Statistics Review: Conceptual Flashcards

0.0(0)
Studied by 0 people
call kaiCall Kai
learnLearn
examPractice Test
spaced repetitionSpaced Repetition
heart puzzleMatch
flashcardsFlashcards
GameKnowt Play
Card Sorting

1/20

flashcard set

Earn XP

Description and Tags

Comprehensive vocabulary flashcards covering conceptual topics across confidence intervals, hypothesis testing, proportions, chi-square tests, linear regression, and one-way ANOVA.

21 Terms

1
New cards

95% Confidence Level

The property that 95% of intervals constructed from repeated random samples will capture the true population parameter; once calculated, the parameter is fixed either inside or outside the interval.

2
New cards

Margin of Error (Sample Size and Confidence Level Effects)

The interval half-width, which narrows when increasing sample size nn (because standard error is inversely proportional to n\sqrt{n}) and widens when increasing confidence level (due to a larger critical value z∗z^*).

3
New cards

Duality of Confidence Intervals and Hypothesis Tests

The principle that a (1−α)(1 - \alpha) confidence interval contains all plausible values of p0p_0 that would not be rejected in a two-sided test at significance level α\alpha; if p0p_0 falls outside the interval, H0H_0 is rejected.

4
New cards

pp-value

The probability of observing sample results as extreme as or more extreme than the observed data, assuming the null hypothesis (H0H_0) is true.

5
New cards

Failing to Reject H0H_0

A statistical decision meaning there is insufficient statistically significant evidence against H0H_0; it never proves that the null hypothesis is true.

6
New cards

Type I Error

A false positive error occurring when a researcher rejects a null hypothesis (H0H_0) that is actually true, with the maximum tolerable probability set by significance level α\alpha.

7
New cards

Statistical Significance vs. Practical Importance

The distinction that a statistically significant result (p<0.05p < 0.05) does not inherently mean an effect is large or practically meaningful, as very large sample sizes can produce tiny pp-values for trivial differences.

8
New cards

Normality Condition for Sample Proportions

The Central Limit Theorem condition requiring at least 10 expected successes (np≥10np \ge 10) and at least 10 expected failures (n(1−p)≥10n(1 - p) \ge 10) for the sampling distribution of p^\hat{p} to be approximately normal.

9
New cards

Sample Size Rounding Rule for Proportions

The requirement to always round the calculated sample size nn up to the next whole integer so that the margin of error does not exceed the specified bound.

10
New cards

Pooled Proportion (p∗p^*)

An estimate combining successes across total trials from both samples, used in the standard error formula for a two-sample proportion hypothesis test under the null hypothesis assumption that p1=p2p_1 = p_2.

11
New cards

Expected Cell Counts Condition (χ2\chi^2 Test)

The validity condition for a χ2\chi^2 test requiring that all expected cell counts must be at least 5, rather than the observed counts.

12
New cards

χ2\chi^2 Distribution Properties

A distribution defined by χ2=∑(O−E)2E≥0\chi^2 = \sum \frac{(O - E)^2}{E} \ge 0 that is right-skewed and always evaluated as an upper-tail test because large values provide evidence against H0H_0.

13
New cards

Correlation Coefficient (rr)

A unitless, symmetric statistic measuring linear association between two variables; it is sensitive to extreme outliers, unaffected by unit changes or swapping xx and yy, and does not prove causation without a randomized controlled experiment.

14
New cards

Zero Correlation (r=0r = 0)

The condition indicating an absence of a linear relationship between two variables, which does not rule out the presence of a strong non-linear (curved) relationship.

15
New cards

Extrapolation

The practice of predicting response values (yy) for explanatory values (xx) outside the domain of observed data, which is unreliable because the linear pattern may not continue.

16
New cards

Test of Regression Slope (H0:β1=0H_0: \beta_1 = 0)

A hypothesis test where failing to reject H0H_0 indicates no significant linear relationship exists, meaning xx does not help explain yy and the mean yˉ\bar{y} is just as good a predictor as the regression line y^\hat{y}.

17
New cards

Mean Response Interval vs. Prediction Interval

A comparison where a confidence interval for the mean response is narrower than a prediction interval for an individual observation because averages vary less than individual data points.

18
New cards

Coefficient of Determination (r2r^2)

The square of the correlation coefficient, representing the proportion of variability in yy that is explained by its linear relationship with xx.

19
New cards

Equivalence of One-Way ANOVA and Linear Regression

The mathematical equivalence showing that a one-way ANOVA across kk groups yields the exact same FF-statistic, pp-value, and R2R^2 as a linear regression model utilizing indicator (dummy/factor) variables.

20
New cards

Alternative Hypothesis (HaH_a) in One-Way ANOVA

The hypothesis stating that at least one group population mean differs from the others, rather than asserting that every group mean is different.

21
New cards

R2R^2 in ANOVA

The ratio calculated as SSModel/SSTotalSS_{\text{Model}} / SS_{\text{Total}}, representing the proportion of total variation in the response variable explained by group differences.