Lecture 7: Data analysis Flashcards
What does the analysis you do depend on?
It depends on
- the data itself: how is it recorded & distributed?
- the research Q: will the analysis answer what I’m trying to ask?
What is categorical data?
Categorical data is data sub grouped by categories
It can be nominal or ordinal
How do you describe categorical data?
- show how often each category is seen
- counts or percentages
- tables and graphs
What is numerical data?
Quantitative data that can be discrete (whole number) or continuous (decimals)
Quantitative data that can be discrete (whole number) or continuous (decimals)
- can use mathematics to describe
- give a summary of the size of the values: mean, median
- give a summary of the spread of the values: variances, standard deviation, IQR
- report some sort of extreme: maximum, minimum, modal value
- consider the specific reason for comparing groups or populations
What can left skewed also be called?
Negative skewed
What can left skewed data not be modelled by?
Left skewed data can’t be modelled by a normal distribution
What are examination/test scores an example of?
Example of left/negative skewness
What do you need to consider when choosing a correlation test?
Need to consider the distribution of data
When would you use Pearson’s correlation?
Pearson’s = normally distributed data
When would you use Spearman’s rank?
Spearman’s = skewed data
What are the limitations of correlation?
- can’t comment on an exposure:outcome relationship, or make projections
- can only test for linear relationships
- can be an oversimplification - things look similar that are not
What can regression analysis do?
- can specify exposure and outcome
- can include non-linear relationships
- can specify multiple exposures
What does regression analysis allow you to quantify?
Allows you to quantify the relationship with the actual values and make projections of the expected values
