L10 - Factors Affecting Choice Of Statistical Tests Including Levels Of Measurement & Design Flashcards
Descriptive statistics
Descriptive statistics can give summaries of data that we have collected from our research; and an indication of what the statistical analysis might reveal about our results.
What are levels of measurement and what are the three levels of measurement
Levels of Measurement are used to try to categorise our data into one of four types, so that we can correctly select the most appropriate statistical test to analyse our results, they are:
- nominal
- ordinal
- interval
Nominal data
- The data consists of the numbers of participants that might fall into different categories, and a person can be placed in one category only and not the other
- e.g no of males & females
Ordinal
- The data can be placed in rank order from lowest to highest. The ordinal scale can consist of measurements that are of unequal intervals e.g. 1.20, 1.25, 1.27.
- The data is concerned with the order that the data can be presented in
- The order from lowest to highest in terms of how quickly participants finished a 100 metre
swimming race, 1st, 2nd, 3rd etc.
Interval data
- The data has fixed and even intervals (and this differs from ordinal data). The units of data are fixed and have the same distance) throughout the range.
- Examples include height and weight which all have fixed intervals between each unit of measurement
Two types of test
- parametric vs non-parametric
Parametric vs non-parametric tests
- Parametric tests are more robust and powerful than non-parametric tests
- rely on the actual data collected rather than just examining the rank order of the data.
- Parametric tests are also more likely to detect if the data is significant or not.
- There are three factors that mean a parametric test can be conducted:
a) Interval level of measurement:
The data must be interval rather than ordinal in terms of level of measurement
b) Normal distribution:
The data collected should be taken from a population that shows a normal distribution curve rather than a skewed distribution
c) variance of data
The data should have similar variance or spread of scores, it can be examined by looking at the dispersion of the data & the standard deviations for both conditions & seeing if they are similar
Non-Parametric tests
- Chi squared
- Spearman’s Rho
- Mann Whitney
- Wilcoxon
Parametric tests
- Pearson’s r
- Related t-test
- Unrelated t-test
How do I decide what test to use?
Ask 3 questions
Q1
Does the research involve a correlation, test of difference, or an association?
- If using a correlation then you should use Spearman’s Rho or Pearson’s r
- If looking for a test of difference, then you should use one of these tests Mann Whitney, Chi Squared, Wilcoxon, Unrelated t-test, or Related t-test
- If looking for an association between variables then you would use Chi squared
Q2
Which research design is being used?
- Independent measures
- repeated measures
- matched participants design
Q3
which level of measurement is being used in the research
- nominal
- ordinal
- interval
What helps decide test to use?
Table (pictures)
Carrots Should Come Mashed With Swede Under Roast Potatoes