Normal Distribution and Standard Deviation — Study Notes

The Normal Curve and Its Significance

  • Many measured values form distributions that look unimodal, symmetric, and bell-shaped, i.e., resemble the normal curve. Examples discussed: daily cheeseburger sales at McDonald’s, weights of adult guinea pigs, and yearly rainfall in Bloomington. The shapes look similar even though the measured quantities are very different in real units.

  • Why these distributions share the same shape:

    • The world is noisy, introducing random fluctuations around a central tendency.

    • There is a central value (mean) around which most observations cluster; some days are busier or quieter due to random variation.

    • Extreme values occur infrequently, producing tails that taper off toward zero (or near-zero in the histogram).

  • Visual representations:

    • Histograms show a peak in the middle with symmetric tails.

    • A smooth normal curve can be drawn over the histogram as a mathematical representation of the data pattern.

  • Important takeaway: the normal curve is a well-known, frequently occurring shape, hence it is called the normal distribution or normal curve.

Normal Curve and the Role of Standard Deviation

  • The normal curve is a very specific shape with a precise mathematical definition (though the exact formula is not required here).

  • We often discuss the normal curve in relation to the standard deviation, denoted by σ\sigma (population) or ss (sample).

  • On the horizontal axis, we frequently view distance from the mean in units of standard deviation:

    • One standard deviation to the right is extmean+σext{mean} + \sigma; two standard deviations to the right is extmean+2σext{mean} + 2\sigma; three standard deviations to the right is extmean+3σext{mean} + 3\sigma.

    • The distribution is symmetric, so the same applies to the left side (negative direction).

  • Practical geometric insight:

    • At about one standard deviation away from the mean, the curve has descended to roughly halfway down the peak.

    • At about three standard deviations, the tails are near the far ends of the curve; for a practical view, the values beyond three standard deviations are very rare.

  • This “three-sigma” idea helps in estimating the spread of a distribution when you assume normality, even if you don’t know the exact formula.

Rule-of-Thumb: Estimating the Standard Deviation from a Roughly Normal Distribution

  • If a distribution is roughly normal, you can estimate the mean (center) and the standard deviation from the spread to the tails.

  • Step 1: Find the mean (for a symmetric unimodal distribution, it lies at the center).

  • Step 2: Identify how far you have to go from the mean to reach the tails where values are very small (i.e., where the curve is near the tail).

  • Step 3: Use the fact that in a normal distribution, these tail distances are about three standard deviations, so

    • If the tail distance is DD, then the estimated standard deviation is approximately racD3rac{D}{3}.

  • Example 1:

    • A distribution has mean around extmean=50ext{mean} = 50.

    • Tails drop to near-zero around values 8080 (positive) and 2020 (negative).

    • Tail distance from the mean is roughly D=80−50=30D = 80 - 50 = 30 (positive side) and 50−20=3050 - 20 = 30 (negative side).

    • Since the tails are about three standard deviations away, 3σ≈D=303\sigma \approx D = 30, so

    • σ≈rac303=10\sigma \approx rac{30}{3} = 10.

  • Example 2:

    • A distribution with mean around extmean=100ext{mean} = 100.

    • Tails reach near-zero around values 145145 (positive) and 5555 (negative).

