MaizePrepSTATS 250
Exam 3 · Tue, Dec 15, 7:30 PM ·  

ANOVA, Regression and Chi-square

One-way ANOVA and Tukey; correlation and simple regression; slope inference; multiple regression; categorical predictors; interactions; two categorical variables; chi-square. Covers Lectures 15–22, Exam Prep 07–10, Labs 07–09.

Sections8
Practice questions75
Reading205min

Suggested study split

Not point weights: nothing official is posted yet

Patterns that repeat every time

  1. 01

    Read the output, not the arithmetic

    Lecture 15 said you will never compute F by hand: read F and its p-value from the ANOVA table and ignore the intermediate columns. Exams 1 and 2 also asked no hand calculation beyond short arithmetic. Expect R output to read and define, not to produce.

  2. 02

    Define the statistic in words

    Exams 1 and 2 asked for test-statistic and p-value definitions in context. Lecture 15 gave the F version: say the variation among the sample means was F times as large as the variation within groups (lecture: "10 times as large" for F = 10). It is a ratio, not a count of standard errors.

  3. 03

    Omnibus first, then pairs

    An ANOVA conclusion only says at least one population mean differs. Lecture called it underwhelming on purpose: Tukey intervals (lwr, upr) tell you which pairs differ, by whether each contains 0.

  4. 04

    Quantity A versus B and "what changes"

    Both earlier practice exams used Quantity A/B grids and "what happens if…" items. For Exam 3 the natural versions are: which data set gives the larger F, which model has the larger adjusted R², how a group gap changes with x in an interaction model.

  5. 05

    Part II: one study, several research questions

    Exams 1 and 2 gave one study with Research Questions A–C and about 5 written prompts graded by rubric elements: define a parameter, define a statistic or p-value, write a conclusion, and take a position supported by two results. Case Study 3 (week of Dec 7) will show the Exam 3 version.

Exam format and strategy

Tue, Dec 15, 7:30 PM

Format

  • Tuesday, December 15, 2026 · 7:30–9:30 pm · 120 minutes for both parts (syllabus)
  • Part I: Concepts and Skills, primarily multiple choice; Part II: Written Statistical Interpretation. Scored separately: Part I counts toward Category 1, Part II toward Category 2 (Written Statistical Interpretation 03 is 8% of the course grade)
  • No Exam 3 practice exam or help card is posted yet. On Exams 1 and 2 the suggested split was about 80 minutes for Part I and 40 for Part II, and Part II was one study with new R output and about 5 prompts (10 minutes to read, 30 to write). Expect the same; check the review materials posted one week before the exam
  • Case Study 3 runs in lab the week of December 7: that is the Part II practice
  • The topic shares above are a study split (equal time per lecture), not point weights

Allowed

  • Official help card provided (the Exam 3 card is not posted yet)
  • No calculator: the syllabus, the case-study lab slides, the practice-exam information pages and lecture all say so. The printed Exam 1 and 2 covers said calculators were not required but allowed; confirm with the instructors
  • Closed notes and closed book; photo ID at submission (as on Exams 1 and 2)

Strategy

  • Budget about 80 minutes for Part I and protect about 40 for Part II, as on Exams 1 and 2, unless the Exam 3 cover says otherwise.
  • For any ANOVA output, write three things before answering: the hypotheses (all means equal versus at least one differs), F as a ratio in words, and the right-tail p-value on the course evidence scale.
  • For Tukey output, read only lwr and upr: contains 0 means no evidence of a difference for that pair; entirely positive or negative means a difference in the direction of the subtraction order.
  • For regression output, find the right row first, say what is held fixed, and give units per unit of x.
  • Use the course evidence words (not enough, some, strong, very strong) and keep the design straight: random assignment for causation, random sampling for generalizing.
  • Do Case Study 3 in lab and the review materials posted one week before the exam; they will show the real Exam 3 format.

Sections

In the order they are tested

Formula sheet

Lecture 15: F in words
F=variation between the group meansvariation within the groupsF=\frac{\text{variation between the group means}}{\text{variation within the groups}}
Never computed by hand in STATS 250. F is never negative; the p-value is the right-tail area P(F≥Fobs∣H0)P(F\ge F_{obs}\mid H_0). The Tuesday lecture explained the two F degrees of freedom (k−1k-1, n−kn-k); the Monday lecture said to ignore them. Both said to read only F and the p-value from the table.
Lecture 15: ANOVA hypotheses
H0:μ1=μ2=⋯=μkHa: at least one μi differsH_0:\mu_1=\mu_2=\cdots=\mu_k\qquad H_a:\ \text{at least one }\mu_i\text{ differs}
Lecture 15: Tukey reading
(lwr, upr)∋0⇒no evidence of a difference;0∉(lwr, upr)⇒evidence of a difference(\text{lwr},\ \text{upr})\ni 0\Rightarrow\text{no evidence of a difference};\quad 0\notin(\text{lwr},\ \text{upr})\Rightarrow\text{evidence of a difference}
Course pack: simple regression
y^=β^0+β^1x,e=y−y^,R2=r2,t=β^1SE(β^1), df=n−2\hat y=\hat\beta_0+\hat\beta_1x,\quad e=y-\hat y,\quad R^2=r^2,\quad t=\frac{\hat\beta_1}{SE(\hat\beta_1)},\ df=n-2
Course-pack preview; Lec 16–17 have not happened yet.
Course pack: slope interval
β^1±t∗ SE(β^1)\hat\beta_1\pm t^*\,SE(\hat\beta_1)
Course pack: interaction model
y^=b0+b1x+b2I+b3xI:slopes b1, b1+b3; gap at x=b2+b3x\hat y=b_0+b_1x+b_2I+b_3xI:\quad \text{slopes } b_1,\ b_1+b_3;\ \text{gap at }x = b_2+b_3x
Course pack: chi-square
E=(row total)(column total)n,χ2=∑(O−E)2E,df=(r−1)(c−1)E=\frac{(\text{row total})(\text{column total})}{n},\quad \chi^2=\sum\frac{(O-E)^2}{E},\quad df=(r-1)(c-1)
Condition: at least 80% of expected counts above 5 and none below 1.

MaizePrep is an independent study resource. It is not affiliated with, sponsored by, or endorsed by the University of Michigan. Course names and numbers are used only to describe which class the material covers. Every practice question is original, written in the style of the course.