Sampling Epidemiology

Question 1

Epidemiological approaches require?

Answer 1

that we gather information about populations
- understand info at pop level, particualrly health status
- we can not sample everyone in apopulation - that is impossble, so we must sample.

Question 2

When you want to conduct a study, what exactly do you need to start?

Answer 2

You need to have a question. Some examples of the types of questions are:
1. Establish level of occurrence of a disease e.g. prevalence
2. Test hypotheses
* has the level of disease changed (Changes in level of occurrence)
* Identification of risk factors associated with disease. Sample in a way we can find differences in disease among various groups.
3. Detection of disease
* Particularly important to trade and Declaring freedom from disease.
- a lot of rules and standards for a country to declare that they are free of a disease.

Question 3

How are you going to measure disease within a population?

Answer 3

• What’s going on in the population?
  
  • What’s the best way to get this information?
    
    • Sample every animal? (aka a census) If you have everyone in the population you do not need statistics. Very powerful, expensive, and difficult to do.
    • Sample a portion of the animals?

Question 4

Define a census

Answer 4

• Census
• Every individual in the population
  is sampled
• Exact measurement of what is
  going on

Question 5

Define a sample.

Answer 5

• Sample
• Take a sample from the population
• Make inferences about the population based on information in the sample
• Quicker, less expensive, and practical!

Question 6

How would you define target population.

Answer 6

• Population you want to make inferences about with respect to your objectives. E.g. Feral cats in US and Canada

Population you are trying to understand.

Question 7

Source population

Answer 7

• Subset of the target population from which you will draw your sample. E.g. Feral cats on Long Island

Taking information from that to develop a study population.

Question 8

Study population

Answer 8

• Individuals actually measured or enrolled in your study. E.g. Feral cats spayed or neutered by LIU’s Team in 2021

Samples you can actually get your hands on.

Question 9

Define internal validity.

Answer 9

• Whether or not the study results (obtained from the study sample) are valid for the source population.

Question 10

Define external validity.

Answer 10

• How well the study results can be generalized to the target population
• Subjective assessment of whether or not the source population and sample are broadly representative of the target population

Tends to refer to the reader and they are wondering is this study relevant to me. Do I care about the results of this study? Does it impact me?

Question 11

Define sampling frame.

Answer 11

• Complete list of all sampling units.
• Required for most probabilistic sampling methods.

Question 12

Define sampling unit.

Answer 12

• Unit of interest…
• Animals (usually the unit in vet med)
• Litters
• Pens
• Farms
• etc. …

Question 13

What are the two approaches to sampling?

Answer 13

• Two approaches to sampling
  1. Probability sampling
  2. Non-probability sampling

Question 14

Define probability sampling.

Answer 14

1. Probability sampling
   * Every element in the population has a know probability of being included in the sample. Where stats and assumptions hold true.
   * Formal process of random selection used to ensure representativeness of the sample.
   * Statistical inference can be made about the population from which you drew your sample.

Question 15

Name the types of probability sampling.

Answer 15

• Five types of probability sample:
  1. Simple random = the easiest one. (also called SRS in stats).
  2. Systematic random
  3. Stratified
  4. Cluster
  5. Multistage

Every element in the population has a known non-zero probability of being included in the sample.

Question 16

Define simple random probability sampling.

Answer 16

• AKA Simple Random Sample (SRS)
• Forms the basis of most probability sampling methods
• Every subject has equal probability of being included in the sample
• Requires a list of all subjects (sampling frame)
• Not always easy or possible
• Procedure
• Generate list of all animals
• Assign a unique number (ID) to each animal
• Generate a list of random numbers assigned to IDs

Question 17

What are the limitations of simple random probability sampling? Provide an example.

Answer 17

• Limitations
• Difficult to obtain a complete sampling frame
• Costs may be higher than other methods

Question 18

Define systematic random probability sampling.

Answer 18

• Systematic random
• Does not require sampling frame, but need all animals presented sequentially
• E.g. fish processing, or appointment book (or old folder filing system)
• Do not necessarily need to know the number of sampling units (but need a rough guess)

Question 19

Describe the procedure used for systematic random probability sampling. Provide an example. (see slideshow)

Answer 19

• Procedure
• Select your sampling interval, j
• Randomly select a number within your sampling interval
• Sample every j th animal after that

Question 20

What are the limitations of systemic random probability sampling.

Answer 20

• Limitations
• Could be subject to bias in the interval
• E.g. (not a great example, but) if you have 5 pens and the producer takes one cow at a
  time from each pen in a sequence for them to run through a chute… you shouldn’t
  sample every 5th animal going through the chute!

Question 21

Define stratified random probability sampling.

Answer 21

• Stratified random
• Sampling frame is divided into groups (strata) of defined common characteristic
• Breed, sex, age (parity), etc.
• Randomize within each stratum
• Simple random or systematic random
• Proportional to the total number, within the stratum
• Ensures all strata are present and properly represented in the study

Question 22

What is an example of stratified random probability sampling. (see slideshow)

Answer 22

• E.g. You want to estimate the prevalence of a disease in a population of cats
• Suspect Siamese cats have a much higher prevalence of disease than DLH/DSH cats
• Want a fair representation of breeds in your study to prevent bias in estimate of prevalence

Question 23

What are the limitations of stratified random probability sampling.

Answer 23

• Need to know about stratification variable in advance
• Less efficient than SRS if the strata do not explain variance within the population (i.e. strata
  are not risk factors)

Question 24

Define cluster probability sampling.

