​ odule 4.1.

1 Data Management​
M
​-​ ​Practice​ ​of​ ​organizing​ ​and​ ​maintaining​
​data​ ​processes​ ​to​ ​meet​ ​ongoing​
​information lifecycle needs​

​ egend:​
L
​𝑥 = repeating value​
​𝑤 = number of occurrences of weight​
​𝑥 (w/ a line on top) = mean​

​ .​ M
2 ​ edian​
​-​ ​Middle​ ​score​ ​for​ ​a​ ​set​ ​of​​data​​that​​has​
​been arranged in order of magnitude​
​-​ ​Formula (location of median):​
​Measures of central tendency​ ​median = (n+1)/2​
​ .​ M
1 ​ ean​ ​Legend:​
​-​ ​The​ ​sum​ ​of​ ​all​ ​values​ ​in​ ​a​ ​dataset​ ​n = total number of data values in the sample​
​divided by the total number of values​ ​Note:​
​*if n is odd, median is the middle value​
​pros​ ​cons​
​*​ ​if​ ​n​ ​is​ ​even,​ ​median​ ​is​ ​the​ ​average​ ​of​ ​the​ ​2​
​ an be used with​
C I​t is susceptible to​ ​middle values​
​both discrete and​ ​the influence of​
​continuous data​ ​outliers​ ​ .​ M
3 ​ ode​
​-​ ​The​ ​most​ ​frequent​ ​score​ ​in​ ​the​ ​data​
​(what value pops up the most)​

​Type of variable (based​ ​Best measure​

​of level of measurement)​

​nominal​ ​mode​

​ordinal​ ​median​

​interval/ratio (not skewed)​ ​mean​

​interval/ratio (skewed)​ ​median​

​ .1 Arithmetic Mean (population mean)​
1
​-​ ​Average​ ​of​ ​a​ ​complete​ ​set​ ​of​ ​data​ ​ egend:​
L
​(population)​ ​Nominal = data can be categorized​
​1.2 Sample Mean​ ​Ordinal = can be categorized and ranked​
​-​ ​The​ ​average​ ​of​ ​a​ ​subset​ ​of​ ​data​ ​taken​ ​Interval​ ​=​ ​can​ ​be​ ​categorized,​ ​ranked,​ ​and​
​from a larger population (sample)​ ​evenly spaced​
​1.3 Weighted Mean​ ​Ratio​ ​=​ ​can​ ​be​ ​categorized,​ ​ranked,​ ​evenly​
​-​ ​An​ ​average​ ​computed​ ​by​ ​giving​ ​spaced, and has a natural zero​
​different​ ​weights​ ​to​ ​some​ ​of​ ​the​
​individual values.​
 ​Examples w/ explanation​
​-​ ​It is a non-negative​

​nominal​ ​Ethnicity​
​-​ ​Can't be ranked​

​ordinal​ ​Top 5 olympic medalist​

​-​ ​Does not tell you how close or far​
​they are in terms of number of wins​

​interval​ ​Temp in celsius​

​-​ ​There are equal intervals of one​
​degree, but the zero point is not​
​true as there that measurement can​
​reach negative degree celsius​ ​ egend:​
L
​ratio​ ​height​
​xi = each individual value​
​μ = population mean​
​xˉ = sample mean​
​Measures of variation​
​N = number of values in the population​
​-​ ​Gives​ ​information​ ​on​ ​the​ ​spread​ ​or​
​n = number of values in the sample​
​variability of the data values​
​-​ ​Although​ ​2​ ​dataset’s​ ​could​ ​have​ ​the​
​ teps on how to calculate:​
S
​same​ ​center,​ ​there​ ​variation​ ​could​ ​be​
​1.​ ​Calculate the mean​
​very different​
​2.​ ​Subtract the mean from each value​
​3.​ ​Square each deviation​
​4.​ ​Add up all the squared deviations​
​5.​ ​Divide by the number of values​
​5.1 if population, do it as is​
​5.2 for sample, do n-1​
​ .​ R
1 ​ ange​
​6.​ ​For standard deviation:​
​-​ ​Used to compare obvious data sets​
​6.1 square root the variance​
​-​ ​Formula:​​Range​​=​​highest​​value​​-​​lowest​
​value​
​4.​​Coefficient of variation​
​-​ ​Considered​ ​as​ ​the​ ​weakest​​measure​​of​
​-​ ​This​ ​kind​ ​of​​measure​​allows​​to​​or​​more​
​spread​
​distributions​ ​measured​ ​in​ ​same​ ​r​
​Note:​ ​heavily​ ​influenced​ ​by​ ​extreme​ ​values​
​different units to be compared​
​(outliers) and only compares 2 values​
​-​ ​How​ ​big​ ​is​ ​the​ ​standard​ ​deviation​
​compared to the mean​
​ .​ V
2 ​ ariance (​​σ​​2)​
​-​ ​Formula:​​CV =​​σ​/(x w/ a line on top) x 100 %​
​-​ ​A​ ​measure​ ​of​ ​how​​far​​a​​set​​of​​data​​are​
​Legend:​
​dispersed from the mean​
​σ = St Dev​
​-​ ​It​​is​​non-negative​​since​​each​​term​​in​​the​
​x w/ a line on top = mean​
​variance is squared​
​-​ ​All​​units​​is​​squared,​​ex:​​a​​set​​of​​weights​
​ ote:​​the​​lower​​the​​CV,​​the​​lesser​​the​​dispersion​
N
​in kg will be given in kg squared​
​of data values​

