1

• Variation in Clinical Medicine can be caused by biological differences, presence or absence of disease, stages or extent of disease, different conditions and techniques of measurement, measurement error, and random variation
• Types of errors in data collection:

  • Systematic error: caused by variations that can distort data in one direction and introduce bias (e.g., weighing patients while wearing shoes)
  • Random error: caused by random variation and slight inaccuracies in measurement, not introducing bias (e.g., blood pressure)
• Quantitative characteristics are characterized using a defined, continuous measurement scale, like systolic & diastolic blood pressure and serum sodium level
• Qualitative characteristics are described by features or words rather than numbers, like normal skin color varying from pinkish white through tan to dark brown or black
• Nominal variables are naming or categoric variables not based on measurement scales or rank orders, like blood groups, occupation, food groups, and skin color
• Dichotomous (Binary) variables have only two levels, like normal/abnormal, male/female, well/sick, living/dead
• Ordinal (Ranked) variables are characterized by three or more qualitative values with a clearly implied direction from better to worse, like satisfaction with care (very satisfied, fairly satisfied, not satisfied)
• Continuous (Dimensional) variables are measured on continuous measurement scales, enabling detailed inferences compared to ordinal or nominal data, like height, weight, and serum glucose levels
• Ratio variables are derived from a continuous scale with a true 0 point, like Kelvin temperature
• Risk and proportions share characteristics of discrete and continuous variables, created by the ratio of counts in the numerator to counts in the denominator
• Frequency distributions can be shown by creating a table listing variable values according to their frequency of occurrence, or by plotting histograms to illustrate the frequency distribution
• The range of a variable is the distance between the lowest and highest observations, calculated as the highest value minus the lowest value
• Real frequency distributions are obtained from actual data or a sample, while theoretical frequency distributions are calculated using assumptions about the population from which the sample was obtained
• Histograms, frequency polygons, and line graphs are used to represent frequency distributions, with histograms illustrating frequency distribution through vertical bars
• Parameters of a frequency distribution involve examining central tendency, mode, median, and mean, as well as determining the spread or dispersion of the data
• Steps in examining a distribution:

  • Look for the central tendency of the observations
  • Examine the mode, median, and mean in detail
  • Determine how spread out (dispersed) the numbers are
• Measures of Central Tendency:

  • Mode: most commonly observed value, can be more than one
  • Median: middle observation when data are arranged from lowest to highest, not sensitive to extreme values
  • Mean: average value, calculated by summing all observed values and dividing by the total number of observations
• Measures of Dispersion:

  • Percentile: point at which a certain percentage of observations lie below the indicated point when all observations rank in descending order
  • Mean absolute deviation: average of the absolute value of deviations from the mean
  • Variance: sum of squared deviations from the mean, divided by the number of observations minus 1
  • Standard deviation: square root of the variance, used to describe the spread in the frequency distribution
• In a normal (Gaussian) distribution, the bell-shaped curve can be fully described using only the mean (central tendency) and standard deviation (dispersion)
• Problems in Analyzing a Frequency Distribution:

  • Skewness (s3): horizontal stretching of a frequency distribution to one side, skew = 0 symmetrical, skew < 0 negatively skewed, skew > 0 positively skewed
  • Kurtosis: vertical stretching or flattening of the frequency distribution, kurtosis = 3 mesokurtic, kurtosis > 3 leptokurtic, kurtosis < 3 platykurtic
• Types of Variables:

  • Qualitative variables: categories expressed as labels, e.g., sex, educational level
  • Quantitative variables: values indicate quantity or amount, e.g., age, weight, can be discrete or continuous
• Relationship between Variables:

  • Dependent variable: affected by independent variables
  • Independent variable: determines the value of the dependent variable
  • Confounding variable: affects the dependent variable but is not primarily of interest to the researcher
• Levels of Measurement:

  • Nominal: names or numbers representing mutually exclusive and exhaustive classes, e.g., sex, geographic regions
  • Ordinal: classes can be ordered, e.g., educational attainment
  • Interval: distances between all adjacent classes are equal, e.g., temperature measurement
  • Ratio: meaningful zero exists, e.g., weight, height
• Defining Variables:

  • Conceptual definition: taken from the dictionary
  • Operational definition: defines how the variable was used in the study, may vary from one study to another