THE VR SCHOOL
  • Schools & Districts
    • License the academy
    • Free program for your students
    • Request a demo
  • Catalog
    • AI & Data
    • Cybersecurity & Web
    • Programming
    • Health Science
    • All Guilds
    • All courses
  • Families
  • Accreditation
  • Log In
  • Request a demo
  • Home
  • Courses

RLP Statistics B — Probability Models and Statistical Inference

Curriculum

  • 4 Sections
  • 20 Lessons
  • 10 Weeks
Expand all sectionsCollapse all sections
  • Unit 1: Continuous Distributions and Sampling Distributions
    5
    • 1.1
      Continuous Random Variables and Density Curves
      50 mins
    • 1.2
      The Normal Distribution
      50 mins
    • 1.3
      When Is a Normal Model Appropriate?
      50 mins
    • 1.4
      Sampling Distributions and the Central Limit Theorem
      50 mins
    • 1.5
      Performance Task — Sampling Distribution Simulation Study
      150 mins
  • Unit 2: Estimating with Confidence
    5
    • 2.1
      Estimating a Proportion with Confidence
      50 mins
    • 2.2
      Confidence Intervals for a Mean
      50 mins
    • 2.3
      What Confidence Means, and What It Does Not
      50 mins
    • 2.4
      Comparing Two Groups with Confidence Intervals
      50 mins
    • 2.5
      Performance Task — Estimation Study
      150 mins
  • Unit 3: Testing Claims: Significance Tests
    5
    • 3.1
      The Logic of a Significance Test
      50 mins
    • 3.2
      t-Tests for Means and Paired Data
      50 mins
    • 3.3
      Errors, Power and Significance in the Real World
      50 mins
    • 3.4
      Two-Sample Tests from Experiments and Observational Data
      50 mins
    • 3.5
      Performance Task — Significance Test Investigation
      150 mins
  • Unit 4: Chi-Square, Regression Inference and the Capstone Study
    5
    • 4.1
      Chi-Square Goodness of Fit
      50 mins
    • 4.2
      Chi-Square Tests for Two-Way Tables
      50 mins
    • 4.3
      Inference for the Slope of a Regression Line
      50 mins
    • 4.4
      Choosing the Right Procedure and Writing It Up
      50 mins
    • 4.5
      Capstone — Full Statistical Inference Study
      200 mins

Inference for the Slope of a Regression Line

Unit 4  ·  Statistical Discourse Lab  ·  Lesson 18 of 20

Inference for the Slope of a Regression Line

PS.12RLP-STATSB
Student examines a rising temperature scatterplot with residuals beside a weather station and thermometer.
By the end of this lesson I can…

test and estimate the slope of a population regression line, check its conditions with residuals, and argue about what the result shows.

Instruction

In RLP Statistics A the least-squares slope b described a data set. Now treat the data as coming from a model: y = α + βx + error, where the true slope β is a parameter and b is its estimate.

The standard error of the slope. Let s = √(Σ residual2/(n − 2)) be the typical size of a residual (n − 2 because two parameters were estimated). Then

SEb = s / (sx√(n − 1))

The slope is estimated more precisely when points sit close to the line (small s) and when the x-values are spread out (large sx). Inference uses t with df = n − 2: the test statistic for H0: β = 0 is t = b/SEb, and the interval is b ± t*·SEb.

Example 1: Springfield, annual mean temperature against year, 1950-2024 (NOAA; n = 75). Technology gives b = 0.0303°F per year, s = 1.337°F, SEb = 0.00713, r = 0.445.

  • Test: t = 0.0303/0.00713 ≈ 4.25, df = 73, P < 0.001 (0.00006).
  • 95% interval (t* = 1.993): 0.0303 ± 1.993(0.00713) = 0.0161 to 0.0445°F per year, about 0.16 to 0.45°F per decade.

Example 2: St. Louis, same years (n = 75). b = 0.0524°F per year, SEb = 0.00799, t ≈ 6.55, P < 0.001, 95% interval 0.036 to 0.068°F per year (0.36 to 0.68°F per decade), r = 0.608.

