Term
| Hypothesis testing is an extension of what? |
|
Definition
|
|
Term
|
Definition
| A statement (prediction) about one or more populations. |
|
|
Term
| What is the difference between research and statistical hypotheses? |
|
Definition
Research - the conjecture that motivates the research (these lead to stat. hyp.)
Statistical - are stated in such a way that they may be evaluated by appropriate statistical techniques |
|
|
Term
| Describe "the whole process" for research |
|
Definition
1.State the biological problem and define your population of interest
2.Decide on what kind of data, treatments, methods, assumptions, etc. are necessary
3.Formulate a protocol
4.Perform a Pilot Study
5.Collect Data
6.Analyze the Data
7.Interpret the Results |
|
|
Term
| Our hypothesis comes in a set of what two hypotheses? |
|
Definition
Null Hypothesis and Alt hypothesis
**Hyp are always based on pop parameters, not sample stats** |
|
|
Term
| Hypotheses can be formulated to test for what 3 patterns? |
|
Definition
An effect in either direction (not equal)
An effect in one direction (< or >)
A directional effect of a certain minimum size (A-B=2) |
|
|
Term
|
Definition
| rejecting Ho when it is true |
|
|
Term
|
Definition
| accepting Ho when it is untrue |
|
|
Term
| What does a non-directional hypothesis specify? |
|
Definition
| That only an effect exists (the p-value is expected to be different) |
|
|
Term
| What does a directional hypothesis specify? |
|
Definition
| The value of one group is expected to be greater than another group |
|
|
Term
| Is a directional or non-directional hypothesis more powerful? |
|
Definition
| A directional hypothesis is more powerful, but if there is a difference in the direction not specified, the difference won't be found |
|
|
Term
| Define "significantly different" |
|
Definition
| More different than we would expect simply due to random chance (he says we must specify this carefully) |
|
|
Term
| What happens to stat sig as N increases? |
|
Definition
| We expect less variation due to chance |
|
|
Term
| Absence of evidence is not evidence of absence |
|
Definition
| That is all. I don't know how to put that in a question. |
|
|
Term
| We can only make statistical claims regarding which hypothesis? |
|
Definition
| The null one (fail to reject Ho and reject Ho) his example was the judicial system |
|
|
Term
|
Definition
α = Pr(Type I Error) = Pr(false positive)
β = Pr(Type II Error) = Pr(false negative)
Classical hypothesis testing controls α and β is decided afterward – ‘Post-hoc’
These values are inversely proportional, but as N increases, both a and B decrease |
|
|
Term
|
Definition
Power = 1 – β
And power is the prob of rejecting the Ho given that the Ho is actually false |
|
|
Term
| Describe the biological assumptions for a statistical test |
|
Definition
1) all observations are random and unbiased
2) all obs are random
3)All samples are representative of the population
4) The populations are normally distr
5) pops are identical in all characteristics aside from the mean |
|
|
Term
| Which statistical test do you use for mean? Median? |
|
Definition
| t statistic; Mann-Whitney U statistic |
|
|
Term
| What happens to the t distribution as you increase df? |
|
Definition
| The curve becomes more narrow and tall. |
|
|
Term
| What does your p value mean? |
|
Definition
| The probability of obtaining a test-statistic value as extreme or more extreme than the observed value from the data, assuming the null hypothesis is true. |
|
|
Term
| What is the posterior power of a test? |
|
Definition
| Symbolized by beta, it is the inverse of p-value and indicates the power of a test. |
|
|
Term
| When do you fail to reject Ho? |
|
Definition
| When p value is greater than a |
|
|
Term
| What kind of hypothesis test do you run on non-independent samples? |
|
Definition
|
|
Term
| Name 3 methods of pairing data for paired comparison test |
|
Definition
- same experimental unit before and after treatment
-littermates of same sex randomly assigned to 2 treatments
- matching experimental units based on characteristic related to the measure of interest (weight, blood pressure) |
|
|
Term
| Disadvantages of paired comparisons are: |
|
Definition
- lots of time and expense finding pairs
- loss of df (would have been 2n-2, now it's n-1) |
|
|
Term
|
Definition
| is a statistical technique where the total variation present in a set of data is partitioned into 2 or more components. You can determine the magnitude of different class variables contributing to the toal variation. |
|
|
Term
| What are the 2 purposes of ANOVA? |
|
Definition
(1) Estimate and test hypotheses about population variances
(2) Estimate and test hypotheses about population means (we’ll focus mostly on this) |
|
|
Term
| What are 2 sources of variation? |
|
Definition
| Total variation = model variation + residual variation |
|
|
Term
| What are the assumptions of ANOVA? |
|
Definition
1. Normal distribution
2. Independence of errors (obs are not infl by each other)
3. Homoscedasticity (homogeneity of variances) |
|
|
