Embed Code - If you would like this activity on your web page, copy the script below and paste it into your web page.

Normal Size Small Size show me how

Normal Size Small Size show me how

# Statistics

### Interpreting and Utilizing Clinical Statistics

Question | Answer |
---|---|

T:FA pooly designed study can not be statistically evaluated | False. It can be evaluated |

T:F, A poorly appied statistical test can never be evaluated | True. |

What type of statistics makes an assumption from a sample to a population | Inferential Statistics |

What type of stats summarizes the essential characteristics of a data set | Descriptive |

P-value and Confidence interval are used in what type of statistics | Inferential |

What are 2 examples of measurements that descriptive stats evaluates | central tendency and variability |

Average for a set of numbers | mean |

middlemost value ina set of ranked data | median |

most frequently occuring number in a data set | mode |

What are 3 measures of variability and be able to describe all 3 (page 16 of notes) | range, standard deviation, standard error of the mean (SEM) |

What measure of variability relates the sample to the population | SEM |

Does hypothesis testing establish "causality?" | yes |

Does hypothesis testing establish certainty? | No |

T:F, Causality and certainty are synonomous | False |

What does the null hypothesis state? | That the treatment being evaluated has no affect on the outcome of interest |

What type of error is described by "A false positive" | Type 1 |

What is a type 2 error | when you fasly conclude that there is no difference when in fact there really is. false negative |

What type of error is seen when you reject the null when you really should have accepted it | type 1 |

What is alpha | Risk of experiencing a type 1 error. it is the risk we are willing to take that we will find a chance result to be significant |

What is usually the accepted alpha value | 5% |

T:F, alpha needs to be established a priori, however beta does not have to be established a priori | Flase. Both need to be established a priori |

What is the risk we take at making a type 1 error | alpha |

What is the risk of making a type 2 error | Beta |

How would you calculate Beta | 1- power |

What is usually the normal designation for a value of beta | 20 % |

T:F, Hypothesis testing establishes whether the outcome of interest is due to chance alone or another factor | true |

What is another name for hpothesis of no difference | null hypothesis |

What are two ways to increase power | increase alpha (significance level) or increase sample size OR make assumptions (specifically one sided assumptions) or reduce variability) |

Which type of tailed test accounts for the possibility that the drug will be inferior to the control | 2 sided test |

What's one example of a downfall of a one-sided test | it only tests a benefit from a certain study, but does not relay information on a downfalls. Eg. a drug is beneficial will be seen, however it if is harmful it will not be seen |

LDL cholesteril is an example of what kind of measurement | continuous |

In a normal distribution what are the percentages of information found within 1, 2, and 3 standard deviations | 1 SD- 68.3%, 2 SD- 95.5%, 3SD- 99.7% |

What is the mean and SD in a normal distribution? | mean = 0, SD = 1 |

T:F, the central limit theorem states that as the sample size decreases and becomes sufficiently large, the distribution of the sample means tend toward normal distribution | False. This will happen as the sample size INCREASES |

If a study were to be conducted on whether placeo or drug A was better. A sample of 20 people were randomly assined. Results were p<.005 and confidence interval was 95%. Could you conclude that the values were normally distributed? | NO. normal distribution is determined by sample size. there has to be a sample size of at least 30 to be normally distributed |

What is the difference between parametric and nonparametric? | parametric = normall distributed, continuous data. Nonparametric = not nor. dist., nominal data or ordinal data |

What is the probability of the observed result or a more extreme result occurring by chance alone | p-value |

the p-value is the risk of experiencing what type of error, a fale positive or a fale negative? | false positive |

How would you interpret a p<.05. What does ttha mean? How would you describe that value in words? | It means that there is a less than 5% chance that what we observed was due to chance alone. |

If the p-value is to high, what should we do to the null hypothesis | you have insufficient evidence to reject the null = accept the null hypothesis |

T:F, The p-value is an important tool in determing clinical significance | FALSE! The p-value infers nothing about clinical significance! It relates to statistical significance! |

Is the p-calue determined post-hoc or a priori | post hoc |

T:F, the smaller the sample size the more narrow the CI | FaLSE, the larger the sample size the more narrow the confidence interval |

Do you want the CI to be small or large | small |

What value gives you statistical significance | p-value |

what value gives you clinical significance | CI |

What is the equation used to determine relative risk reduction? | (High - low)/ high |

What is the equation for absolute risk reduction? | hi-low |

what is the equation for number needed to treat | 1/absolute risk reduction |

What is the relative risk reduction? | the reduction of risk from 1 therapy to anoter. |

what is the aboslute risk reduction | absolute differences between the probabilities of the treatment even rate and control even rate |

How would you define NNT | the number if subjects needed to treat over a period of time in order to see the benefits of a therapy in 1 subject |

Which one of the following is used to make clinical decisions, NNT, ARR, RRR. | NNT |

What does the number needed to harm mean | the number needed to treat before you experience one adverse reaction |

What is the equation for NNH | 1-ARR |

What is the relationship between sample size and the ability to detect a difference? | Smaller sample sizes detect a large difference. large sample sizes detect small differences |

Created by:
ajasmine