Parametric vs. Nonparametric Flashcards
Mann - Whitney U
Ordinal data, but not limited to Compares two independent groups (Medians) Parametric Counterpart = 2 (Independent) Samples t-test
Wilcoxon signed rank
Ordinal data Compares 1 median to a specified value / Para counterpart = z-test, 1-sample t-test Compares 2 dependent (paired) medians/ Para counterpart = Paired samples t-test
Kruskal-Wallis
Ordinal data Compares 3 or more medians, 1 variable Parametric Counterpart = 1-way ANOVA
Friedman
Compares 3 or more medians, 2 variables Parametric Counterpart = 2-way ANOVA
Chi-Square Test of Independence
Tests 2 categorical variables for independence (lack of association) No parametric counterpart
Primary citation for non-parametric test info
Martella, Nelson, Marchand-Martella (1999)
Three general reasons to use a non-parametric test
- Area of study is better represented by the median 2. You have a very small sample size 3. You have ordinal data, ranked data, or outliers that you can’t remove. 4. Data is not normal
Reasons to use a parametric test.
- We have continuous, “normally” distributed data 2. Usually have more statistical power
General rule of thumb for parametrics
It is recommend that parametric tests be used even in those cases in which data do not meet the assumptions of normal distribution or homogeneity of variance but are derived from interval or ratio scores. Martella, Nelson, Marchand-Martella (1999)
Non-parametric vs. Parametric Table
Ordinal data, but not limited to Compares two independent groups (Medians) Parametric Counterpart = 2 (Independent) Samples t-test
Mann - Whitney U
Ordinal data Compares 1 median to a specified value / Para counterpart = z-test, 1-sample t-test Compares 2 dependent (paired) medians/ Para counterpart = Paired samples t-test
Wilcoxon signed rank
Ordinal data Compares 3 or more medians, 1 variable Parametric Counterpart = 1-way ANOVA
Kruskal-Wallis
Compares 3 or more medians, 2 variables Parametric Counterpart = 2-way ANOVA
Friedman
Tests 2 categorical variables for independence (lack of association) No parametric counterpart
Chi-Square Test of Independence
Martella, Nelson, Marchand-Martella (1999)
Primary citation for non-parametric test info
- Area of study is better represented by the median 2. You have a very small sample size 3. You have ordinal data, ranked data, or outliers that you can’t remove. 4. Data is not normal
Three general reasons to use a non-parametric test
- We have continuous, “normally” distributed data 2. Usually have more statistical power
Reasons to use a parametric test.
It is recommend that parametric tests be used even in those cases in which data do not meet the assumptions of normal distribution or homogeneity of variance but are derived from interval or ratio scores. Martella, Nelson, Marchand-Martella (1999)
General rule of thumb for parametrics
Non-parametric vs. Parametric Table
Nominal
A Nominal Scale is a measurement scale in which numbers serve as “tags” or “labels” only, to identify or classify an object. A nominal scale measurement normally deals only with non-numeric (quantitative) variables or where numbers have no value.
Ordinal
the level of measurement that reports the ranking and ordering of the data without actually establishing the degree of variation between them.
quantitative data which have naturally occurring orders and the difference between is unknown. It can be named, grouped and also ranked
EX: Likert scales
Interval
numeric scales in which we know both the order and the exact differences between the values.
Without a true zero, it is impossible to compute ratios. With interval data, we can add and subtract, but cannot multiply or divide.
Celsius example
10 degrees plus 10 degress celsius equals 20 degrees C but that doesn’t mean 10 is twice as hot. 10C = 50 F, 20 C = 68F
Ratio
the ultimate nirvana when it comes to data measurement scales because they tell us about the order, they tell us the exact value between units, AND they also have an absolute zero–which allows for a wide range of both descriptive and inferential statistics to be applied.
