Cheatography

# Statistical Tests Cheat Sheet (DRAFT) by Robyn.jll

Overview of statistical tests an when to use them

This is a draft cheat sheet. It is a work in progress and is not finished yet.

### 1 Dependent Variable & 0 IVs (1 Popula­tion)

 DV Test interval & normal One-sample t-test tests if a sample mean differs sig. from a hypoth­esized value ordinal or interval One-sample median test tests if a sample median differs sig. from a hypoth­esized value categorial (2 catego­ries) Binominal Test tests if the proportion of successes on a two-level categorial dependent variable differs sig. from a hypoth­esized value categorial Chi-square goodne­ss-­of-fit tests if the observed propor­tions for a categorial variable differ from hypoth­esized propor­tions

### 1 DV & 1IV with 2 levels (indep­endent groups)

 DV Test interval & normal 2 indepe­ndent sample t-test compares the means of a normally distri­buted interval DV for two indepe­ndent groups ordinal or interval Wilcox­on-Mann Whitney test is a non-pa­ram­etric analog to the indepe­ndent samples t-test used, when you do not assume that the DV is a normally distri­buted interval variable categorial Chi-square test to see if there is a relati­onship between 2 categorial varibales assumes that each cell has an expected frequency of 5 or more categorial Fischer`s exact test same as Chi-square test, but can be used regardless of the expected frequency (expected frequency of 5 or less)

### 1 DV & 1IV with 2 or more levels (indep. groups)

 DV Test interval & normal One-Way ANOVA test for differ­ences in the means of the DV broken down by the levels of the IV used when categorial IV (with one or more catego­ries) an normally distri­buted interval DV ordinal or interval Kruskal Wallis test is non-pa­ram­etric version of ANOVA and a genera­lized form of the Mann-W­hitney test since it permits two or more groups categorial Chi-square test

### 1 DV & 1IV with 2 or more levels (indep. groups)

 DV Test interval & normal One-Way ANOVA test for differ­ences in the means of the DV broken down by the levels of the IV used when categorial IV (with one or more catego­ries) an normally distri­buted interval DV ordinal or interval Kruskal Wallis test is non-pa­ram­etric version of ANOVA and a genera­lized form of the Mann-W­hitney test since it permits two or more groups categorial Chi-square test

### 1 DV & 1IV with 2 (depen­den­t/m­atched groups)

 DV Test interval & normal Paired t-test used when you have two related observ­ations and want to see if the means on these two normally distri­buted interval variables differ from one another ordinal or interval Wilcoxon signed rank sum test is non-pa­ram­etric version of a paired sample t-test used, when you do not wish to assume that the difference between the two variables is the interval and normally distri­buted categorial McNemar test use if interested in the marginal freque­ncies of two binary outcomes

### 1 DV & 1 IV with 2 or m. lev. (dep./­matched g.)

 DV Test interval & normal One-Way repeated measures ANOVA is the equivalent of paired t-test, but allows for 2 or more levels of the categorial variable ordinal or interval Friedman test use when you have one within­-su­bjects IV with 2 or more levels and a DV that is not interval or normally distri­buted categorial (2 catego­ries) Repeated measures logistic regression use if you have a binary outcome measured repeatedly for each subject and wish to run a logistic regression that accounts for the effects of multiple measures from a single subject

### 1 DV & 2 or more IVs (indepen. groups)

 DV Test interval & normal factorial ANOVA use if you have 2 or more categorial IV and a single normally distri­buted interval DV ordinal or interval Ordered logistic regression used, when the DV is ordered, but not continuous categorial (2 categories Factorial logistic regression used, when you have 2 or more categorial IV but a dichot­omous DV

### 1 DV & 1 interval IV

 DV Test interval & normal Correl­ation used, when you want to see the relati­onship between two (or more) normally distri­buted interval variables interval & normal Simple linear regression allows us to look at the linear relati­onship between one normally distri­buted interval IV and one normally distri­buted interval DV ordinal or interval Non-pa­ram­etric correl­ation (Spearman) used, when one or both of the variables are not assumed to be normally distri­buted and interval the values of the variables are converted in ranks and then correlated categorial Simple logistic regression assumes that the outcome variable is binary

### 1 DV & 1 or m. interval IV/ 1 or m. categ. IVs

 DV Test interval & normal Multiple Regression similar to simple regres­sion, except that in multiple regression you have more that one IV in the equation interval & normal Analysis of Covariance like ANOVA, except in addition to the categorial IV you also have continuous IV categorial Multiple logistic regression like simple regres­sion, except that there are 2 or more IV IV can be dummy or interval variables, but cannot be categorial variables (if, should be coded into 1 or more dummy variables) categorial Discri­minant analysis used, when you have one or more normally distri­buted interval IV and a categorial DV is a multiv­ariate technique that considers the latent dimensions in the IV for predicting group membership in the categorial DV

### 2+ DV & 1 IV with 2 or more levels (indep. groups)

 DV Test interval & normal One-way MANOVA like ANOVA, except that there are 2 or more DV. there is one categorial IV and two or more DV interval & normal Multiv­ariate multiple linear regression used, when you have two or more DV that are to be predicted from two or more IV interval & normal Factor analysis is a form of explor­atory multiv­ariate analysis that is used to either reduce the number of variables in a model or to detect relati­onships amongst variables all variabales need to be interval and assumed to be normally distri­buted goal is to try to identify factors which underlie the variables

### 2 sets of 2+ DV & 0 IV

 DV Test interval & normal Canonical correl­ation is a multiv­ariate technique used to examine the relati­onship between two groups of variables for each set of variables, it creates latent variables and looks at the relati­onship among the latent variables assumes that all variables in the model are interval and normally distri­buted