Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
55 Cards in this Set
- Front
- Back
Specify what a "95% C.I." is and define what it represents.
|
A C.I. is a confidence interval. For each group, it represents a range of plausible values containing true population parameters with a probability of 0.95.
It allows for inclusion of information on both magnitude and precision of effects. |
|
What does overlap between control and case(s) confidence intervals mean?
|
Overlap means possible or no difference in direct comparison.
|
|
What does no overlap of confidence intervals mean?
|
No overlap means there is a definite difference in direct comparison. Statistically significantly different.
|
|
What is a "p-value?"
|
A p-value represents evidence accumulated from a hypothesis test. It measures the probability of obtaining a test statistic of value equal to that of the sample or something more extreme if there is no effect/difference in the population.
|
|
What are the steps taken in a hypothesis test?
|
1. A hypothesis test (or significance test) requires specification of the null hypothesis (H0).
2. Calculation of the appropriate test statistic 3. Probabilistic assessment of the test statistic under H0 |
|
What measure is (often) used in a hypothesis test?
|
P-value
|
|
Reject H0 when p-value is less than
|
0.05
|
|
P-value less than 0,05 means
|
reject H0
|
|
P-value more than 0.05 means
|
accept H0
|
|
Accept H0 when p-value is greater than
|
0.05
|
|
When are Chi Squared tests used?
|
For categorical data
|
|
When are Fisher's Exact test used?
|
For categorical data with no more than 2 categories
|
|
What is a table of results with columns and rows as seen in SPSS called?
|
A cross-tabulation with tests of association (i.e. "via chi-squared tests and Fisher's exact tests).
|
|
When should odds-ratio be used?
|
In case-control studies
|
|
What does it mean if the odds ratio equals 1?
|
No statistical difference
|
|
An odds ratio of 1 means...
|
The event is equally likely to happen in both groups
|
|
What is an odds ratio?
|
A way of comparing whether the probability of a certain event is the same for two groups
|
|
An odds ratio greater than 1 means...
(i.e. low/high) |
The event is more likely for the FIRST group
(i.e. more likely in the low group) |
|
An odds ratio less than 1 means...
(i.e. low/high) |
The event is more likely for the SECOND group
(i.e. more likely in the high group) |
|
Explain the concept of an odds ratio
|
An odds ratio is a ratio of two quantities. The odds of a specific event (being a case v. a control) in one categorical variable versus the odds of a specific event (being a case v. a control) in another categorical variable, where the odds are the number of cases/number of controls in one group.
|
|
What is unordered qualitative data?
|
Female, Male
|
|
What is ordered qualitative data?
|
Small, Medium, Large
|
|
What is discrete quantitative data?
|
Data has to be an integer (whole number)
Such as number of children in a family |
|
What is continuous quantitative data?
|
Data that can have any number of decimal places
Height |
|
In Levene's Test for Equality of Variances what do you look for FIRST?
|
Whether the "Sig." of the "Equality of Variances Assumed" row is above or below 0.05. If it is BELOW 0.05 then you assume the two groups do not have equality of variances and look to the row labeled "Equality of Variances not assumed"
|
|
What does a t-test assume?
|
Equal Variance (aka equal standard deviations)
|
|
Multiple Testing
|
a challenge for statistical interference that can occur in any instance that involves the simultaneous testing of more than one hypothesis
|
|
What is a type I error?
|
When we reject the null hypothesis (H0) even though it is true
|
|
What is a type II error?
|
When we accept the null hypothesis (H0) even though it is false
|
|
What can lead to a type I error?
|
1. large number of Exposure variables with respect to the risk of a Single outcome
2. Study designs comparing several groups (multiple comparisons) 3. Examining subgroups within a single analysis 4. Implicit or explicit selection of a result from a larger pool of possible analyses |
|
How can a type I error be avoided?
|
1. Careful study design
2. The Bonferroni correction (changes the level at which the p-value is accepted, so 0.005 instead of 0.05) 3. Use of an overall test of significance to detect evidence of any differences between groups |
|
What are some assumptions of a two-sided t-test?
|
Data is independently sampled from a normal distribution
|
|
Sampling Variance
|
The uncertainty introduced by using a sample to estimate the population
|
|
When should a one-way ANOVA be used?
|
For comparing mean measurements in three or more groups
|
|
Tests used for continuous data and what do they assume?
|
One-sided t-test
Two-sided t-test Paired t-test One-way ANOVA They assume normal distribution in the data |
|
Tests used for data that does not appear to be normally distributed?
|
Wilcoxon Signed Rank
Mann-Whitney Kruskal Wallis |
|
Tests for categorical data?
|
Chi-squared
Fishers Exact |
|
Test for paired categorical data?
|
McNemar's Test
|
|
Confounder or confounding factor
|
A factor that may be related to both exposure and outcome separately, and lead to evidence of association between exposure and outcome, even when there is no genuine relationship.
|
|
Relative Risk
|
Measures the change in risk associated with Group 1 relative to that of Group 2
|
|
Simple Random Sampling
|
Most basic form; every member of the population has an equal chance of being included in the sample where inclusion or exclusion is determined by chance
|
|
Stratified Sampling
|
Ensures that certain subsets of the population are represented in the sample in the same proportions in which they occur in the population
|
|
Multi-stage Sampling
|
Random sampling at levels - classes within schools within education authorities
|
|
Cluster Sampling
|
Simplification of multi-stage sampling by including everyone at the lowest stage rather than random selection
|
|
Systematic Sampling
|
Selecting every nth person but patterns can risk bias
|
|
Quota Sampling
|
Specifying the proportion of the sample who should fall into groups defined by characteristics thought to be relevant
|
|
Statistical Significance
|
When we reject H0 then it is statistically significance but that does not necessarily mean practical significance.
|
|
Standard Error relates to...
|
Confidence Intervals
|
|
Standard Deviation relates to...
|
Hypothesis Tests
|
|
Confidence interval containing zero means...
|
It is not statistically significant
|
|
When should a scatterplot be used
|
To compare to quantitative variables directly
|
|
Answering a question about confidence intervals
|
In this context, the best estimate of the mean difference is _____, but the data are consistent with the mean change lying between _____ and _____ (aka the numbers provided by the confidence interval).
|
|
What should we always consider when analyzing data
|
The number of participants in the study as sample sizes that are too small can make it difficult to draw meaningful conclusions
|
|
Practical or Clinical Significance...
|
Requires more information than simply statistical significance, is it worth it to implement the change? Depends on things such as cost, etc. outside the realm of statistics
|
|
A Confidence Interval allows for...
|
clinical relevance to be assessed rather than just the strength of evidence
|