​ .​ S
3 ​ tandard deviation (​​σ)​
​5.​​Interquartile Range (IQR)​
​-​ ​Measured​​the​​deviation​​of​​data​​from​​its​
​-​ ​Defines​​the​​difference​​between​​the​​third​
​mean​
​and the first quartile​
 ​-​ ​ ormula:​ ​upper​ ​quartile​ ​-​​lower​​quartile​
F
​= q3 - q1​
​How to calculate the quartiles:​
​Note:​
​Q1 = 25% of data falls below this value​
​Q2 = 50% of the data falls below this value​
​Q3 = 75% of the data falls below this value​
​1.​ ​Arrange the data (smallest to largest)​
​2.​ ​Find q2 (median)​
​3.​ ​Find​ ​Q1​ ​(median​ ​of​ ​the​ ​lower​ ​half​ ​of​
​data)​
​4.​ ​Find​ ​Q3​ ​(median​ ​of​ ​the​ ​upper​ ​half​ ​of​
​data)​

​Correlation (Pearson)​
​-​ ​ easure​ ​describing​ ​the​ ​way​ ​two​
M
​variables vary together​
​-​ ​Statistic​​that​​measures​​the​​strength​​and​
​direction​ ​of​ ​a​ ​linear​ ​relationship​
​between two quantitative variables​
​-​ ​r represents the correlation coefficient​

​Size of Correlation​ ​Interpretation​

​.90 to 1 (-0.90 to -1)​ ​ ery high positive​

V
​(negative) correlation​

​.70 to .90 (-.70 to -.90)​ ​ igh positive (negative)​

H
​correlation​

​.50 to .70 (-.50 to -.70)​ ​ oderate positive​

M ​ odule 4.4 Simple Regression​
M
​(negative) correlation​
​-​ ​A​ ​statistical​ ​tool​ ​that​ ​is​ ​used​ ​in​ ​the​
​.30 to .50 (-.30 to -.50)​ ​ ow positive (negative)​
L ​quantification​ ​of​ ​the​ ​relationship​
​correlation​ ​between​ ​a​ ​single​ ​independent​ ​variable​
​.00 to .30 (.00 to -.30)​ ​Negligible correlation​ ​and​ ​a​ ​single​ ​dependent​ ​variable​ ​on​
​observations​​that​​have​​been​​carried​​out​
​in the past​

​Formula for regression: y = a + bx​

​Legend!​ ​ lternative Hypothesis (H₁ or Ha)​
A
​y = dependent variable​ ​-​ ​Represents what you aim to support or​
​x = independent variable​ ​prove​
​a = intercept (value of y when x = 0)​ ​-​ ​Indicates the presence of an effect,​
​b​ ​=​ ​slope​ ​(change​ ​in​ ​y​ ​for​ ​every​ ​1​ ​unit​ ​difference, or relationship​
​increase in x)​ ​-​ ​Always contains inequality ( ≠, >, < )​
​—----------------------------------------------------​ ​Purpose:​​proposed if H₀ is rejected​
​Formula for slope:​ ​Examples:​
​●​ ​H₁: μ ≠ 50 (the mean is not 50)​
​ = number of variables​
n ​●​ ​H₁: p < 0.7 (the proportion is less than​
​—----------------------------------------------------​ ​70%)​
​Formula for intercept a:​ ​Note: Direction depends on research​
​question​
​ and y (w/ a line): represents the means​
x ​●​ ​One-tailed: H₁ uses < or >​
​*Means​​must​​be​​calculated​​first​​before​​you​​can​
​get the intercept​ ​●​ ​Two-tailed: H₁ uses ≠​
​—----------------------------------------------------​
​Interpretation of Regression​ ​Steps in Hypothesis Testing​
​“For every 1 extra unit of smth, the dependent​ ​ .​ S
1 ​ tate H₀ and H₁​
​variable (y) will increase by (b). If independent​ ​2.​ ​Choose a significance level (α, usually​
​variable (x) stays at 0, the predicted value will be​ ​0.05)​
​(a)”​ ​3.​ ​Collect and analyze sample data​
​4.​ ​Compute test statistic (e.g., z, t)​
​Hypothesis Testing​ ​5.​ ​Compare with critical value or use​
​p-value​
-​ A method of making statistical decisions using​ ​6.​ ​Make a decision:​
​experimental data​ ​○​ ​If p-value ≤ α →​​Reject H₀​
​Used to test assumptions (claims) about a​ ​○​ ​If p-value > α →​​Fail to reject​
​population parameter based on sample data​ ​H₀​