    • Tail distance from the mean is about D = 45\

    • 3\sigma \approx D = 45,so</p></li><li><p>, so</p></li><li><p>\sigma \approx rac{45}{3} = 15.</p></li></ul></li><li><p>Importantcaveat:</p><ul><li><p>Thisisarule−of−thumbforestimating.</p></li></ul></li><li><p>Important caveat:</p><ul><li><p>This is a rule-of-thumb for estimating\sigmawhenthedataareroughlynormallyshaped.</p></li><li><p>Itisnottheformaldefinitionofstandarddeviation.Thestandarddeviationistheaveragedistancefromthemeantoadatapoint,whichisdefinedmorepreciselybythestandarddeviationformulas(seebelow).</p></li><li><p>Themethodreliesonnormality;itdoesnotworkwellforhighlyskewed,bimodal,orotherwisenon−normaldistributions.</p></li></ul></li><li><p>Quickreminder:thedistance−to−tailinterpretationisapracticaltoolforquickestimation,notarigorouscalculation.</p></li></ul><h3id="fb0ac40c−1447−4b82−aaed−a52b0fecf894"data−toc−id="fb0ac40c−1447−4b82−aaed−a52b0fecf894"collapsed="false"seolevelmigrated="true">ClarifyingtheDefinitionofStandardDeviation</h3><ul><li><p>Thestandarddeviationmeasuresthespreadofdataaroundthemean.Aprecise,formaldefinition(population)is:</p><ul><li><p>when the data are roughly normally shaped.</p></li><li><p>It is not the formal definition of standard deviation. The standard deviation is the average distance from the mean to a data point, which is defined more precisely by the standard deviation formulas (see below).</p></li><li><p>The method relies on normality; it does not work well for highly skewed, bimodal, or otherwise non-normal distributions.</p></li></ul></li><li><p>Quick reminder: the distance-to-tail interpretation is a practical tool for quick estimation, not a rigorous calculation.</p></li></ul><h3 id="fb0ac40c-1447-4b82-aaed-a52b0fecf894" data-toc-id="fb0ac40c-1447-4b82-aaed-a52b0fecf894" collapsed="false" seolevelmigrated="true">Clarifying the Definition of Standard Deviation</h3><ul><li><p>The standard deviation measures the spread of data around the mean. A precise, formal definition (population) is:</p><ul><li><p>\sigma = \sqrt{\frac{1}{N} \sum{i=1}^{N} (xi - \\mu)^2}<br>where<br>where\\muisthepopulationmean.</p></li></ul></li><li><p>Forsamples,thestandarddeviationis:</p><ul><li><p>is the population mean.</p></li></ul></li><li><p>For samples, the standard deviation is:</p><ul><li><p>s = \sqrt{\frac{1}{n-1} \sum{i=1}^{n} (xi - \bar{x})^2}<br>where<br>where\bar{x}isthesamplemean.</p></li></ul></li><li><p>Theearlierthree−sigmaruleandthe“one−thirdofthetaildistance”heuristicarepracticaltoolsbuiltonthenormalshape,notthefundamentaldefinitionofis the sample mean.</p></li></ul></li><li><p>The earlier three-sigma rule and the “one-third of the tail distance” heuristic are practical tools built on the normal shape, not the fundamental definition of\sigma.</p></li></ul><h3id="9a3cbf89−5a93−4f24−bdf1−0f611d5723b3"data−toc−id="9a3cbf89−5a93−4f24−bdf1−0f611d5723b3"collapsed="false"seolevelmigrated="true">WorkedExamplesandPracticeScenarios</h3><ul><li><p>Reviewofaroughlynormaldistributiontoestimatethestandarddeviationusingthemeanandtaildistance,applyingtheruleofthree:</p><ul><li><p>Findthemean(centerofthedistribution).</p></li><li><p>Determinethedistancetothetailswherevaluesbecomeverysmall.</p></li><li><p>Estimate.</p></li></ul><h3 id="9a3cbf89-5a93-4f24-bdf1-0f611d5723b3" data-toc-id="9a3cbf89-5a93-4f24-bdf1-0f611d5723b3" collapsed="false" seolevelmigrated="true">Worked Examples and Practice Scenarios</h3><ul><li><p>Review of a roughly normal distribution to estimate the standard deviation using the mean and tail distance, applying the rule of three:</p><ul><li><p>Find the mean (center of the distribution).</p></li><li><p>Determine the distance to the tails where values become very small.</p></li><li><p>Estimate\sigma \\approx \frac{Distance}{3}.</p></li></ul></li><li><p>Exerciseprompts(asdiscussedinthematerial):</p><ul><li><p>Givenadistributionthatappearsroughlynormal,estimate.