Upgrade to remove ads
Only $35.99/year

Hypothesis Testing

Terms in this set (35)

What is hypothesis testing
Determining whether results of a scientific study are meaningful
What is the main question you are trying to answer when using hypothesis testing
Are the findings real or due to chance
How do you interpret the probabilities that come out of hypothesis testing
Is dependent on data given duh
Depend on the difference between the means of normal and not
The scatter of data
Number of subjects are tested
When looking at a scatter plot graph, how can you determine if something was due to chance or or not
If there is a huge size difference between two means of the two subgroups, you can infer that the result is not due to chance.
The closer the means the more likely it is to be by chance
What can you determine if the data points on the graph are scattered widely and mixing together
More likely that the finding is due to chance and less likely to be real
Which is better, having a large sample size or a smaller sample in a scatter graph
Larger, it makes it easier to determine if the points are clustering or not. Therefore, making it easier to determine if the data is real or by chance
What is H0 called
Null hypothesis
What is H1 called
Alternative hypothesis
In reality which is correct: H0 or H1
Determined by the experiment
What are the 4 possible outcomes from hypothesis testing
#1: There is a difference in reality, detected by the experiment
#2: There is no difference in reality, experiment finds no difference
#3: There is no difference in reality, but the study finds a difference
#4: There is a difference in reality, but the study misses it
What is a type 1 (alpha) error
(#3) when there is no difference in reality, but our study finds one
What is a type 2 (beta) error
(#4) when there is a difference in reality, but the study misses it
What does Power mean in regards to hypothesis testing
Chance of detecting difference
What does alpha mean in regards to hypothesis testing
Chance of seeing difference that is not real
What does beta mean in regards to hypothesis testing
Chance of missing a difference that is really there
What does power equal
Power= 1- Beta
What is the P-value
Used to accept or reject an hypothesis
Calculated based on difference between groups (magnitude, scatter, number of samples)
If there is no difference between means what does that determine
Represents chance that the null hypothesis was correct
What does it mean if the p-value is greater than 0.05
Usually the null hypothesis is rejected, deems h1 as correct
What does P-value represent
Represents the cancer of a False positive, no difference in reality
What is the difference between the p-value and alpha
The alpha level is set by study design
The p-value is calculated by a comparison of the two groups
What can Power also be called when dealing with hypothesis testing
Rejecting the null hypothesis because when you identify a difference in the study you will reject the null hypothesis
Things related to power calculation
Is used to "maximize power of study to detect a true difference"
Helps determine if conducting the study is worth your time
Used if you have little or no control over scatter of data and difference between means
You do have control over number of subjects
What is power calculation
Determining the number of subjects needed to give a high power, power of 80%
What does detecting a difference mean
When the P-value is below 0.05 → "the study detected a difference"
When the P-value is above 0.05 → "the study detected no difference"
Which statistical error is a false positive
Type 1 (alpha) error
Which statistical error is being described: A difference or effect is found, but there is none in reality
Type 1 (alpha) error
Which statistical error is being described: the result of this error will be rejecting the null hypothesis (H0) when you should not have
Type 1 (alpha) error
What are some causes of a type 1 error
Random chance
Improper research techniques
What is a random chance
Most studies have an alpha error of 5%
1/20 times error occur
One of the possibilities for an alpha error to occur
Which statistical error is a false negative
Type 2 (beta) error
Which statistical error is being described: there is no difference or effect found, when there is one in reality
Type 2 (beta) error
Which statistical error is being described: the result of this error will be accepting the null hypothesis when you should not have
Type 2 (beta) error
List the cause for Type 2 (beta) errors to occur
Too few patients
Why would having a low number of patients/ subjects cause a Type 2 (beta) error
You need lots of patients to have a sufficient P-value, you wont find a difference if there are too few subjects.
Students also viewed