Inferential Statistics Flashcards
What do inferential statistics allow us to do?
They allow us to draw conclusions based on the probability that our results could have arisen by chance.
What information is needed to choose a statistical test?
Whether the study is a difference or correlation/association,
What experimental design was used,
What level of measurement was used.
Nominal data.
Most basic level of measurement.
Category data.
Used to label categories without any quantitative value.
Frequency data- tells us how many people in each group.
Discrete data- only one item in each category.
Example: hair color, nationalities, names of people
Ordinal data.
Intervals between each rank are not even.
Based on subjective opinion rather than objective measurement.
No clearly defined interval between the ranks.
Example: scale of 1-5, measuring economic status using the hierarchy: ‘wealthy’, ‘middle income’ or ‘poor.
Interval data.
Equal gap between each unit on the scale.
Precise.
Can go into minus values.
Can be obtained using measures of dispersion.
Example: Temperature
Ratio data.
Equal gap between each unit on the scale.
Precise.
Starts at zero.
Examples: income, height, weight
Level of significance.
We can never be 100% certain that our hypothesis is correct. There is the probability that it might be due to chance.
Example: 99.9% certain before accepting H1 and rejecting H0. This is allowing 0.1% possibility that the result is due to chance. Level of significance here is 0.001
Type 1 error.
Occur when level of significance is too lenient.
This gives us a false positive because we accept the hypothesis when actually the null hypothesis was correct.
Type 2 errors.
Occur when level of significance is too strict.
This gives us a false negative because we accept null hypothesis when actually the hypothesis was correct.
How to find critical value
Identify if the hypothesis is directional or non-directional.
What level of significance to use.
Number of data sets OR number of participants in first condition and number in second condition OR degrees of freedom
When is the sign test used.
Test of difference.
Repeated measures / matched pairs.
Nominal data.
Find critical value (number of data sets),
Observed value smaller than critical value = significant.
When is Mann-Whitney used.
Test of difference.
Independent measures.
Ordinal data.
Find critical value (number of participants in first condition and number in second condition).
observed smaller than critical = significant.
When is the Wilcoxon test used.
Test of difference.
Repeated measures / matched pairs.
Ordinal data.
Find critical value (number of data sets),
observed smaller than critical = significant.
When is Chi-Squared used.
Test of difference.
Independent design.
Nominal data.
Find critical value (degrees of freedom - rows -1 x columns -1)
Observed bigger than critical = significant.
when is Unrelated t-Test used.
Test of difference.
Independent design.
Interval / ratio data.
Find critical value (degrees of freedom - number of people in group 1 + number of people in group 2 - 2)
Observed bigger than critical = significant.
