Choosing Statistical Tests Flashcards
How do you decide which test to use? It depends upon the question you're asking and the data you have. [Q & D]. [These are inferential statistics because they involve hypothesis testing.]
Terms : Hide Images [1]
| 154335404 | correlation | Q: Is there a RELATIONSHIP between 2 factors? D: Continuous x and y variables, but variables are interchangeable with regard to predictor and response; investigating a potential association but NOT a cause-and-effect relationship | 1 | |
| 154335405 | linear regression | Q: Is there a RELATIONSHIP between 2 factors? D: Continuous predictor (x) and response (y) variables; investigating a potential cause-and-effect relationship | 2 | |
| 154335406 | chi-square test of association | Q: Is there a RELATIONSHIP between 2 factors? D: Categorical x and y (predictor and response) variables; may be investigating a potential cause-and-effect relationship or not. Setup varies with number of levels within each factor. Investigation of more than 2 factors also possible. | 3 | |
| 154335407 | t-test (single sample) | Q: Is there a DIFFERENCE between 2 groups? D: The 2 groups are categorical predictors, and response (y) data is continuous; investigating a potential difference between one sample and a known population | 4 | |
| 154335408 | unpaired t-test | Q: Is there a DIFFERENCE between 2 groups? D: The 2 groups are categorical predictors, and response (y) data is continuous; investigating a potential difference between two independent samples | 5 | |
| 154335409 | paired t-test | Q: Is there a DIFFERENCE between 2 groups? D: The 2 groups are categorical predictors, and response (y) data is continuous; investigating a potential difference between two related samples (e.g., before and after). | 6 | |
| 154335410 | Mann-Whitney U test | Q: Is there a DIFFERENCE between 2 groups? D: The 2 groups are categorical predictors, and response (y) data is continuous; investigating a potential difference between two independent samples. NON-PARAMETRIC: have converted continuous response data to rank data [analogous to unpaired t-test] | 7 | |
| 154335411 | Wilcoxon signed-rank test | Q: Is there a DIFFERENCE between 2 groups? D: The 2 groups are categorical predictors, and response (y) data is continuous; investigating a potential difference between two related samples (e.g., before and after). NON-PARAMETRIC: have converted continuous response data to rank data and retrieved difference signs (+ or -) [analogous to paired t-test] | 8 | |
| 154335412 | ANOVA (analysis of variance) | Q: Is there a DIFFERENCE between 3 or more groups? D: The groups are categorical predictors, and response (y) data is continuous. [specific type of ANOVA depends upon experimental design] | 9 | |
| 154335413 | one-way ANOVA | Q: Is there a DIFFERENCE between 3 or more groups? D: The groups are categorical predictors, and response (y) data is continuous; only one factor (also known as effect or treatment variable) is being tested, but it can have multiple levels (and typically has at least 3 - hence, "3 or more groups"). | 10 | |
| 154335414 | two-way ANOVA | Q: Is there a DIFFERENCE between 3 or more groups? D: The groups are categorical predictors, and response (y) data is continuous; two separate factors are being tested, each with multiple levels; replicates are required for each combination of levels within factors. | 11 | |
| 154335415 | three-way ANOVA (or higher multifactorial design) | Q: Is there a DIFFERENCE between 3 or more groups? D: The groups are categorical predictors, and response (y) data is continuous; three separate factors are being tested, each with multiple levels; replicates are required for each combination of levels within factors. Larger multifactorial ANOVAs can also be designed. | 12 |