Medical Statistics: Probability, Distributions, and Statistical Inference

0.0(0)
Studied by 0 people
call kaiCall Kai
learnLearn
examPractice Test
spaced repetitionSpaced Repetition
heart puzzleMatch
flashcardsFlashcards
GameKnowt Play
Card Sorting

1/71

flashcard set

Earn XP

Description and Tags

Vocabulary practice flashcards covering fundamental definitions, probability laws, binomial and normal distributions, and principles of statistical inference from Lectures 1-4 of Medical Statistics.

72 Terms

1
New cards

According to the course syllabus, what is the reference textbook for the Medical Statistics module?

Essential Medical Statistics by B.R. Kirkwood and J.A.C. Sterne (Blackwell publishing, second edition).

2
New cards

How is probability defined under the frequentist approach?

The proportion of times that an event would be observed if the experiment were repeated a large number of times.

3
New cards

Compared to a driver with a zero blood alcohol level, how much more likely to have an accident are drivers with blood alcohol levels of 0.050.05, 0.10.1, and 0.150.15?

Those with a level of 0.050.05 are 33 times more likely, with a level of 0.10.1 are 88 times more likely, and with a level of 0.150.15 are 2929 times more likely.

4
New cards

In a fair die throw experiment, approximately how many throws are required before the proportion of each face landing on top stabilizes near one in six (16.7%16.7\%)?

About 20002000 throws.

5
New cards
<p>What concept does this graph demonstrate regarding the proportion of throwing a 3 over repeated experiments?</p>

What concept does this graph demonstrate regarding the proportion of throwing a 3 over repeated experiments?

Random variation: in the short run the proportion varies widely above and below 16.7%16.7\% (0.1670.167), but in the long run it stabilizes closer to one in six.

6
New cards

What are mutually exclusive events, and what does the additive law of probability state for them?

Mutually exclusive events cannot occur at the same time (they have no intersection). The additive law states that the probability of at least one occurring is the sum of their individual probabilities: Prob(A or B)=Prob(A)+Prob(B)\text{Prob}(A \text{ or } B) = \text{Prob}(A) + \text{Prob}(B).

7
New cards

What are the numerical probabilities assigned to an event that is certain to happen and an event that cannot happen?

An event certain to happen has a probability of 100%100\% (or 11), and an impossible event has a probability of 0%0\% (or 00). Probabilities can never be greater than 11 or less than 00.

8
New cards

When are two events considered independent, and what is the multiplicative law of probability for independent events?

Two events are independent when the outcome of the first tells us nothing about the outcome of the other, and vice versa. The multiplicative law states that prob(A and B)=prob(A)×prob(B)\text{prob}(A \text{ and } B) = \text{prob}(A) \times \text{prob}(B).

9
New cards

In a sample of 200 blood donors, 46% are group O and 3% are group AB. Using the additive law, what is the probability that the next donor is either group O or group AB?

46%+3%=49%46\% + 3\% = 49\% (or 0.490.49), because being group O and being group AB are mutually exclusive.

10
New cards

At a Rome hospital in 2001, singleton births comprised 52% boys and 48% girls. What is the probability that the first three singleton babies born are all girls?

prob(GGG)=0.48×0.48×0.48=0.11\text{prob}(GGG) = 0.48 \times 0.48 \times 0.48 = 0.11 (or 11%11\%), since the sex of each birth is independent.

11
New cards

What is the general formula for the additive rule when two events AA and BB are not mutually exclusive?

P(A∪B)=P(A)+P(B)−P(A∩B)P(A \cup B) = P(A) + P(B) - P(A \cap B)

12
New cards

What is the formula for conditional probability P(B∣A)P(B|A)?

P(B∣A)=P(A∩B)P(A)P(B|A) = \frac{P(A \cap B)}{P(A)}, meaning the probability of event BB occurring given that event AA has occurred.

13
New cards

What is the mathematical formulation of Bayes' Theorem as presented in the lecture?

