Basic concepts Flashcards
Why are statistical tests important?
- Inherent in process of doing quantitative research
- Essential in designing research and interpreting results
What is a statistical test?
- A procedure for addressing a problem with a specified sequence of steps
- Relies on assumptions
- Requires the formulation of a hypothesis
What is a hypothesis?
An initial idea of expected patterns
Null hypothesis
H0 - The default - always assumed to be correct
- There is no statistically significant relationship/pattern
Alternative hypothesis
H1 - There is a statistically significant relationship/pattern
What is a confidence level?
- The probability that H1 is correct
- e.g. 99% or 95%
How does a table of critical values work?
If the output is greater than the critical value, you reject the null
Why do we ‘fail to reject’ the null instead of ‘accepting’ the alternative?
Leaves room to re-test and improve
The area under a distribution curve…
Represents the probability of an outcome in that range
What is significance level?
How likely it is that the results are due to chance/The probability that the null hypothesis is correct
When do we accept H1?
If the significance level is sufficiently low (usually 0.05 or 0.01)
When is the critical value determined?
Before the test is carried out
What is the population?
The whole body of individuals we’re interested in
What is the sample?
A collection of individuals drawn from a population
- two samples from the same population are expected to have the same characteristics
What is a sampling strategy?
A method applied to choose a sample from a population
What are the main requirements in a sample?
- Large enough to ensure precision
- Careful consideration to avoid bias
What is accuracy?
A measure of correctness - how close the results are to the true value. This requires the sample to be representative of the population and measurements to be correct.
What is precision?
A measure of reproducibility - how well a result has been determined without reference to its closeness to the true value. This requires the sample to be large enough to compensate for variability within the population.
What is bias?
Over or underrepresentation of a particular characteristic of a population.
How might bias occur?
- Results from subtle aspects in our sampling strategy
- Depends on the target population
(e. g. door knocking)
What are the fundamentals of sampling?
- Every individual in the population has equal and independent probability of being included in the sample
- Inclusion of one individual does not affect the selection of another
What is random sampling?
- Random selection
- Often by assigning labels to a population
- e.g. random number generator
What is systematic sampling?
Selecting every nth individual from a population
Why might interviewing people ‘at random’ be problematic and how could it be improved?
- Might be influenced by conscious/subconscious bias
- We can’t test for or correct this so it is not acceptable for practice in academic studies
- Sample the nth person passing by (systematic) - the order of individuals is assumed to be random (unlike choosing every nth house)
What is assumed normal distribution?
- We assume samples and populations are approximately normally distributed
- This is where we would use parametric statistics
Ardoyne Road, Belfast…
There are patterns to the houses in that area based on what religion you are – you will get very different results from different parts of that area so you could systematically sample all protestant, or all catholic houses.
What is discrete data?
- Can only take particular values
- May potentially be an infinite number of those values, but each is distinct and there’s no grey area in between. - Can be numeric (like numbers of apples) but it can also be categorical (like red or blue, or male or female, or good or bad)
What is continuous data?
- Not restricted to defined separate values, but can occupy any value over a continuous range
- Between any two continuous data values there may be an infinite number of others
- Continuous data are always essentially numeric