What should be ensured in the experimental design with regards to bias?
Any uncontrolled influences should not favour any condition over another systematically
What happens if a potential source of influence is not controlled?
It could vary systematically
In this case, the results become ambiguous as the outcome has more than one possible explanation
This uncontrolled factor is called the confounding variable
What is experimental control?
The idea that the conditions being investigated should differ only with respect to the factor under investigation
Attempts to eliminate or minimise bias by holding any potential sources of bias constant
What are the drawbacks of using a natural predictor variable?
Random assignment of participants to conditions is impossible
This means we must take great care in drawing causal conclusions
What is causal ambiguity?
When the causation or direction of causality is not known to have been influenced by the predictor
A third factor (extraneous variable) could be at play
What are descriptive statistics?
The use of graphs and summary stats to capture the basic features of data
What are the uses of descriptive statistics?
Communication of data to others (making it more comprehensible)
Estimation of unseen data (used to extrapolate to the general population)
A health check of the data
What are parameters?
The estimation, using summary statistics from sample data, of the corresponding values for the general population
What is a health check used for?
To ensure that the data do not contain misleading factors (this is especially important when using stats as estimators of parameters)
To ensure that the data meet any requirements of the model or statistical test employed
To identify outliers
What are outliers?
Data points that lie way out of line with the rest of the data points
Also known as anomalous results
How do you find the frequency of data?
Count up the number of times the experimental condition has been met (how many times the coin landed on heads)
How do you find the relative frequency of data?
Count up the number of times the experimental condition has been met before dividing by the total number of outcomes
BSC Psychology stats