Testing the slope is testing the correlation (PS.13). The same t can be written t = r√(n − 2)/√(1 − r2). For Springfield: 0.445 × √73/√(1 − 0.198) ≈ 4.25. A slope of zero and a correlation of zero are the same hypothesis. Note that r2 = 0.198 at Springfield: the trend is statistically clear but explains only about 20% of the year-to-year variation. Significance and strength are different things (A1.DS.A.7).

Conditions (LINER). Linear form (scatterplot and residual plot); Independent residuals; Normal residuals (histogram or normal probability plot); Equal variance of residuals across x; Random sample or random assignment.

Where these examples strain the conditions. Years are not a random sample, and the residuals are not independent. Consecutive years’ residuals are correlated (lag-1 correlation about 0.29 at Springfield and 0.40 at St. Louis), because warm and cool spells last several years. Positively correlated residuals make SEb too small, and so make t too large and the P-value too small. For St. Louis, with t near 6.6, the conclusion that the recorded temperatures trend upward survives any reasonable correction. For Springfield, with t = 4.25, the evidence is real but weaker than the P-value suggests. Climate scientists use methods that model this autocorrelation, and a careful report says the simple t-test is an approximation here.

Interpreting the slope in context (A1.DS.A.6). The slope is an average rate of change in the recorded values at this station. The St. Louis slope is about 1.7 times the Springfield slope. Candidate explanations include differences in regional climate, the growth of the St. Louis area around the airport, and station and instrument changes. The regression cannot choose between them.

For the discourse lab (MP.3). Claim for debate: “The Springfield test proves that Springfield has warmed.” Argue for or against, using the interval, r2, the residual autocorrelation and the scope of the data. Then write the strongest objection to your own argument, and answer it.

📖 Read for this lesson

The reading opens over the lesson; close it to come back to exactly where you were. The checks below draw on it.

  • The Slope, Its Standard Error and Its Autocorrelated Residuals, Worked for Three StationsNOAA NCEI Global Summary of the Year, element TAVG, stations USW00013994 (St. Louis), USW00013995 (Springfield) and USW00013987 (Joplin), 1950-2024; regression sums, residual autocorrelation and the effective-sample-size correction computed for this course; retrieved 2026-11-10 · about 5 min

    Every sum behind the lesson’s ‘technology gives’, a third station in full, and the autocorrelation correction turned into numbers: St. Louis stays strongly significant, Springfield stays significant, and Joplin’s drops to the edge. The argument in the lab starts here.

  • Seventy-Five Years at Two Missouri Airports, and the Normal Table to Read Them WithNOAA NCEI Global Summary of the Year, element TAVG, stations USW00013994 (St. Louis Lambert International Airport) and USW00013995 (Springfield-Branson National Airport), 1950-2024; standard normal table computed for this course; retrieved 2026-11-10 · about 6 min

    The 75-year St. Louis and Springfield series the regressions are fitted to.

Vocabulary in context

  • population regression line — The model’s true line, μ_y = α + βx; the least-squares line estimates it.
  • standard error of the slope — SE_b = s/(s_x√(n − 1)); the typical sampling variation of b.
  • residual standard deviation (s) — √(Σresidual²/(n − 2)); the typical size of a prediction error.
  • autocorrelation — Correlation between a series and itself shifted in time; positive autocorrelation in residuals makes simple standard errors too small.

Formative check

Work through these before moving on. They are not graded — they tell you whether the standard below has landed yet.

Fill in the blank

For Springfield, t = 0.0303/0.00713 ≈ with df = .

+50 XP

The Springfield slope test gives P < 0.001 but r² = 0.198. Which statement is correct?

Statistical significance of the slope and the strength of the relationship are different questions.
Myth or Fact?

Positive autocorrelation in the residuals makes the simple slope test’s P-value too small.

Correlated residuals carry less independent information than n suggests, so SE_b is underestimated and t is inflated.

