a correlation where X causes changes in Y?
Age with Height.
Which is an example where a third variable can influence X and Y?
Ice cream sales and No. of shark sightings.
Third variable = temperature.
Which is an example of a correlation being a coincidence?
Birth rate.
No of stock sightings.
For what is correlation a good method?
To explore a dataset.