P(B∣A)=P(B)×P(A∣B)P(A)P(B|A) = \frac{P(B) \times P(A|B)}{P(A)}

14
New cards

What three sequential steps are required to calculate complex probabilities in epidemiology?

  1. Count: Count all possible combinations of events leading to the outcome of interest.
  2. Calculate: Calculate the probability of each combination.
  3. Add: Add the separate probabilities across all combinations.
15
New cards

What is the combinatorial formula used to calculate the number of combinations of xx events in nn trials?

n!x!(n−x)!\frac{n!}{x!(n-x)!}, where n!=1×2×3×⋯×nn! = 1 \times 2 \times 3 \times \dots \times n.

16
New cards

What is the short-cut method to calculate the probability of having at least one boy among three newborn babies?

Subtract the probability of the all-girl combination (GGGGGG) from 11: prob(at least one boy)=1−prob(GGG)=1−(0.48)3=1−0.11=0.89\text{prob}(\text{at least one boy}) = 1 - \text{prob}(GGG) = 1 - (0.48)^3 = 1 - 0.11 = 0.89 (or 89%89\%).

17
New cards

What type of data does the binomial distribution describe?

Discrete binary data—characteristics that can take one of only two possible values (e.g., boy/girl, vaccinated/not vaccinated, HPV positive/negative).

18
New cards

How does the location of the most probable value in a binomial distribution with a small sample size (n=10n = 10) affect its shape?

When the most probable value is near the middle of the range (like 33 or 55), the distribution is almost symmetrical. When the most probable value is near 00, it is skewed to the right; when near 1010, it is skewed to the left.

19
New cards

What happens to the shape of a binomial distribution as the sample size increases from 1010 to 100100?

The distribution becomes more symmetrical and its shape approaches that of the Normal distribution.

20
New cards

What is the distinction between categorical variables and quantitative variables?

Categorical variables take a limited number of values (such as marital status or country of birth), whereas quantitative variables take many numerical values (such as weight or cholesterol).

21
New cards

Who first applied the Normal distribution in astronomy, giving it the alternative name Gaussian distribution?

Carl Friedrich Gauss (17771777-18551855).

22
New cards

What two parameters completely define a Normal distribution, and what roles do they play?

The mean (μ\mu), which defines the location (shifting the curve left or right), and the standard deviation (SD\text{SD}), which defines the shape/spread (larger SD\text{SD} increases spread and reduces peak height).

23
New cards
<p>What effect does changing the standard deviation have on the appearance of the Normal curve, as illustrated here?</p>

What effect does changing the standard deviation have on the appearance of the Normal curve, as illustrated here?

A smaller standard deviation (SD=25\text{SD} = 25) produces a narrower spread and a taller curve, whereas a larger standard deviation (SD=90\text{SD} = 90) produces a wider spread and a flatter curve.

24
New cards

What is the total area under any Normal distribution curve?

11 (or 100%100\%), representing all possible values in the population.

25
New cards

What are the mean and standard deviation of the Standard Normal distribution?

A mean equal to 00 and a standard deviation equal to 11 (zz-scores).

26
New cards

What is the formula to convert any normally distributed observation xx into a standardized normal score (zz-score)?

z=x−meanSDz = \frac{x - \text{mean}}{\text{SD}}

27
New cards

What percentage of a Normal distribution lies within ±1 SD\pm 1\,\text{SD}, ±1.96 SD\pm 1.96\,\text{SD}, and ±2 SD\pm 2\,\text{SD} of the mean?

±1 SD\pm 1\,\text{SD} contains 68.3%68.3\%, ±1.96 SD\pm 1.96\,\text{SD} contains 95.0%95.0\%, and ±2 SD\pm 2\,\text{SD} contains 95.4%95.4\%.

28
New cards

What proportion of the area under the Standard Normal curve lies outside the range of ±1 SD\pm 1\,\text{SD} (two-tailed area)?

100%−68.3%=31.7%100\% - 68.3\% = 31.7\% (or 0.31730.3173).

29
New cards

How do you calculate the upper-tail probability (P-upper) for a negative zz-score, such as z=−2.00z = -2.00, using the Standard Normal table?