Record (or write) a 90-second argument for or against: ‘The Springfield test proves that Springfield has warmed.’ Use the 95% interval for the slope, r², the residual autocorrelation and the scope of a single-station record. Then fit the same model to Joplin (its 75 years are printed in the Lesson 4.3 reading) or another Missouri station from NOAA’s Global Summary of the Year, report b, SE_b, t and the interval, and write the strongest objection to your own argument and your answer to it.

0 words
Quick self-check

How confident are you that you can test and estimate a regression slope, check its conditions, and argue about what it shows?

Not yetVery confident

Practice

Work these on paper or in your notebook, then open Check your answer. Aim for all of Fluency and Application; try at least one Challenge.

Printable version: this unit’s practice workbook (PDF)

Fluency

Build speed and accuracy with the core skill.

  1. A slope estimate b = 0.30 has SE 0.12 with n = 27. Find t and the df for H₀: β = 0.
    Check your answer
    Answer: t = 2.5, df = 25.
  2. Find the two-sided P-value for t = 2.5 with df = 25.
    Check your answer
    Answer: ≈ 0.0193.
  3. Find the 95% interval for β when b = 0.30, SE = 0.12 and df = 25.
    Check your answer
    Answer: 0.30 ± 2.060 × 0.12 ≈ (0.053, 0.547).
  4. A slope test has residuals whose lag-1 correlation is 0.4. Is the reported P-value too large or too small, and why?
    Check your answer
    Answer: Too small: positively correlated residuals carry less independent information, so the SE of the slope is underestimated and t is inflated.
  5. What do the letters in LINER stand for?
    Check your answer
    Answer: Linear form, Independent residuals, Normal residuals, Equal variance, Random sample or assignment.
  6. r = 0.5 with n = 27. Find the t statistic for H₀: β = 0 from r.
    Check your answer
    Answer: t = 0.5√25/√0.75 ≈ 2.887.
    t = r√(n − 2)/√(1 − r²).

Application

Use the skill in context. Show your reasoning.

  1. Joplin annual mean temperature against year, 1950-2024 (75 years): b = 0.0243°F per year, SE = 0.00727, r = 0.364. Test H₀: β = 0 and give a 95% interval per decade.
    Check your answer
    Answer: t ≈ 3.34, df = 73, P ≈ 0.0013. Interval ≈ 0.10 to 0.39°F per decade. r² ≈ 0.133, so the trend explains about 13% of year-to-year variation.
  2. St. Louis, same analysis (75 years): b = 0.0524°F per year, SE = 0.00799, r = 0.608. Compute t both from b/SE and from r, and interpret.
    Check your answer
    Answer: b/SE ≈ 6.55; r√(n − 2)/√(1 − r²) ≈ 6.55. The two agree (they are the same test). P < 0.001: convincing evidence of an upward trend in the recorded values, and the trend explains about 37% of the variation.
  3. St. Louis precipitation against year, 1950-2024: b = 0.1810 in per year, SE = 0.0399, P ≈ 0.00. A student writes, ‘The test proves St. Louis’s rainfall has not changed.’ Correct the conclusion.
    Check your answer
    Answer: Fail to reject H₀: there is no convincing evidence of a linear trend in St. Louis’s annual rainfall. That does not prove no change: the 95% interval for the slope, about 1.01 to 2.61 in per decade, allows modest trends either way, because rainfall varies so much from year to year.

Challenge

Stretch problems. Expect to think before you write.

  1. Missouri’s age-adjusted cancer death rate against year, 1999-2017 (CDC): b = -2.353 per 100,000 per year, SE = 0.0824, n = 19. Compute the 95% interval for β, then explain why a ‘population’ regression model is a delicate idea for these data.
    Check your answer
    Answer: df = 17, interval ≈ (-2.527, -2.179). The 19 rates are not a random sample: they are the complete record for those years. Inference here treats the yearly rates as a steady trend plus random year-to-year variation; that is a model assumption, and correlated residuals would make the interval too narrow.

Review

Keep earlier skills sharp.

