Descriptive Statistics

📚 04 PSM & Preventive Pediatrics • 🎯 31 High-Yield MCQs • Nelson 22nd Edition Standard

📊 Vital Statistics & Demographic Data Sources

  • Annual Demographic Data Source: ((((STAT001)) · ((STAT005)))): The Sample Registration System (SRS) conducted by the Registrar General of India provides continuous, yearly national-level estimates for Crude Birth Rate (CBR), Crude Death Rate (CDR), and Infant Mortality Rate (IMR).
  • Multifactorial & Nutritional Survey: ((((STAT001)) · ((STAT005)))): The National Family Health Survey (NFHS) operates on a 5-year cycle and assesses fertility, nutrition, ANC coverage, and family planning; it is not an annual source.
  • Registration Completeness & Vital Events: ((STAT005)): The Civil Registration System (CRS) is the official mechanism used to assess the completeness of birth and death registrations.
  • Targeted Surveillance: ((STAT005)): Sentinel surveillance is utilized specifically to monitor longitudinal disease trends (e.g., HIV prevalence) among designated high-risk groups. ---

📐 Sampling Techniques & Study Variables

  • Stratified Random Sampling: ((STAT002)): Indicated when a population contains distinct subgroups (strata) based on a variable closely linked to the outcome (e.g., religion and food habits); samples are drawn proportionate to stratum size to ensure representation.
  • Variable Classifications: ((STAT006)):
  • Continuous Variables: : Measurements like fasting blood glucose, birth weight, and mid-arm circumference that can take any value along a continuum.
  • Count / Discrete Variables: : Whole-number values like the number of emergency department visits in a year (0, 1, 2...) requiring discrete probability models such as Poisson or negative binomial regression.
  • Scales of Measurement: ((STAT007)): Categorizing disease severity into normal, mild, moderate, and severe represents an ordinal scale (meaningful rank order without mathematically equal intervals). Nominal scales lack order (e.g., blood groups); interval and ratio scales feature equal quantitative distances with ratio possessing a true absolute zero. ---

📈 Graphical Data Representation

  • Geographical Disease Mapping: ((STAT008)): A dot-density map (spot map) is the gold standard for displaying spatial distribution, clusters, and hotspots during disease outbreaks (e.g., cholera, dengue).
  • Hierarchical/Branching Clinical Data: ((STAT009)): A tree diagram is utilized to explain multi-level progressions, such as systematic joint involvement pathways in a diabetic patient.
  • Arranged Quantitative Data: ((((STAT010)) · ((STAT011)))): Arranging biological values (e.g., heights or birth weights) in ascending/descending order directly yields the median / 50th centile (the middle value splitting the dataset into two equal halves).
  • Visual Statistical Displays: ((((STAT024)) · ((STAT025)) · ((STAT028)) · ((STAT029)))): - Boxplot: Displays minimum, Q1, median, Q3, maximum, and outliers. - Component bar graph: Divides a single bar into proportional stacked segments to show total composition and changes over time (e.g., yearly population structure). - Frequency polygon: Connects the midpoints of class intervals with straight lines to show continuous distribution shapes. - Line diagram: Ideal for tracking continuous quantitative trends across time. ---

🧮 Descriptive Statistics & Measures of Central Tendency

  • Correlation Coefficient Limits: ((STAT004)): Pearson's $r$ is strictly bounded between -1 and +1; any reported value exceeding these limits (e.g., $r = 1.1$) indicates a computation mistake.
  • Skewed Distributions: ((((STAT012)) · ((STAT016)))):
  • Positively Skewed (Right-Skewed): : Characterized by a long right tail of extreme high values. In this distribution: Mean > Median > Mode.
  • Negatively Skewed (Left-Skewed): : Characterized by a long left tail. In this distribution: Mean < Median < Mode.
  • Incubation Period Variability: ((STAT013)): For highly variable biological durations with extreme outliers (e.g., leprosy incubation ranging from months to decades), the mode is the preferred measure of central tendency.
  • Correcting Mean Calculations: ((STAT014)): To find the true mean after an entry error, recalculate the total sum by subtracting the incorrect value and adding the true value, then divide by the total sample size ($N$).
  • Coefficient of Variation (CV): ((STAT022)): Calculated as $\text{CV} = (\text{SD} / \text{Mean}) \times 100$ to measure relative dispersion.
  • Distribution Curves & Bimodality: ((STAT030)): Symmetrical unimodal curves have equal mean, median, and mode; bimodal curves possess two distinct peaks (indicating underlying subpopulations) where the mean, median, and mode are not equal. ---

📉 Probability, Confidence Intervals & Hypothesis Testing

  • Confidence Interval for the Mean: ((STAT015)): Calculated for large, normally distributed samples using the formula: $\text{Mean} \pm 1.96 \times (\text{SD} / \sqrt{n})$.
  • Empirical Rule (Normal Distribution): ((((STAT018)) · ((STAT019)))):
  • 95% of observations: fall within $\text{Mean} \pm 2\text{ SD}$.
  • 5th percentile cutoff: in a normal distribution corresponds to a z-score of -1.645, calculated via $X = \mu + (Z \times \sigma)$.
  • Statistical Significance & p-values: ((((STAT020)) · ((STAT021)))): A $p\text{-value} < 0.05$ indicates that the observed finding is unlikely due to chance, leading to the rejection of the null hypothesis ($H_0$) and acceptance of the alternative/study hypothesis ($H_1$).
  • Study Power & Errors: ((STAT017)): To minimize the risk of missing a true clinical difference, investigators must maximize statistical power, defined as $1 - \beta$ (where $\beta$ is the probability of a Type II error). ---

🧬 Advanced Survival Analysis & Special Plots

  • Comparing Survival Curves: ((STAT023)): Time-to-event outcomes with censored data displayed via Kaplan-Meier curves are statistically compared using the log-rank test / Cox proportional hazards test. (ANOVA, chi-square, and Bland-Altman analyses are inappropriate for time-to-event survival data).
  • Kaplan-Meier Confidence Bands: ((STAT031)): Hall–Wellner confidence bands present smooth, continuous shaded regions that narrow near the center and widen dramatically at the early and late time extremes to maintain simultaneous confidence coverage.