| A | B |
|---|
| Normal distribution percents | 68-95-99.7 |
| Mean | Average |
| Median | Middle observation |
| Correlation | Strength of linear relationship between two numerical variables |
| R - squared | Percent of variation in response variable explained by the model |
| Outlier | Outside 1.5XIQR range |
| Law of Large Numbers | In the long run, a statistic approaches the population parameter |
| Central Limit Theorem | As x bar increases, the sampling distribution get more normal |
| Segmented bar graph | Used to examine categorical variables |
| Probability | Chances of an event occurring |
| Essentials to designing experiments | Replication, control, randomnization |
| Z - test for means | When you know the population standard deviation |
| T - test for means | When you do not know the population standard deviation |
| Skewed right distribution | When the average is greater than the median |
| Skewed left distribution | When the mean is smaller than the mdeian |
| Bias | When something is or could be systematically wrong with the design of a survey of experiment |
| Parameter | Number concerning a population |
| Statisitc | Number concerning a sample |
| Influential Observation | Outlier in the X variable |
| Residuals | Diference between obs. - exp. (always sum to 0) |
| IQR | Q3 - Q1 |
| Standard Deviation | Measure of the spread of data about the mean |
| Standard Normal Curve | Has mean 0 and std. dev. 1 |
| Categorical variable | Groups without numerical values |
| Quantitative variable | Variable which takes on numerical values |
| Lurking variable | A variable that may have a large influence on the response, but is not part of the explanatory variables |
| Curve in the residual plot | Indicates posible presence of exponential growth |
| Experiment | Deliberately imposing a treatment upon something or someone to observe the response |
| Sample | A selection of possible items which should represent the larger population being studied |
| Bias | Design flaws that systematically favor certain outcomes |
| Randomization | Using chance to assign subject to treatments to eliminate bias |
| Replication | Using many subjects in an experiment to reduce chance variation |
| Matched Pair | When one serves as its own control in an experiment |
| SRS | Simple random sample, like drawing names out of a hat |
| Stratified random sample | Dividing the population into similar units (strata), then getting an SRS from each |
| Variability | The spread of the sampling distribution (dependent on sample size) |
| Disjoint or mutually exclusive | Events that share no outcomes |
| Independence | How knowing one thing gives no information on another thing |
| Union | Could be in this group OR that group |
| Intersection | Must be in this group AND that group |
| Binomial coefficient | n choose k or n!/k!(n-k)! |
| Critical value | The number that has probability p lying to the right of it on a density curve |
| Confidence level | The probability that the method used will give correct answer C% of the time |
| Null hypothesis | A statement of "no effect" or no difference |
| Alternative hypothesis | If chosen, it asserts that there is an effect present |
| P-value | Assuming Ho true, the probability that results as extreme as these obtained were gotten by chance |
| Matched pair design | When one often serves as their own control in an experiment |
| Robust | Resistant to outliers and violations of assumptions |
| Cell | A certain row and column combination |
| Margin of error | How accurate a particular estimate is to otbaining the true parameter |
