Sampling strategy Flashcards
1
Q
why is sampling strategy important
A
- for efficiency
- need to know how to select the appropriate sampling strategy
2
Q
sampling scheme
A
- sampling strategy
- population determination
- sampling plan
- sampling procedure
3
Q
an appropriate sampling strategy is dependent on what
A
- The purpose of the investigation
- The customer’s request
- The anticipated use of the results
All this should be taken into account when designing a sampling scheme
4
Q
what does a sampling plan provide
A
- Adequate basis for answering questions of applicable law
e. g. is there a drug present in the population - if an inference about the whole population is to be drawn from a sample then the plan shall be statistically based, and limits of the inference shall be documented
5
Q
statistical sampling strategies
A
frequentist
bayesian
6
Q
non- statistical sampling strategies
A
square root N
management directive
judicial requirements
7
Q
what forms the foundation of most sampling strategies
A
laws and legal practice
8
Q
what two approaches may a sampling strategy take
A
statistical or non statistical approach
In many cases, a non statistical approach may suffice
9
Q
what does a sampling strategy provide
A
- Adequate basis for answering questions of applicable law
e. g. is there a drug present in the population - if an inference about the whole population is to be drawn from a sample then the plan shall be statistically based, and limits of the inference shall be documented
10
Q
what needs to be addressed when a single unit or bulk population needs to be analysed
A
the issue of homogeneity
- one sample is sufficient if the bulk material is homogeneous
11
Q
what if the bulk material is not homogenous
A
several samples from different locations may be necessary to ensure that the test results are representative (to avoid false negative results)
- Depending upon the inference to be drawn/ what needs to be known form the analysis for a multiple unit population (may be statistical or non statistical)
12
Q
statistical approaches
A
- Applicable when inferences are made about the whole population for example:
The probability that a given % of the population contains the drug of interest or is positive for a given characteristic
13
Q
non statistical approaches
A
appropriate if no inference is to be made about the whole population
You just want to know if a drug is present or not
14
Q
sampling plans advantages
A
- decrease time per case
- decrease use of costly chemicals and instrumentation
- widely used in the forensic community
- usually sufficient to prove possession/supply go a controlled substance
15
Q
sampling plans disadvantages
A
- means that some items are not tested
- can be confusing to explain
- in the legal community, there is a lack of understanding/ communication - may be challenged in court
16
Q
basic sampling methods
A
- Random
- Systematic
- Stratified
- Cluster
- Convenience
17
Q
what is random sampling
A
- in Simple Random Sample of a given size
All such subsets are given an equal proability
Or the pick a name out of the hat technique
18
Q
how to make sure you sample is random in random sampling
A
Random number table
Random number generator
19
Q
advantages of random sampling
A
- Minimises bias and simplifies analysis of results
- The variance between individual results within the sample is a good indicator of variance in the overall population
Making it relatively easy to estimate the accuracy of results
20
Q
disadvantages of random sampling
A
- SRS can be vulnerable to sampling error because of the randomness of the selection
May result in a sample that doesn’t reflect the makeup of the population
21
Q
population sampling guidelines
A
no of units -> no of sample 1-> 1 2-5 -> 2 6-15 -> 3 15-25 -> 4 >25 --> 5
22
Q
mixing and sampling
A
- In coning and quartering, the sample is manually mixed in on itself for a period of time and the material is formed into a cone
- A particle could roll down the cone in any direction, hence the mixing effect
- The top of the cone is flattened and it is divided into 4 quarters
- Opposite quarters are combined to give half the sample, and the process is repeated until a suitable sample size is obtained.
23
Q
systematic sampling
A
- Relies on arranging the study population according to some ordering scheme
- Then selecting elements at regular intervals through the ordered list
- Involves a random start and the proceeds with the selection of every kth element form then onwards
- It is important that the starting point is randomly chosen
24
Q
stratified sampling
A
- Population is divided into subgroups (strata)
- Strata are based on specific characteristics
Age
Appearance
Education level
Etc. - Use random sampling within each strata
25
cluster sampling
- more cost-effective to select respondents in groups
- Sampling is often clustered by geography or by time periods
- Random sampling used to choose clusters
- All data used from selected clusters
26
convenience sampling
- Sometimes known as grab, accidental or opportunity sampling
- A type of non-probability sampling which involves the sample being drawn from the part of the population which is close to hand
- Be wary of bias
27
when sampling drugs
- Each group will be considered as a whole population and will be sampled alone
- In some rare cases, although the external characteristics look the same, upon opening the units (sampling), difference in the powder appearance among the units may be seen
28
getting a random sample
- The theoretical way to select a truly random, unbiased representative sample from a population is to individually number each item in the population
- Then use a number generation to choose which item to select
- This is not possible in practice, especially for large populations containing many thousands of units
29
when sampling we must ensure that two principles are maintained
- The properties of the sample are a true reflection of the properties of the population from which the samples were taken.
- Each unit in the population has an equal chance of being selected.
30
what is arbitrary sampling
- They are often used in practice and work well in many situations
- However they have no statistical foundation and may lead to a very large sample to analyse in case of large seizures
e.g all (n=N)
- Advantages: 100% certainty about the composition of the population
- Disadvantage: excessive sample sizes for larger populations
n = 0.05N, n = 0.1N etc
- Advantage: simple approach
- Disadvantage: excessive sample sizes for larger populations
n = √N, n = 0.5 √N, n = √N/2
- Advantage: widely accepted approach
- Disadvantage: the number of samples may be too small when the population is small
- Disadvantage: excessive sample sizes for larger populations
n = 1
- Advantage: minimum amount of work
- Disadvantage: least amount of information on the characteristics of the seizure
- Two methods concern a frequentist approach, while the third method describes a Bayesian approach
31
statistical sampling method
- Two methods concern a frequentist approach, while the third method describes a Bayesian approach
32
frequent sampling methods
- The assumption a fixed but unknown proportion of the seizure (population) contains drugs
- The proportion of drugs in a sample (= the sampled units) can estimate this seizure population
- The proportion of drugs in the sample will however vary over different samples
- Therefore the frequentist methods provide a confidence, (1-α)100%
- For instance 95% if α is selected to be 0.05
- That with a given sample proportion of the seizure proportion is at least k 100% (for instance 90% if k is selected to be 0.9)
- In another words, one would be correct about a seizure containing at least 90% drugs in 95% out of 100 cases
33
bayesian sampling methods
- The assumption behind a Bayesian approach is that the sample proportion is known and fixed
- This proportion is used to calculate probabilities on certain value of the unknown seizure proportion that at that point is still assumed variable
34
they hypergeometric distribution
* The probability that a sample size n contains X positives (units containing illegal drugs)
* Given that the population of size N contains N1 positives, can be calculated by
(equation in powerpoint)
and see notes for examples
35
the binomial distribution
- The binomial distribution can be used to calculate a sample size n such that with (1-α)100% confidence can be stated at least a proportion of k 100% is positive
- The calculations with the binomial distribution are easier than the ones with the hypergeometric distribution
- However binomial distributions are approximations
- Sample size estimated with it will be slightly overestimated
- Only in very large seizures (sometimes of several thousand) will the sample sizes calculated from both distributions be equal
see notes for examples