​Null Hypothesis (H₀)​ ​ ype I and Type II Errors​

T
​-​ ​ epresents the default or status quo​
R ​-​ ​Types of incorrect decisions that may​
​assumption​ ​occur in hypothesis testing​
​-​ ​Assumes no effect, no difference, or no​ ​-​ ​Related to the truth or falsity of the null​
​relationship between variables​ ​hypothesis (H₀) and what decision is​
​-​ ​Always contains equality ( =, ≥, ≤ )​ ​made based on the data​
​Purpose:​​to test whether there's​
​enough evidence​​against​​it​ ​ ype I Error (α)​
T
​Examples:​ ​-​ ​Occurs when we reject the null​
​●​ ​H₀: μ = 50 (the population mean is 50)​ ​hypothesis (H₀) even though it is​
​●​ ​H₀: p ≥ 0.7 (the population proportion is​ ​actually true​
​at least 70%)​ -​ ​ ​Also called a false positive​
​Note: If evidence is strong, we reject H₀​ ​-​ ​We detect an effect that isn’t really there​
​Example:​ ​ inomial Distribution​
B
​-​ ​A person is diagnosed with a disease​ ​– Used when there are repeated trials, each with​
​(reject H₀) but is actually healthy (H₀ is​ ​two possible outcomes: success or failure​
​true)​ ​– The probability of success stays the same for​
​Controlled by: Significance level (α)​ ​each trial​
​– Common value: α = 0.05​ ​– Each trial is independent (one doesn’t affect​
​the others)​
​ ype II Error (β)​
T ​– Think: “yes or no” situations, repeated several​
​-​ ​Occurs when we fail to reject the null​ ​times​
​hypothesis (H₀) even though it is​
​actually false​ ​ xample:​
E
​-​ ​Also called a false negative​ ​– Flipping a coin 10 times and counting how​
​-​ ​We fail to detect an effect that is really​ ​many heads​
​there​ ​– Guessing on a multiple-choice quiz with 5​
​Example:​ ​questions​
​-​ ​A person is told they’re healthy (fail to​ ​Conditions to use binomial:​
​reject H₀) but actually has the disease​ ​– Fixed number of trials (n)​
​(H₀ is false)​ ​– Two outcomes: success/failure​
​Related to: Power of the test​ ​– Constant probability (p)​
​– Power = 1 - β​ ​– Trials are independent​
​Formula:​
​ rror Summary Table​
E
​Decision Made​
​– H₀ is True → Type I Error (α)​
​– H₀ is False → Type II Error (β)​
​– Reject H₀ → may lead to Type I Error if H₀ is​
​true​
​– Fail to Reject H₀ → may lead to Type II Error if​
​H₀ is false​

​ egend:​
L
​– H₀ = Null Hypothesis​ ​Poisson Distribution​
​– α = Probability of Type I Error​ –​ Used to count how often an event happens​
​– β = Probability of Type II Error​ ​over time or space​
​– Power = Probability of detecting a true effect​ ​– You don’t know the number of trials, but you​
​(1 − β)​ ​know the average rate​
​Note:​ ​– Events happen randomly and independently​
​– Lowering α reduces Type I errors but​ ​– Best for rare events​
​increases the chance of Type II errors​
​– Increasing sample size helps reduce both​ ​ xample:​
E
​errors​ ​- Number of emails received in an hour​
​– Cars arriving at a toll booth in a minute​
​– Number of errors in a book​
​Conditions to use Poisson:​
​– Events occur randomly and independently​
–​ Happen at a constant average rate (λ)​ ​Z-Score​
​– Based on time, area, volume, or distance​
–​ A Z-score tells you how far a data point is​
​from the mean, in terms of standard deviations​
​– Helps standardize different data values for​
​comparison​
​– A positive Z-score means the value is above​
​the mean​
​– A negative Z-score means the value is below​
​the mean​
​Hypergeometric Distribution​
–​ Used when sampling without replacement​
​from a group​
​– The probability of success changes with each​
​draw​
​– The trials are dependent on each other​
​– Useful when the population is small and​
​known​

​ xample:​
E
​- Drawing 5 cards from a deck without​
​replacement and counting the number of kings​
​– Selecting students from a class and counting​
​how many are seniors​
​– Picking colored balls from a bag and not​
​putting them back​ ​ teps to Calculate Z-score:​
S
​Conditions to use hypergeometric:​ ​ .​
1 ​Find the mean of the data​
​– Population is finite (N)​ ​2.​ ​Find the standard deviation​
​– Known number of successes in the​ ​3.​ ​Subtract the mean from the data point​
​population (K)​ ​4.​ ​Divide the result by the standard​
​– Drawing without replacement (n)​ ​deviation​