  1. A 4-category variable is claimed to be uniform. Observed counts: 27, 41, 26, 56. Carry out a chi-square goodness-of-fit test.
    Check your answer
    Answer: expected 37.5 each; χ2 ≈ 15.92, df = 3, P-value ≈ 0.0012. At α = 0.05, reject H0: there is convincing evidence the categories are not equally likely.
    χ2 = Σ(observed − expected)2/expected; all expected counts are at least 5.
  2. (RLP Statistics A) Interpret a slope of 0.0524°F per year for St. Louis annual mean temperature.
    Check your answer
    Answer: The model predicts recorded annual mean temperature rose by about 0.052°F per year (about 0.52°F per decade), on average, from 1950 to 2024.

Missouri Learning Standards — Mathematics (data and probability) addressed: A1.DS.A.5a, A1.DS.A.7, A1.DS.A.6, MP.3

Extends beyond Missouri Statistics (enrichment): content Missouri places after Statistics

Instructional focus: t-procedures for the regression slope and critical evaluation of their conditions

Reading

The Slope, Its Standard Error and Its Autocorrelated Residuals, Worked for Three Stations

NOAA NCEI Global Summary of the Year, element TAVG, stations USW00013994 (St. Louis), USW00013995 (Springfield) and USW00013987 (Joplin), 1950-2024; regression sums, residual autocorrelation and the effective-sample-size correction computed for this course; retrieved 2026-11-10 · Public domain (works of the United States government), with SLS original worked computation

Why this reading. The lesson says “technology gives” the slope, its standard error and the residual autocorrelation for two stations, and then asks you to fit the same model to a third station and to argue about whether the Springfield P-value can be trusted. All of that is arithmetic on the 75-year series printed in the Lesson 1.2 reading (St. Louis and Springfield). This page does the arithmetic once, shows every intermediate sum, prints a third station (Joplin) in full for the open question, and turns the autocorrelation caution into numbers. The argument in the lab starts here.

The sums, and what the year column contributes

For all three stations x is the year, 1950 to 2024, so x̄ = 1987 and Sxx = Σ(x − 1987)2 = 2(12 + 22 + … + 372) = 35,150, with sx = √(35,150/74) = 21.794 years. That is the same for every station, which is why SEb differs between them only through s, the residual standard deviation. From the least-squares formulas: b = Sxy/Sxx, and the sum of squared residuals is SSE = Syy − b·Sxy; then s = √(SSE/73), SEb = s/√Sxx, t = b/SEb on 73 df.

StationSxySyyb (°F/yr)SSEsSEbt95% interval for βrr2Lag-1 residual correlation
St. Louis1840.2260.2580.05235163.921.49850.007996.550.0364 to 0.06830.6080.3700.40
Springfield1064.6162.7430.03029130.501.33700.007134.250.0161 to 0.04450.4450.1980.29
Joplin854.0156.2220.02430135.471.36230.007273.340.0098 to 0.03880.3640.1330.32

Springfield, line by line. ȳ = 56.3680. Sxy = 1064.6, so b = 1064.6/35,150 = 0.03029°F per year. SSE = 162.743 − 0.03029 × 1064.6 = 130.50; s = √(130.50/73) = 1.3370; SEb = 1.3370/√35,150 = 0.00713; t = 0.03029/0.00713 = 4.25, df = 73. Table 2 of the critical-values reading, df-73 row: the tail beyond 4.0 is far below 0.001, so the two-sided P is below 0.001 (< 0.0001 from technology). Interval: 0.03029 ± 1.993(0.00713) = 0.0161 to 0.0445°F per year. The correlation: r = Sxy/√(SxxSyy) = 0.445.

Table 2. Annual mean temperature, Joplin Regional Airport, 1950–2024 (°F)

NOAA Global Summary of the Year, TAVG, station USW00013987. Retrieved 2026-11-10. The third station for the open question, in full so the sums can be checked.