</p></li></ul></li><li><p>Exercise prompts (as discussed in the material):</p><ul><li><p>Given a distribution that appears roughly normal, estimate\sigmabyidentifyingthemeanandthetaildistance.</p></li><li><p>Ifthemeanisatacertainvalueandthetailsreachtoacertaindistance,usethethree−sigmaruletoestimatethespread.</p></li></ul></li></ul><h3id="0bbf9811−cf05−4811−b10f−14d5444ab5fd"data−toc−id="0bbf9811−cf05−4811−b10f−14d5444ab5fd"collapsed="false"seolevelmigrated="true">AreasUndertheNormalCurve:QuickRulesofThumb</h3><ul><li><p>Basicsymmetryfact:</p><ul><li><p>Theareaabovethemeanis50by identifying the mean and the tail distance.</p></li><li><p>If the mean is at a certain value and the tails reach to a certain distance, use the three-sigma rule to estimate the spread.</p></li></ul></li></ul><h3 id="0bbf9811-cf05-4811-b10f-14d5444ab5fd" data-toc-id="0bbf9811-cf05-4811-b10f-14d5444ab5fd" collapsed="false" seolevelmigrated="true">Areas Under the Normal Curve: Quick Rules of Thumb</h3><ul><li><p>Basic symmetry fact:</p><ul><li><p>The area above the mean is 50% of the distribution.</p></li><li><p>The area below the mean is also 50%.</p></li><li><p>Therefore, the area within one standard deviation of the mean (i.e., from ext{mean} - \sigmatoto ext{mean} + \sigma)isabout) is about0.68(i.e.,68(i.e., 68%).</p></li></ul></li><li><p>Piece-by-piece areas (approximately):</p><ul><li><p>From the mean up to one standard deviation above the mean:P( ext{mean} \le X \le ext{mean} + \sigma) \approx 0.34</p></li><li><p>Fromonestandarddeviationabovethemeantotwostandarddeviationsabovethemean:</p></li><li><p>From one standard deviation above the mean to two standard deviations above the mean:P( ext{mean} + \sigma \le X \le ext{mean} + 2\sigma) \approx 0.14</p></li><li><p>Beyondtwostandarddeviationsabovethemean(tothetail):</p></li><li><p>Beyond two standard deviations above the mean (to the tail):P(X \ge ext{mean} + 2\sigma) \approx 0.02</p></li></ul></li><li><p>Bysymmetryonthenegativeside:</p><ul><li><p>Fromthemeandowntoonestandarddeviationbelowthemean:</p></li></ul></li><li><p>By symmetry on the negative side:</p><ul><li><p>From the mean down to one standard deviation below the mean:P( ext{mean} - \sigma \le X \le ext{mean}) \approx 0.34</p></li><li><p>Fromtwostandarddeviationsbelowthemeantoonestandarddeviationbelow:</p></li><li><p>From two standard deviations below the mean to one standard deviation below:P( ext{mean} - 2\sigma \le X \le ext{mean} - \sigma) \approx 0.14</p></li><li><p>Beyondtwostandarddeviationsbelowthemean:</p></li><li><p>Beyond two standard deviations below the mean:P(X \le ext{mean} - 2\sigma) \approx 0.02</p></li></ul></li><li><p>Consolidatedfamiliarpercentages(memorytips):</p><ul><li><p>Withinonestandarddeviation:</p></li></ul></li><li><p>Consolidated familiar percentages (memory tips):</p><ul><li><p>Within one standard deviation:P(|X - \\mu| \le \\sigma) \approx 0.68</p></li><li><p>Withintwostandarddeviations:</p></li><li><p>Within two standard deviations:P(|X - \\mu| \le 2\sigma) \approx 0.95</p></li><li><p>Withinthreestandarddeviations:</p></li><li><p>Within three standard deviations:P(|X - \\mu| \le 3\sigma) \approx 0.997</p></li></ul></li><li><p>Summaryofthethreekeyvaluestomemorize(fromthecontent):</p><ul><li><p>34</p></li></ul></li><li><p>Summary of the three key values to memorize (from the content):</p><ul><li><p>34%, 14%, and 2% are the portions between the mean and the positive (or negative) direction for the intervals [mean, mean+σ], [mean+σ, mean+2σ], and [mean+2σ, tail], respectively. These add up on each side to give approximately 68% within ±1σ, and about 95% within ±2σ, etc.</p></li></ul></li><li><p>Note: These area values are exact for the ideal normal distribution; in practice, they are good approximations for data that are approximately normal.