Comprehensive AP Statistics Study Guide: Univariate Data, Normal Models, Observational Design, Categorical Analysis, and Linear Transformations

Normal Distributions and Standardization

  • City Temperature Models:

    • Phoenix, Arizona daily high summer temperatures are modeled as normal with mean μ=106 ∘F\mu = 106\,^\circ\text{F} and standard deviation σ=6 ∘F\sigma = 6\,^\circ\text{F}.
    • Seattle, Washington daily high summer temperatures are modeled as normal with mean μ=76 ∘F\mu = 76\,^\circ\text{F} and standard deviation σ=7 ∘F\sigma = 7\,^\circ\text{F}.
  • Relative Extremity Comparison:

    • On a specific day, Phoenix reached 115 ∘F115\,^\circ\text{F} and Seattle reached 88 ∘F88\,^\circ\text{F}.
    • Standardized score (ZZ-score) formula:     Z=x−μσZ = \frac{x - \mu}{\sigma}
    • Phoenix ZZ-score calculation:     ZP=115−1066=96=1.50Z_P = \frac{115 - 106}{6} = \frac{9}{6} = 1.50
    • Seattle ZZ-score calculation:     ZS=88−767=127≈1.71Z_S = \frac{88 - 76}{7} = \frac{12}{7} \approx 1.71
    • Cumulative distribution upper tail probabilities:     normal cdf(115,∞,106,6)=0.0668\text{normal cdf}(115, \infty, 106, 6) = 0.0668normal cdf(88,∞,76,7)=0.0432\text{normal cdf}(88, \infty, 76, 7) = 0.0432
    • Seattle experienced the relatively more unusually hot day because its high temperature was 1.711.71 standard deviations above its mean, compared to Phoenix which was 1.501.50 standard deviations above its mean.
  • Percentile Determination:

    • Lower cumulative area for Phoenix at 115 ∘F115\,^\circ\text{F}:     normal cdf(−∞,115,106,6)=0.933\text{normal cdf}(-\infty, 115, 106, 6) = 0.933
    • A temperature of 115 ∘F115\,^\circ\text{F} in Phoenix corresponds to the 93.3rd93.3\text{rd} percentile.
  • Equivalent Relative Temperature:

    • To match Seattle's relative standing (Z=1.71Z = 1.71), solve for Phoenix temperature xx:     1.71=x−10661.71 = \frac{x - 106}{6}x−106=1.71×6x - 106 = 1.71 \times 6x=10.26+106=116.26 ∘Fx = 10.26 + 106 = 116.26\,^\circ\text{F}
  • Anomalous Temperature Threshold:

    • An anomalous temperature falls in the upper 2%2\% of daily high temperatures (lower area = 0.980.98).
    • Inverse normal calculation:     invNorm(0.98,106,6)=118.32 ∘F\text{invNorm}(0.98, 106, 6) = 118.32\,^\circ\text{F}
    • Phoenix must reach a high temperature of 118.32 ∘F118.32\,^\circ\text{F} to be classified as anomalous.

Univariate Quantitative Data Analysis and Outliers

  • Caffeine Content Dataset:

    • Sampling of caffeine amounts (in milligrams per 12 ounce12\,\text{ounce} serving) across 1212 popular beverages (soft drinks, energy drinks, tea, coffee):     7.5 mg,34 mg,35 mg,38 mg,40 mg,45 mg,55 mg,70 mg,72 mg,80 mg,105 mg,165 mg7.5\,\text{mg}, 34\,\text{mg}, 35\,\text{mg}, 38\,\text{mg}, 40\,\text{mg}, 45\,\text{mg}, 55\,\text{mg}, 70\,\text{mg}, 72\,\text{mg}, 80\,\text{mg}, 105\,\text{mg}, 165\,\text{mg}
  • Summary Statistics and Outlier Thresholds:

    • Minimum (Min\text{Min}): 7.5 mg7.5\,\text{mg}
    • First Quartile (Q1Q_1): 35 mg35\,\text{mg}
    • Median (Med\text{Med}): 50 mg50\,\text{mg}
    • Third Quartile (Q3Q_3): 76 mg76\,\text{mg}
    • Maximum (Max\text{Max}): 165 mg165\,\text{mg}
    • Interquartile Range (IQR\text{IQR}):     IQR=Q3−Q1=76−36.5=39.5 mg\text{IQR} = Q_3 - Q_1 = 76 - 36.5 = 39.5\,\text{mg}
    • Lower Fence (LB\text{LB}):     LB=Q1−1.5×IQR=36.5−1.5×39.5=22.75 mg\text{LB} = Q_1 - 1.5 \times \text{IQR} = 36.5 - 1.5 \times 39.5 = 22.75\,\text{mg}
    • Upper Fence (UB\text{UB}):     UB=Q3+1.5×IQR=76+1.5×39.5=135.25 mg\text{UB} = Q_3 + 1.5 \times \text{IQR} = 76 + 1.5 \times 39.5 = 135.25\,\text{mg}
    • Identified Outlier: 165 mg165\,\text{mg} exceeds the upper threshold of 135.25 mg135.25\,\text{mg}.
  • Distribution Description Framework (SCCS):

    • Center: The median caffeine content is 50 mg50\,\text{mg}.
    • Shape: The distribution is slightly right-skewed.
    • Spread: The interquartile range (IQR\text{IQR}) is 39.5 mg39.5\,\text{mg}.
    • Outliers: There is one high outlier located at 165 mg165\,\text{mg}.