Answer 24

• Sampling frame is divided into logical aggregations (clusters) and a random selection of
  clusters is performed
  
  • Clusters are common in veterinary medicine
    
    • Any group that shares common characteristics
      
      • Farm
      • Pen
      • Litter
    • Animals within a cluster are more similar than animals between clusters
• All individual sampling units within the selected clusters are included in the sample
• Random selection of clusters (groups)
• Unit of concern (Primary Sampling Unit) = all individuals within the cluster

Question 25

What are the advantages of cluster probability sampling?

Answer 25

* Only require a sampling frame for groups, but not for the members within the group * Fewer sampling units (clusters) are required to achieve a certain sample size of study units * More economical than SRS

Question 26

What are the limitations of cluster probability sampling?

Answer 26

* Higher variability than SRS, so larger overall sample size (units of concern) is required to achieve the same precision * When between-cluster variation is large relative to within-cluster variation --> need to sample many more clusters! If the shelters are all very similar, you care more about the individual differences of the animals. How similar are the animals to one another in a shelter? How similar are shelters to one another?

Question 27

Define multistage probability sampling.

Answer 27

* Same as Cluster Sampling, but take a sub-sample of subjects within clusters. * Why do this? * Animals within a cluster are more alike than animals between clusters * Don’t gain much information from sampling over and over within a cluster! * Gain more info by sampling more herds and taking a small sub-sample within each herd * Very common in veterinary epidemiology * Cluster is randomly selected * Individuals within each cluster are randomly selected (primary sampling units = PSUs) * Can have multiple stages: * Farm -> Sows -> Litters -> Piglets

Question 28

What are the advantages of multistage probability sampling?

Answer 28

* Advantages * Only requires a list of clusters * Farms, not animals * Sites, not pens, not fish * Most practical for large studies (national) * Adjust number of clusters and primary sampling units (PSUs) based on cost of acquiring samples vs running samples * Also depends on within- and between-cluster variation If within cluster variation is high --> smaller number of clusters, with more PSUs i.e. weak ‘farm’ effect --> need fewer farms, but more animals If between cluster variation is high --> larger number of clusters, with fewer PSUs i.e. strong ‘farm’ effect --> need more farms, but fewer animals

Question 29

What are the limitations of multistage probability sampling?

Answer 29

* Limitations * Larger sample sizes (number of clusters) may be required to achieve the same precision as SRS * Due to dependence (similarity) * Greater complexity of statistical methods to analyze and estimate sample size

Question 30

Define non-probability sampling.

Answer 30

1. Judgement (haphazard) * Investigator chooses subjects that ‘represent’ the population in the judgement of the investigator 2. Convenient * Samples that are easy to obtain * OK if you do not need a group that is representative of the source population 3. Purposive * Chose on attributes (know exposure, cases) * If random samples are taken from this group, then it is a probability sample from the source population 4. Targeted (risk-based) * Specifically used in animal disease surveillance programs * Biased sample... to help: * Find rare disease * Determine freedom from disease

Question 31

What is an example of targeted (risk-based) sampling?

Answer 31

Example: Bovine spongiform encephalopathy (BSE) surveillance: * Cattle over 30 months * Dead, Down, Diseased, Dying (4 Ds) ISA surveillance (fish in Canada): * Mortalities and moribund fish routinely submitted

Question 32

What are the advantages of non-probability sampling?

Answer 32

* Advantages * Convenient and inexpensive * Appropriate under certain circumstances * Exploratory or preliminary investigation * Homogenous populations * Sampling to determine if disease is present in a population

Question 33

What are the disadvantages of non-probability sampling?

Answer 33

* Not possible to know the probability of each unit being selected * Cannot estimate sample size or standard errors * Not appropriate for descriptive studies (at the population-level) * Can lead to biased population estimates, with no way to quantify the extent of the bias (selection bias)!

Question 34

Sample Size requires ‘Basic Statistics’ * Extra statistical resources * CDC: Principles of Epidemiology in Public Health Practice https://www.cdc.gov/csels/dsepd/ss1978/ss1978.pdf * Lesson 2: Summarizing Data (page 109 to 179) Breeze through this

Question 35

What are the different types of data?

Answer 35

1. Continuous: Real numbers such as temperature, weight, decimal values, etc. 2. Dichotomous: Yes/No... 1/0, male vs. female 3. Categorical (Nominal): Ordinal (Type 1, 2, 3) *anything that is more than two possibilities* 4. Count: Integers

Question 36

Histogram

Answer 36

freq on y axis over some type of x axis variable

Question 37

boxplots

Answer 37

see graph in powerpoint

Question 38

What are the different types of distributions?

Answer 38

* Distributions * Binomial vs. Normal * Shapes: skewed (Right vs Left) * Mean, median, mode * Percentiles, Quartiles, and Interquartile ranges

Question 39

What is the difference between a binomial and a normal distribution?

Answer 39

The main difference between the binomial distribution and the normal distribution is that binomial distribution is discrete, whereas the normal distribution is continuous. It means that the binomial distribution has a finite amount of events, whereas the normal distribution has an infinite number of events.

Question 40

What are the different measures of spread?

Answer 40

* Variance vs. Standard Deviation

Question 41

What is the difference between standard deviation and variance?

Answer 41

Standard deviation measures how far apart numbers are in a data set. Variance gives an actual value to how much the numbers in a data set vary from the mean. Standard deviation is the square root of the variance and is expressed in the same units as the data set. Variance can be expressed in squared units or as a percentage.

Question 42

Define standard error of the mean.

Answer 42

Confidence Intervals (e.g. 95%CI)

Question 43

If you increase sample size, what would happen to your precision? standard deviation? sample error?

Answer 43

Question he said is fair game for exam

Question 44

Normal distribution graph