The 7 Questions or less to find the correct statistical method of analysis for your project.


Question Text
Result Text

_ _ _

'Dependent variable' Dependent variables are the variables your project aims to study. They are those variables you are interested in.
Also referd to as 'Outcome variables' or 'Target variables'.

Example: In a study to determine whether drinking milk daily in youth increases a person's lenght, 'length' would be the dependent variable of this study.
|
'Predictor variable' In experimental research this refers to the variable(s) that are 'manipulated' (e.g. medicine or placebo).
In observational research it is the variable that is hypothesized to cause or covary with the independent (outcome) variable.
Also refered to as 'independent variable'.

Example: In a study to determine whether drinking milk daily in youth increases a person's lenght, '(amount of) milk' would be the predictor variable of this study.
|
'Continuous variable' Continous refers to the type of scale on which the variable is measured.
Your variable is of continuous scale when your measurement can (theoretically) be split up to infinitely small amounts.
Also refered to as 'interval' or 'ratio' level.

Example: Length or distances in meters, centimeters, etc, time in minutes, seconds, etc. (However, if your design works by grouping these (e.g. short, medium and long) this is no longer regarded as continuous but as 'categorical')
|
'Categorical variable' Categorical (also known as 'nominal') refers to the type of scale on which the variable is measured.
Your variable is of categorical scale when your measurement are grouped/categorized. Often seen in qualitative research.
Ordinal (nominal with order/'hierarchy') should here be categorized under categorical level.

Example: People's nationality, political party, yes or no, lengths/sizes grouped to 'small', 'medium' and 'large', etc
|
'Assumptions for Parametric tests' The distribution of your data also has a role to play in determining which significance test you can use.
There are two major classes (parametric and non-parametric), in SPSS, mini-tab, matlab and most other tools there are standard procedures that can check which class you need by testing the 'assumptions for parametric tests'.

Example: If your data is not 'normally distributed', only test from the non-parametric class are reliable for your significance calculations. Which exact test is suited (from within the non-paramtric class is determined by the design of the study and the type of variables you used. This will also follow from this tool. However, there is not always a non-parametric test possible. This is the case when the asnwer 'no' is not available to choose.)





What does this little tool do?
This tool simply asks you a question (multiple choice) about your dataset or research design. Based on your answer a second questions is asked (again in multiple choice), this until are relevant information is gathered (maximum of 7 questions) about your data and research design to tell you which hypothesis test is the 'correct' one for your project.

Is it fast?
Yes. It is all JavaScript so each new question does not require a pageload. You load the page once, and that is all.

Which hypothesis test are 'included' in this tool?
"basic t-Test, "Biserial/Point-Biserial Correlation", "Mann-Whitney U Test", "Wilcoxon Matched-Pairs Test", "One Way independent ANOVA", "Kruskal-Wallis Test", "One Way Repeated Measures ANOVA", "Friedman’s ANOVA", "Pearson Correlation, "Regression", "Spearman Correlation, "Kendall’s Tau", "Independent Factorial ANOVA, "Multiple Regression", "Factorial Repeated Measures ANOVA", "Factorial Mixed ANOVA", "ANCOVA", "Pearson Chi-Square, "Likelihood Ratio", "Logistic Regression, "Loglinear Analysis", "MANOVA", "Factorial MANOVA", "MANCOVA"


© stevenpeutz.com