lecture 2- effect size and power 2 Flashcards
strength of association
How much of the variation in the dependent variable can be
explained by the independent variable
partial eta squared
-when is it used
-what does it measure
-is partial or classical eta squared
Can be used for a factorial design (ANOVA)
Measures linear and nonlinear association (in contrast to
correlation)
Discussion in literature whether partial or classical eta squared
is better
what does classical eta square measure
vs partial eta square
Classical eta square measures the proportion of the total variance
in a dependent variable that is associated with the variance of a
given factor in an ANOVA model.
Partial eta square is a similar measure in which the effects of
other independent variables and interactions are partialled out.
Classical η2 adds to no more than one: advantageous for
describing variance within one experiment
Partial η2 can add to >1. Variance due to other factors is
removed, supporting comparisons of effect sizes across
experiments with different factor structures.
three risk estimates
relative risk
odds ratio
risk difference
relative risk
relative risk refers to a measure used to compare the risk of a certain outcome (like developing a condition or experiencing an event) between two groups
odds ratio
An odds ratio (OR) is a measure used to compare the odds of a particular event occurring in one group relative to the odds of it occurring in another group.
- It is commonly used in case-control studies or in situations where you’re comparing the odds of an outcome between two different groups (like those exposed to a factor vs. those not exposed).
When risk is small the odds ratio approximates the relative risk
The absolute difference in risk (probability) of an event between two groups.
-a measure that quantifies the difference in the risk of an outcome between two groups. Shows how much the risk changes between the exposed group (e.g., people who have a particular characteristic or intervention) and the unexposed group.
Difference between proportion of treatment group that contract the
disease and proportion of controls that contract the disease
Can be used to estimate number of cases avoided by a treatment (if
population size is known)
Reflects overall probability of getting disease
Easy to understand, also for non-experts.
recap
type 1 error
type 2 error
Type 1 Error (False Positive):
A Type 1 error occurs when you incorrectly reject a true null hypothesis. In other words, you conclude that there is an effect or a difference when, in fact, there isn’t one.
Type 2 Error (False Negative):
Definition: A Type 2 error occurs when you fail to reject a false null hypothesis. In other words, you conclude that there is no effect or difference when, in reality, there is one.
what is power
Power is the probability of correctly rejecting the null-hypothesis (i.e.
when H0 is incorrect)
-In other words, power is the ability of a test to detect an effect if there is one.
what factors influence power
Effect size
note: unreliable measures reduce effect size by inflating estimate of sigma (e.g., Cohen’s d = (M1 – M2) / SDpooled)
Alpha level (e.g., α=0.05 or α=0.01)
One-tailed test has higher power than two-tailed
Sample size
what is prospective power
-what are the steps
Computed before the study’s data are collected
Three steps for calculating prospective power:
o Hypothesize effect size
o Alpha level
o Planned sample size
Get an estimate of effect size
o Do a pilot experiment and compute effect size
o Do a meta-analysis and compute weighted effect size
o Use Cohen’s estimates for small, medium and large effect size
explain the importance of power
Power determines how likely an effect is detected (assuming there
really is an effect)
Suppose a drug is highly effective, but the researcher only tests 5
participants per group (treatment, control): Drug may never get onto
the market
a power at least _____ is usually considered acceptable
A power of at least 80% is usually considered acceptable
* Underpowered (power<80%) studies are useless and
unethical (waste of resources and people’s time)
computing power - software
G*power 3
Free to download
Runs on different operating systems
what is observed power
-is it useful
Computed after study is completed
Assumes effect size in the sample equals effect size in the population
Generally not very useful
example of when is observed power usueful
excpetion :
In a meta-analysis, observed power is useful
It provides an indication as to which results to assign a higher weight
increasing the power
Adding participants
- Adding participants to groups that are cheaper to run
Choose a less stringent significance level (usually not an option)
Increase the hypothesized effect size
- Strengthening the intervention/manipulation (in experiments)
- ‘Throw out’ middle part of distribution
- Improve reliability of measurement Increasing power
- Use as few groups as possible
Use covariates variables
Use a repeated measures design
Use measures sensitive to change