Year°FYear°FYear°F
195056.0197556.8200058.1
195156.0197654.9200159.2
195258.8197757.4200258.6
195360.3197856.4200358.3
195461.2197955.5200458.6
195558.3198059.2200559.4
195659.0198158.1200660.6
195758.0198257.8200759.1
195856.4198357.1200857.4
195957.7198457.9200956.7
196056.3198556.7201058.3
196156.8198659.0201159.0
196258.1198758.7201261.5
196358.1198857.4201356.3
196457.6198955.7201456.1
196558.9199059.0201558.7
196656.2199158.9201660.3
196756.8199257.0201760.4
196855.8199356.3201858.1
196956.9199458.3201957.7
197057.1199558.1202058.9
197158.4199657.4202159.2
197257.8199757.3202258.9
197357.7199859.7202360.4
197456.5199959.4202461.4

Joplin’s slope, 0.0243°F per year (0.24°F per decade), is about half of St. Louis’s; t = 3.34, P = 0.0013; r2 = 0.133. Significant at 0.01 on the simple test. Hold that verdict until the next section.

The residuals are not independent: measuring it, and what it costs

The lag-1 autocorrelation of the residuals is Σeiei+1 / Σei2: multiply each residual by the next year’s and add, then divide by SSE. For St. Louis the residuals for 2013 through 2018 are -1.60, -2.25, +0.70, +2.24, +1.99, -0.86: a run of years on one side of the line. Warm spells last. For Joplin, 6 consecutive years (1952–1957) lie above the line. When residuals cluster like that, the 75 observations carry less independent information than 75, and the standard error of the slope is too small.

A standard rough correction for a series with lag-1 correlation ρ replaces n by an effective n(1 − ρ)/(1 + ρ), which inflates SEb by √((1 + ρ)/(1 − ρ)) and shrinks t by the same factor:

StationLag-1 correlation ρInflation factor √((1+ρ)/(1−ρ))Adjusted tAdjusted two-sided PEffective n
St. Louis0.401.534.29< 0.000132
Springfield0.291.353.140.00241
Joplin0.321.402.390.01938

St. Louis’s trend survives untouched in practice: t falls from 6.55 to 4.29, still far beyond 0.001. Springfield’s falls from 4.25 to 3.14, still significant at 0.01, with an interval widened to roughly 0.0111 to 0.0495°F per year. Joplin’s falls from 3.34 to 2.39: still significant at 0.05, no longer at 0.01 (P = 0.019), and its interval for β widens to 0.0040 to 0.0445, whose lower end is close to zero. The correction is a rough one, and different methods give somewhat different numbers. What does not change is the ordering: the St. Louis evidence is strong, the Springfield evidence solid, the Joplin evidence moderate.

For the claim under debate

“The Springfield test proves that Springfield has warmed.” Against: the interval for β is 0.0161 to 0.0445°F per year before correction and wider after; r2 = 0.198, so the trend accounts for about a fifth of the year-to-year variation; the residuals are autocorrelated (ρ = 0.29) and the simple P-value is an approximation; one instrument at one airport is not a city, and the station’s surroundings and equipment changed over 75 years; and a straight line imposes a constant rate on a record that might have gained its warmth in one period. For: the slope is positive at all three Missouri stations, the lower end of every interval, even after correction, is above zero for Springfield and St. Louis, and the two-era comparison in Lesson 3.4 gives a consistent answer by a different method. A good argument weighs these, says which carries more weight, and states what the data cannot decide.

Sources: NOAA NCEI, Global Summary of the Year data search (ncei.noaa.gov); NOAA NCEI Access Data Service (the query used here) (ncei.noaa.gov)

Reading

Seventy-Five Years at Two Missouri Airports, and the Normal Table to Read Them With

NOAA NCEI Global Summary of the Year, element TAVG, stations USW00013994 (St. Louis Lambert International Airport) and USW00013995 (Springfield-Branson National Airport), 1950-2024; standard normal table computed for this course; retrieved 2026-11-10 · Public domain (works of the United States government), with an SLS original computed table

Why this reading. The lesson fits a normal model to 75 years of St. Louis temperatures and reports how well it does: 49 of 75 years within one standard deviation, 8 years at 59°F or warmer. It does not print the 75 years, so you could not check a single one of those counts. Here they are, as NOAA publishes them, with a second station (Springfield) beside them for the open question, and the standard normal table the lesson’s areas are read from. Every figure the lesson gets from a spreadsheet can be recovered from this page.