Observational Study Design and Classification of Variables

  • Study Overview:

    • Conducting Institution: Researchers at the University of California, San Francisco.
    • Sample Size: n=12,000n = 12{,}000 preteens across the United States.
    • Target Population: All preteens (NN) in the United States.
    • Initial Age: 9−109-10 years old.
    • Longitudinal Duration: Observed annually over a period of 33 years.
  • Research Focus and Classification:

    • Investigative Question: Determine whether social media use is associated with depression in preteens.
    • Study Type: Prospective study, as researchers tracked subjects forward in time across 33 consecutive years.
    • Variable 1: Time spent using social media per day — Quantitative variable.
    • Variable 2: Level of depressive symptoms — Categorical variable.
  • Empirical Findings:

    • Average daily social media usage increased substantially with age, rising from approximately 7 minutes/day7\,\text{minutes/day} to greater than 70 minutes/day70\,\text{minutes/day}.
    • Substantial year-over-year increases in social media consumption were followed by larger increases in depressive symptoms in subsequent years.
    • Conclusion: Elevated social media usage serves as a potential risk factor for future depressive symptoms.

Categorical Bivariate Data and Association

  • Voter Alignment Survey Data:

    • Sample size: N=550N = 550 voters surveyed regarding political party alignment and preferred voting method.
  • Two-Way Contingency Table:

    • Party K:
    • Election Day: 4545
    • Advance: 145145
    • Absentee by mail: 6060
    • Row Total: 250250
    • Party M:
    • Election Day: 9090
    • Advance: 150150
    • Absentee by mail: 6060
    • Row Total: 300300
    • Column Totals:
    • Election Day Total: 135135
    • Advance Total: 295295
    • Absentee by mail Total: 120120
    • Grand Total: 550550
  • Explanatory vs. Response Variables:

    • Explanatory Variable: Voting method (analyzed as the factor explaining voter party distribution due to targeted campaign strategies).
  • Conditional Relative Frequencies (Segmented Bar Graph Calculations):

    • Election Day Column (Total=135\text{Total} = 135):
    • Party K: 45135≈0.333\frac{45}{135} \approx 0.333 (33.3%33.3\%)
    • Party M: 90135≈0.667\frac{90}{135} \approx 0.667 (66.7%66.7\%)
    • Advance Column (Total=295\text{Total} = 295):
    • Party K: 145295≈0.492\frac{145}{295} \approx 0.492 (49.2%49.2\%)
    • Party M: 150295≈0.508\frac{150}{295} \approx 0.508 (50.8%50.8\%)
    • Absentee by Mail Column (Total=120\text{Total} = 120):
    • Party K: 60120=0.500\frac{60}{120} = 0.500 (50.0%50.0\%)
    • Party M: 60120=0.500\frac{60}{120} = 0.500 (50.0%50.0\%)
  • Association Assessment:

    • An association exists between political party and voting method.
    • Party K members disproportionately utilize Absentee by mail (50.0%50.0\%) and Advance voting (49.2%49.2\%) compared to Election Day (33.3%33.3\%).
    • Party M members represent a substantially higher proportion of Election Day voters (66.7%66.7\%).

Histogram Features, Descriptive Measures, and Linear Transformations

  • Meal Price Histogram Distribution (N=100N = 100 Responses):

    • Bin Breakdown:
    • $5−$9\$5 - \$9: Frequency = 2222
    • $9−$13\$9 - \$13: Frequency = 2525
    • $13−$17\$13 - \$17: Frequency = 2020
    • $17−$21\$17 - \$21: Frequency = 1313
    • $21−$25\$21 - \$25: Frequency = 1111
    • $25−$29\$25 - \$29: Frequency = 88
    • $29−$33\$29 - \$33: Frequency = 55
    • $33−$37\$33 - \$37: Frequency = 11
    • $41−$45\$41 - \$45: Frequency = 11
  • Mean vs. Median in Skewed Distributions:

    • Resistance Property: The median is resistant to extreme values and outliers, whereas the mean is non-resistant.
    • Directional Impact: Right-skewed distributions contain high-value tails that pull the mean to the right, causing the mean to be strictly greater than the median (Mean>Median\text{Mean} > \text{Median}).
    • Estimation Requirement: Determining precise quantitative parameters from grouped frequency distributions requires evaluating interval midpoints rather than reading absolute values directly.
  • Comprehensive Distribution Characteristics:

    • Shape: Unimodal and skewed to the right.
    • Center: Centered within the $13−$15\$13 - \$15 interval.
    • Spread: Spans from approximately $5\$5 to $45\$45.
    • Unusual Features: Long right tail extending up to $45\$45.
  • Standard Deviation and Dispersion:

    • Standard deviation measures the average distance of data points from the center (xˉ\bar{x}).
    • Datasets with identical means (xˉ=5.5\bar{x} = 5.5) display different standard deviations depending on data dispersion:
    • Data tightly clustered around the mean yields a lower standard deviation.
    • Data spread broadly away from the mean yields a higher standard deviation.
  • Linear Transformations (Adding a Constant):

    • Initial Dataset Parameters: Mean μ=18\mu = 18, Standard Deviation σ=3\sigma = 3
    • Shift Transformation: Adding a constant k=10k = 10 to every individual value (xnew=x+10x_{\text{new}} = x + 10).
    • Effect on Measures of Center:
    • The mean increases linearly by the constant added:       μnew=μ+10=18+10=28\mu_{\text{new}} = \mu + 10 = 18 + 10 = 28
    • Effect on Measures of Spread:
    • Adding a constant shifts all data points equally, leaving the absolute distance between values unchanged.
    • The standard deviation is invariant to shifts in location:       σnew=σ=3\sigma_{\text{new}} = \sigma = 3