OBJ - Descriptive Stats Flashcards
Qualitative Scales of Measurement
Categories: Nominal or Ordinal
NOMINAL – no ordering/ranking – just categories
• Names or labels. Numbers, if present, act as labels
• Blood group: A, B, AB, O
ORDINAL - Natural ranking present (differences/spacing between categories, unclear)
• Tumor stage (1-4)
Quantitative Scales of Measurement
Numbers: Continuous or Discrete
COUNT/DISCRETE = Numeric but can only take on specific values
• Number of seizures in a given month
CONTINUOUS = Measured or numerical data that can (in theory) take on any value
• Body mass index (kg/m2); BP, HR, etc
Summarizing Qualitative Medical Dat
Categories
1) Count = Number of cases
2) Proportion = Count in category / total count
3) Percentage = 100 x proportion
4) Ratio = Count in category A / count in category B
5) Rate = Proportion per period time
6) Odds = Proportion / (1- proportion)
Representations/Displaying Quantitative Data
Continuous Data
1) Histogram
2) Box plot (concisely displays Max, Min, 75th%, 25th%, median & **outliers)
Ordinal Data
1) Pie Graph
2) Bar Chart
Summarizing Quantitative Medical Data
Numbers
1) Mean – Arithmetic average (sum / sample size)
2) Median – Middle value of sorted data
3) Mode – Most frequent value
4) Geometric mean – Nth root of the product of the data (Money or titers and other data expressed as powers)
Spread of Variability
How far apart are the data are from each other
1) Range – Maximum – minimum
2) Interquartile range – 75th percentile – 25th percentile (central 50% of data)
3) Variance – Average squared deviation from the mean
Σ ( X – mean(X))2 / (n - 1)
4) Standard deviation – Square root of variance √ [Σ ( X – mean(X))2 / (n - 1)] 1 STDDEV = 68% 2 STDDEV = 95% 3 STDDEV = 98% Good if - Sample size is large - Distribution has (single peak) - Distribution is symmetric
5) Coefficient of variation = Standard deviation / mean (x 100%)
Report - BOTH location and spread - Report corresponding measures of location and spread Mean + standard deviation Median + range or interquartile range
Gaussian Distributions
- Bell-shaped
- σ = standard deviation
- μ = mean ( = median = mode)
- Total area under curve equal 1
- Extends to infinity in both directions
- Probability corresponds to area under the curve
- Completely described if two parameters (μ, σ) known
- Also called “Z” distribution (z = 1.5 ~ z = 1.5 std deviations from mean)
- Probabilities (areas under the curve) are tabled in most texts (Campbell table T1) – probability of BOTH tails
- Reference range if not clarified is 95%
Sources of Variation
Biologic variation
- Between-individual: two individuals have different values
- Within-individual: same individual measured twice has different values
Measurement variation
- Between-observer: two observers obtain different values when they measure the same thing
- Within-observer: same observer obtains different values when measuring the same thing twice
Instrument or analytical error
- Different equipment yields different values when measuring the same thing
Sources of Variability
- Reliability (Precise – to each other)
- Random error
- Measurements close to each other
- Validity (Accuracy – to “target”)
- Systematic error
- Measurements close to the true value