Subtract the upper-tail probability of +z+z from 11: 1−P-upper(+2.00)=1−0.0228=0.97721 - P\text{-upper}(+2.00) = 1 - 0.0228 = 0.9772. By symmetry, P-lower(−z)=P-upper(+z)P\text{-lower}(-z) = P\text{-upper}(+z).

30
New cards

What mathematical transformations are commonly applied to normalize data that are skewed to the right?

1x\frac{1}{x}, log⁡(x)\log(x), and x\sqrt{x}.

31
New cards

What mathematical transformations are commonly applied to normalize data that are skewed to the left?

Squaring (x2x^2), cubing (x3x^3), and exp⁡(x)\exp(x).

32
New cards

In medical statistics, what base is implied when the term 'log' is used?

Natural logarithm (base-ee, corresponding to the 'ln' button on a calculator).

33
New cards

What is the definition of statistical inference?

The process by which we draw conclusions about a population from data collected on a sample.

34
New cards

What is the difference between a target population and a sampled population?

The target population is the population about which we want information. The sampled population is the population about which we can and do obtain information. If they differ, results will be biased.

35
New cards

Why did the Department of Education's smoking survey find a 4% prevalence while the Department of Health found 21% in the same town?

The Department of Education's sampled population only included pupils aged 10–15 (friends of the director's children), missing older pupils aged 16–19 who are more likely to smoke, creating selection bias.

36
New cards

What symbols are used to distinguish population parameters from sample statistics?

Greek letters (such as μ\mu for mean and π\pi for proportion) represent unobserved true population parameters; Roman letters (such as xˉ\bar{x} or pp) represent measured sample statistics.

37
New cards

What is the definition of a sampling distribution?

The probability distribution of how a sample estimate (such as a mean or proportion) varies across many repeated random samples of the same size drawn from a population.

38
New cards

What is Property 1 of sampling distributions regarding their mean?

The mean of a large number of estimates obtained from samples of identical size is approximately equal to the true population value (π\pi or μ\mu).

39
New cards

What is the conceptual difference between a standard deviation (SD) and a standard error (SE)?

The standard deviation represents variability in individual data, whereas the standard error represents variability in sample estimates across repeated samples.

40
New cards

What is the formula for the standard error of a sample proportion, SE(p)\text{SE}(p)?

SE(p)=π(1−π)n\text{SE}(p) = \frac{\sqrt{\pi(1-\pi)}}{\sqrt{n}}

41
New cards

How does increasing the sample size (nn) by a factor of 100 affect the standard error of a sampling distribution?

Because SE\text{SE} depends inversely on n\sqrt{n}, increasing nn by 100100 times makes the SE\text{SE} 1010 times smaller (e.g., decreasing from 0.150.15 to 0.0150.015).

42
New cards

What does the Central Limit Theorem state regarding the shape of a sampling distribution?

When the sample size is large, the sampling distribution of sample estimates is approximately Normal, even if the underlying distribution of the original population data is not normal.

43
New cards

How is a single sample estimate pp used to infer the 95% confidence interval for the true population proportion π\pi?

Because the sampling distribution is approximately Normal and π\pi is unknown, pp is used as the best estimate to compute SE(p)\text{SE}(p), allowing the researcher to be 95%95\% confident that π\pi lies within approximately 2 SE2\,\text{SE} of pp.

44
New cards

Frequentist Probability

The proportion of times an event will be observed to occur if an experiment is repeated a large number of times under identical conditions.

45
New cards

Random Variation

The short-run divergence of observed proportions above and below their true expected probability that disappears only after a large number of repetitions.

46
New cards

Mutually Exclusive Events

Events that cannot occur at the same time and have no intersection, meaning the occurrence of one prevents the occurrence of the other.

47
New cards

Additive Law of Probability

A probability rule stating that for mutually exclusive events, the probability of at least one event occurring is the sum of their individual probabilities: P(A or B)=P(A)+P(B)P(A \text{ or } B) = P(A) + P(B).

48
New cards

Independent Events

Events for which the value or occurrence of one event provides no information about the value or occurrence of the other event, and vice versa.