</p></li></ul><h3 id="ec80f4f2-1283-4e6e-963a-4138bcfbd47d" data-toc-id="ec80f4f2-1283-4e6e-963a-4138bcfbd47d" collapsed="false" seolevelmigrated="true">Practical Implications and Real-World Relevance</h3><ul><li><p>The normal distribution serves as a central model in statistics because many real-world processes, especially sums of independent random effects, tend to be approximately normal (per the intuition from the Central Limit Theorem).</p></li><li><p>The standard deviation is a key descriptor of spread: it tells you how far, on average, observations lie from the mean.</p></li><li><p>The rule-of-thumb estimation of\sigmafromthetaildistanceishandyforquickjudgmentandroughcomparisons,butrememberitreliesonapproximatenormalityandisnotaformalcalculation.</p></li><li><p>Ethicalandpracticalimplications:</p><ul><li><p>Whenusingnormal−curveassumptionsininference(e.g.,confidenceintervals,hypothesistests),checkfornormalityorapplyrobustmethodsifdataareskewedormultimodal.</p></li><li><p>Misusingnormal−curveassumptionsonnon−normaldatacanleadtobiasedconclusionsoroverstatedprecision.</p></li></ul></li></ul><h3id="198afc0e−ae25−4b64−a6f8−6c00f9fa32c2"data−toc−id="198afc0e−ae25−4b64−a6f8−6c00f9fa32c2"collapsed="false"seolevelmigrated="true">QuickReference:KeyTakeaways</h3><ul><li><p>Normaldistributionfeatures:</p><ul><li><p>Unimodal,symmetric,bell−shapedcurve;smoothrepresentationisthenormalcurve.</p></li><li><p>Centraltendencyaroundthemeanwithspreadgovernedbythestandarddeviationfrom the tail distance is handy for quick judgment and rough comparisons, but remember it relies on approximate normality and is not a formal calculation.</p></li><li><p>Ethical and practical implications:</p><ul><li><p>When using normal-curve assumptions in inference (e.g., confidence intervals, hypothesis tests), check for normality or apply robust methods if data are skewed or multimodal.</p></li><li><p>Misusing normal-curve assumptions on non-normal data can lead to biased conclusions or overstated precision.</p></li></ul></li></ul><h3 id="198afc0e-ae25-4b64-a6f8-6c00f9fa32c2" data-toc-id="198afc0e-ae25-4b64-a6f8-6c00f9fa32c2" collapsed="false" seolevelmigrated="true">Quick Reference: Key Takeaways</h3><ul><li><p>Normal distribution features:</p><ul><li><p>Unimodal, symmetric, bell-shaped curve; smooth representation is the normal curve.</p></li><li><p>Central tendency around the mean with spread governed by the standard deviation\sigma.</p></li></ul></li><li><p>Estimatingspreadfromaroughlynormalshape:</p><ul><li><p>Identifythemean.</p></li></ul></li><li><p>Estimating spread from a roughly normal shape:</p><ul><li><p>Identify the mean\\muandtheapproximatedistancetothetails;estimateand the approximate distance to the tails; estimate\sigma \approx \frac{Distance}{3}(ruleofthumb;nottheformaldefinition).</p></li></ul></li><li><p>Formaldefinitions(forreference):</p><ul><li><p>Populationstandarddeviation:(rule of thumb; not the formal definition).</p></li></ul></li><li><p>Formal definitions (for reference):</p><ul><li><p>Population standard deviation:\sigma = \sqrt{\frac{1}{N} \sum{i=1}^{N} (xi - \mu)^2}</p></li><li><p>Samplestandarddeviation:</p></li><li><p>Sample standard deviation:s = \sqrt{\frac{1}{n-1} \sum{i=1}^{n} (xi - \bar{x})^2}</p></li></ul></li><li><p>Areasunderthenormalcurve(practicalrules):</p><ul><li><p>Abovethemean:</p></li></ul></li><li><p>Areas under the normal curve (practical rules):</p><ul><li><p>Above the mean:0.5;within±1σ:; within ±1σ:0.68;betweenmeanand+1σ:; between mean and +1σ:0.34;between+1σand+2σ:; between +1σ and +2σ:0.14;beyond+2σ:; beyond +2σ:0.02;symmetriconthenegativeside.</p></li><li><p>Therefore:; symmetric on the negative side.</p></li><li><p>Therefore:P(|X - \\mu| \le \sigma) \approx 0.68,,P(|X - \\mu| \le 2\sigma) \approx 0.95,,P(|X - \\mu| \le 3\sigma) \approx 0.997$$.