Posted June 7, 2013
Overview of Intellectus Statistics: Level of Measurement
Nominal: In this level of measurement, numbers are used to categorize the data. For example, if gender is your variable, the responses will be male or female. A dichotomous nominal variable has only two categories.
Ordinal: Ordinal level variables have a meaningful order to them such as rank. For example there is an order to “drink size” (small, medium, large), however there is not a consistent distance between sizes.
Scale: Numeric variables that have equal intervals between each value, for example age.
When uploading a .sav file to the application all variables with ordinal level of measurement selected will default to scale. If you would like to change the level of measurement back to ordinal, you can do so using the drop down menu next to your variable name.
When uploading a .csv file to the application all variables with text data will be default to nominal and all variables with numeric data will default to scale. Ordinal level variables in .csv files should be in numeric format.
Level of Measurement and Your Analysis
When your data is uploaded to the Intellectus Statistics application you should confirm your levels of measurement are as follows for each analysis:
Pearson Correlation: All variables are scale level of measurement
One way ANOVA: Independent variable is nominal level of measurement; dependent variable is scale
Independent samples t-test: Independent variable is a dichotomous nominal; dependent variable is scale
Dependent samples t-test: both variables are scale
Chi-square: x variable (factor) is nominal; Y variable (factor) is nominal
Linear regression: Dependent variable is scale; predictor(s) scale or nominal
Mediation: All variables are scale level of measurement
Moderation: All variables are scale level of measurement
Reliability: All variables should be the same level of measurement with themselves (e.g., only ordinal variables used together)
Logistic regression: Independent variables are nominal or scale; dependent variable is dichotomous
Wilcoxon signed rank: Both variables are ordinal (or scale)
Mann Whitney U: Independent variable is dichotomous; dependent variable is ordinal (or scale)
Kruskal Wallis: Independent variable is nominal; dependent variable is ordinal (or scale)
Friedman Test: All variables are ordinal (or scale)
Repeated measures ANOVA: All variables are scale
One-within One between ANOVA: Independent variable is nominal; dependent variables are scale
ANCOVA: Independent variable is nominal; covariates are scale; dependent variable is scale