Question
click below
click below
Question
Normal Size Small Size show me how
BCPS study guide
biostatistics
Question | Answer |
---|---|
type of data: gender, group members, categories that are mutually exclusive with no order. qualitative non-numerical data. | Nominal - think "n for naming" |
type of data: categorical with order. used to measure attitudes or perceptions such as a likert scale or pain scale but the difference between is arbitrary. numbers only establish order but not amounts. | Ordinal data |
type of data: ordered and equal distance between values. zero is arbitrary and is simply used as any number rather than the lack of value. can have positive and negative values. tells distance between values but cant do ratios. | Interval |
type of data: ordered and equal distance between values but with absolute zero. | Ratio |
mean can be used with what types of data | interval and ratio |
median can be used with what types of data | ordinal and continuous (interval and ratio) |
mode can be used with what types of data | all. only one that can be used for nominal |
one standard deviation in percent, two and three | 68.26%, 95%, 99% respectively |
when does the mean equal the median | in normally distributed data |
what are the examples of tests for Nominal Data | for independent samples the Chi square or (Fisher exact test). if three or groups more you have to do a contingency table. for dependent samples use McNemar test. if more than 3 groups you use Cochran Q. |
what are the examples of tests for Ordinal Data | for independent samples do Mann Whitney U or Wilcoxon rank sum. if 3+ groups use Kuskal Wallis. for dependent samples use wilcoxon signed rank or Friedman if 3+ groups |
what are the examples of tests for Continuous Data | for independent samples do student t. for 3+ groups use ANOVA for dependent samples do paired student t or ANOVA for 3+ groups |
examples of parametric tests - for normally distributed data | student t tests, ANOVA, ANCOVA, Pearson correlation, Linear regression |
examples of non-parametric tests - for non-normally distributed data | Wilcoxan Rank Sum, Mann Whitney U, Kruskal-Wallace, Friedman test. |