Table 1. Standard normal areas Φ(z) = P(Z < z)

SLS original, computed for this course. Read the first decimal of z down the left column and the second across the top: Φ(1.26) is in row 1.2, column .06. For negative z use symmetry, Φ(−z) = 1 − Φ(z). The area to the right of z is 1 − Φ(z).

z.00.01.02.03.04.05.06.07.08.09
0.00.50000.50400.50800.51200.51600.51990.52390.52790.53190.5359
0.10.53980.54380.54780.55170.55570.55960.56360.56750.57140.5753
0.20.57930.58320.58710.59100.59480.59870.60260.60640.61030.6141
0.30.61790.62170.62550.62930.63310.63680.64060.64430.64800.6517
0.40.65540.65910.66280.66640.67000.67360.67720.68080.68440.6879
0.50.69150.69500.69850.70190.70540.70880.71230.71570.71900.7224
0.60.72570.72910.73240.73570.73890.74220.74540.74860.75170.7549
0.70.75800.76110.76420.76730.77040.77340.77640.77940.78230.7852
0.80.78810.79100.79390.79670.79950.80230.80510.80780.81060.8133
0.90.81590.81860.82120.82380.82640.82890.83150.83400.83650.8389
1.00.84130.84380.84610.84850.85080.85310.85540.85770.85990.8621
1.10.86430.86650.86860.87080.87290.87490.87700.87900.88100.8830
1.20.88490.88690.88880.89070.89250.89440.89620.89800.89970.9015
1.30.90320.90490.90660.90820.90990.91150.91310.91470.91620.9177
1.40.91920.92070.92220.92360.92510.92650.92790.92920.93060.9319
1.50.93320.93450.93570.93700.93820.93940.94060.94180.94290.9441
1.60.94520.94630.94740.94840.94950.95050.95150.95250.95350.9545
1.70.95540.95640.95730.95820.95910.95990.96080.96160.96250.9633
1.80.96410.96490.96560.96640.96710.96780.96860.96930.96990.9706
1.90.97130.97190.97260.97320.97380.97440.97500.97560.97610.9767
2.00.97720.97780.97830.97880.97930.97980.98030.98080.98120.9817
2.10.98210.98260.98300.98340.98380.98420.98460.98500.98540.9857
2.20.98610.98640.98680.98710.98750.98780.98810.98840.98870.9890
2.30.98930.98960.98980.99010.99040.99060.99090.99110.99130.9916
2.40.99180.99200.99220.99250.99270.99290.99310.99320.99340.9936
2.50.99380.99400.99410.99430.99450.99460.99480.99490.99510.9952
2.60.99530.99550.99560.99570.99590.99600.99610.99620.99630.9964
2.70.99650.99660.99670.99680.99690.99700.99710.99720.99730.9974
2.80.99740.99750.99760.99770.99770.99780.99790.99790.99800.9981
2.90.99810.99820.99820.99830.99840.99840.99850.99850.99860.9986
3.00.99870.99870.99870.99880.99880.99890.99890.99890.99900.9990
3.10.99900.99910.99910.99910.99920.99920.99920.99920.99930.9993
3.20.99930.99930.99940.99940.99940.99940.99940.99950.99950.9995
3.30.99950.99950.99950.99960.99960.99960.99960.99960.99960.9997
3.40.99970.99970.99970.99970.99970.99970.99970.99970.99970.9998

Critical values you will use for the rest of the course:

Confidence level80%90%95%98%99%
z* (two-sided)1.2821.6451.9602.3262.576
Φ(z*)0.90000.95000.97500.99000.9950

Table 2. Annual mean temperature (°F), 1950–2024, two NOAA stations

NOAA NCEI Global Summary of the Year, element TAVG, stations USW00013994 (St. Louis Lambert International Airport) and USW00013995 (Springfield-Branson National Airport, Missouri). Retrieved 2026-11-10. No year is missing at either station.

