Term
| Confidence intervals enable us to quantify the amount of uncertainty in our estimates of population parameters based on statistics calculated from samples. |
|
Definition
| Confidence intervals enable us to state that we are X% confident that the pop'n parameter of interest is AT MOST a specified interval from the sample statistic (e.g. if the 95% CI for the AVG of a dataset was 10+/-1, we would be 95% confidence that the mean of the pop'n from which the sample was drawn was between 9 and 11. |
|
|
Term
|
Definition
describes the relationship between the samplling distribution of sample means and the pop'n from which the samples are taken. As you get more and more samples, the CI shrinks, and also the sample mean moves closer and closer to the pop'n mean.
As n (sample size) gets large, the samplling distribution of the sample mean (averages) approximates a normal distribution, if n are draw from any pop'n with a mean u and SD sigma. |
|
|
Term
| Inferential statistics - enables us to move from describing what has happened (descriptive statistics) to predicting what is likely to happen (inferential stats) |
|
Definition
| The Central Limit Theorem describes the distribution of averages of samples take from pop'ns |
|
|
Term
Calculate Confidence interval of
- Mean
- sigma
- process capability
- proportion |
|
Definition
| highlight formulas - indicate page |
|
|
Term
| Basic steps in hypothesis testing |
|
Definition
Type I risk: beta B (usually 10-20%)
Type II risk: alpha a level (usually 5%) |
|
|
Term
Type I error (producer's risk): error of rejecting a good-quality lot creates a problem for the producer; examine a crate of oranges, conclude oranges are defective and refuse to buy but in reality less than 5% of oranges are defective;
Type II error (consumer's risk): examine a sample of the oranges conclude there are 5% or less defective and buy them, but get them home to realize that the oranges are much worse than your sample indicated. |
|
Definition
alpha a is the risk of finding a difference when one really doesn't exist
beta B: the risk of not finding a difference when one really does exist |
|
|
Term
| if p value alpha (p>0.05), then not sig diff. |
|
Definition
|
|
Term
|
Definition
| T - test is used to if the means of two populations are equal (assuming similar variance) whereas F-test is used to test if the variances of two populations are equal. F - test can also be extended to check whether the means of three or more groups are different or not (ANOVA F-test). |
|
|
Term
| ANOVA - be able to complete an ANOVA table - if they gave an Alpha - |
|
Definition
|
|
Term
|
Definition
| n=1 (sometimes indicated as v) |
|
|
Term
| one-way-anova: you'll see output of an ANOVA - and asked which values are "E" significantly different from? |
|
Definition
|
|
Term
| What does it mean to "reject the null hypothesis"? |
|
Definition
| reject that the groups are the same. meaning they are different. |
|
|
Term
| Tukey method - which 3 machines are all not statistically different? (see grouping of Bs, all the same) |
|
Definition
|
|
Term
| Correlation coefficient (sample) |
|
Definition
| memorize formula (determine location of formula and how to use it) |
|
|
Term
|
Definition
|
|
Term
|
Definition
| get p value, then if it's less than the alpha (e.g. 0.05), what do we do next |
|
|
Term
|
Definition
| if it's .99, about 99% of the x can explain/contributes to the y (the closer to 1 that the R^2 is, the more accurate that equation is at representing the data); if the R^2 is low, there are probably other inputs than just x going on, and your equation isn't representing |
|
|