49
New cards

Multiplicative Law of Probability

A probability rule stating that for independent events, the probability of both occurring together is the product of their separate individual probabilities: P(A and B)=P(A)×P(B)P(A \text{ and } B) = P(A) \times P(B).

50
New cards

Exhaustive Events

A set of events that encompasses all possible outcomes of an experiment, such that their combined probability equals 11 (100%100\%).

51
New cards

Conditional Probability

The probability of an event BB occurring given the condition that another event AA has already occurred, denoted as P(B∣A)P(B|A).

52
New cards

Bayes' Theorem

A theorem allowing the calculation of inverse conditional probabilities using the formula: P(B∣A)=P(B)×P(A∣B)P(A)P(B|A) = \frac{P(B) \times P(A|B)}{P(A)}.

53
New cards

Factorial

The product of all positive integers less than or equal to nn, denoted as n!=1×2×3×⋯×nn! = 1 \times 2 \times 3 \times \dots \times n.

54
New cards

Combinations Formula

A combinatorial formula used to calculate the number of ways to obtain xx successes in nn trials without regard to order: n!x!(n−x)!\frac{n!}{x!(n-x)!}.

55
New cards

Binomial Probability Distribution

A discrete probability distribution that describes the probabilities of obtaining each possible count of successes for binary characteristics across a fixed number of independent trials.

56
New cards

Categorical Variable

A variable that can take only a limited number of distinct non-continuous categories or values, such as blood group, marital status, or country of birth.

57
New cards

Quantitative Variable

A variable that can take many different continuous numerical values, such as height, weight, or blood cholesterol level.

58
New cards

Normal Distribution

A symmetrical, bell-shaped continuous probability distribution completely defined by its mean and standard deviation, also known as the Gaussian distribution.

59
New cards

Location Parameter

The mean (μ\mu) of a Normal distribution, which dictates the central position of the distribution curve along the horizontal axis.

60
New cards

Spread Parameter

The standard deviation (SD\text{SD}) of a Normal distribution, which governs the width and height of the curve around the mean.

61
New cards

Standard Normal Distribution

A specific Normal distribution that has a mean equal to 00 and a standard deviation equal to 11.

62
New cards

Z-Score

A standardized value indicating how many standard deviations an individual observation lies above or below the distribution mean: z=x−meanSDz = \frac{x - \text{mean}}{\text{SD}}.

63
New cards

Two-Tailed Area

The proportion of the area under the Normal distribution curve that lies in both extreme tails outside a specified symmetric interval (−z,+z)(-z, +z).

64
New cards

Logarithmic Transformation

The process of taking the natural logarithm (ln⁡\ln) of right-skewed data values to compress extreme values and make the distribution more symmetrical and approximately normal.

65
New cards

Statistical Inference

The process of drawing generalized conclusions about an unobserved target population based on data measured from a representative sample.

66
New cards

Target Population

The entire group or collection of individuals about which a researcher ultimately seeks to gain information.

67
New cards

Sampled Population

The actual population from which sample individuals are eligible to be selected and from which data are collected.

68
New cards

Sampling Distribution

The theoretical probability distribution of a sample statistic (such as a sample mean or sample proportion) obtained from repeatedly taking random samples of the same size from a population.

69
New cards

Standard Error (SE)

The standard deviation of a sampling distribution, which quantifies the variability of sample estimates across different samples of identical size.

70
New cards

Standard Error of a Proportion

The standard error quantifying sampling variability of a sample proportion pp, defined as SE(p)=π(1−π)n\text{SE}(p) = \frac{\sqrt{\pi(1-\pi)}}{\sqrt{n}}, which is inversely related to n\sqrt{n}.

71
New cards

Central Limit Theorem

The statistical theorem stating that when sample size is large, the sampling distribution of a sample estimate is approximately normal, even if the underlying data variable is not normally distributed.

72
New cards

Confidence Interval

A range defined by two limits around a sample estimate that provides a designated degree of confidence (e.g., 95%95\%) of containing the true population parameter.