YearSt. LouisSpringfieldYearSt. LouisSpringfieldYearSt. LouisSpringfield
195054.454.3197555.355.7200056.155.9
195154.454.2197654.055.0200157.756.6
195257.256.7197755.456.9200257.856.2
195358.558.2197853.253.9200356.555.8
195458.558.7197954.153.4200457.556.7
195557.256.1198056.457.1200558.057.8
195657.056.9198156.157.1200658.558.8
195756.356.0198254.856.3200758.358.1
195854.154.2198355.956.0200855.455.1
195956.355.5198456.456.7200956.655.2
196054.654.7198555.255.1201058.056.5
196153.854.7198657.657.1201158.757.0
196254.655.5198758.257.2201261.259.5
196354.357.2198856.955.3201356.455.4
196456.056.7198955.753.9201455.855.3
196556.056.9199059.057.7201558.857.8
196654.154.5199159.257.5201660.459.4
196754.655.3199257.156.3201760.259.1
196854.554.2199355.554.9201857.457.6
196954.655.1199457.757.1201957.257.4
197055.054.6199557.056.2202058.257.2
197156.256.7199654.954.3202159.057.7
197254.655.8199755.154.4202257.857.1
197355.956.6199858.757.9202360.459.1
197454.855.7199957.957.6202461.259.7

Check the lesson’s St. Louis figures

Sum of the 75 St. Louis values: 4247.9, so x̄ = 4247.9/75 = 56.6387, which rounds to the lesson’s 56.64. The sample standard deviation is 1.8754 (rounds to 1.88). Now count the years:

IntervalBounds (°F)Years insidePercentNormal model
μ ± 1σ54.76 to 58.5149 of 7565.3%68.3%
μ ± 2σ52.89 to 60.3971 of 7594.7%95.4%
μ ± 3σ51.01 to 62.2675 of 75100%99.7%

Years at 59°F or warmer: 1990 (59.0), 1991 (59.2), 2012 (61.2), 2016 (60.4), 2017 (60.2), 2021 (59.0), 2023 (60.4), 2024 (61.2); 8 of 75, or 0.107. The model: z = (59 − 56.64)/1.88 = 1.26, and Table 1 gives Φ(1.26) = 0.8962, so P(X ≥ 59) = 0.1038. The 90th percentile of the data: 0.9 × 75 = 67.5, and the 68th and 69th values in sorted order are both 59.0, so the data’s 90th percentile is 59.0°F, against the model’s 56.64 + 1.2816(1.88) = 59.04°F. The data’s 10th percentile (the 8th value in sorted order) is 54.4°F, against the model’s 54.24°F.

The open question, worked for the second station

Springfield, the same 75 years: sum 4227.6, x̄ = 56.37°F, s = 1.483°F, median 56.5. Within 1, 2 and 3 SDs: 52, 71 and 75 of 75 (69.3%, 94.7%, 100%), against 68–95–99.7. The counts pass. Choose a threshold: years at 59°F or warmer. Model: z = (59 − 56.37)/1.483 = 1.77, area to the right 0.038, so about 2.8 years expected. Data: 5 of 75.

Now look at the years, not just the pile. Springfield’s first thirty years (1950–1979) never exceed 58.7°F, and 7 of the 30 years from 1995 to 2024 reach 58°F or more, against 2 of the first 30. The percentages pass because a normal model only asks how values are spread, not when they happened. A model of “a typical Springfield year” fitted to 1950–2024 describes a mixture of a cooler era and a warmer one, and it will predict the next year badly if the drift continues. The same is true of St. Louis, where the mean of 1950–1979 is 55.3°F and of 1995–2024 is 57.9°F (Unit 2 and Unit 4 return to this).

If you would rather do the open question on a station that is not printed here, the source line above gives the NOAA query; change the station code. The two printed stations are enough to complete the lesson.

Sources: NOAA NCEI, Global Summary of the Year data search (ncei.noaa.gov); NOAA NCEI Access Data Service (the query used here) (ncei.noaa.gov)

Up NextComplete & Continue →

Leave a Reply Cancel reply

Chi-Square Tests for Two-Way Tables
Prev
Choosing the Right Procedure and Writing It Up
Next

© 2026 The VR School | Career content by Wiley | Powered by Sejal Learning Systems.

Join free · Accreditation · Privacy Policy · Terms of Service

Modal title

Main Content