main index

P00: frame around

P01: olicognography

P02: addictions


Registers of application docs

*statistics sampling errors *

*statistics processes *

*procedure bias*


Similar user docs

*entropy measuring *

*sampling mistakes *



Kinds of Sampling

Sampling Kinds


Brief Description



Simple random

Assign to each population member a unique number; select sample items by use of random numbers

1. Requires a minimum knowledge of population in advance,

2. Free of possible classification errors

3. Easy to analyze data and compute errors

1. Does not make use of knowledge of population which researcher may have

2. Larger errors for same sample size than in stratified sampling


Use natural ordering or order population, select random starting point between 1 and the nearest integer to the sampling ratio (N/n); select items at interval of nearest integer to sampling ratio

1. If the population is ordered with respect to pertinent property, gives stratification effect , and hence reduces variability compared to simple random sampling

2. Simplicity of drawing sample; easy to check

1. If sampling interval is related to a periodic ordering of the population, increased variability may be introduced,

2. Estimates of error likely to e high where there is stratification effect

Multistage random

* prob. proportio-nate to size

Use a form of random sampling in each of the sampling stages where there are at least 2 stages

select sampling units with probability proportionate to their size

1. Sampling lists, identification, and numbering required only for members of sampling units selected in sample

2. If sampling units are geographically cuts down field costs,

1. Reduce variability

1. Error likely to be larger than in previous 2 sampling for same sampling size

2. Errors increase as number of sampling units selected decreases

1. Lack of knowledge of size of each sampling unit before selection increases variability


1. Proportionate

Select from every sampling unit at other than last stage a random sample proportionate to size of sampling unit

1. Assures representativeness with respect to property which forms basis of classifying units; therefore yields less variability than first and third sampling. 2. Decreases chance of failing to include members of population because of classification process

3. Characteristics of each stratum can be estimated, hence comparisons can be made

1. Requires accurate information on proportion of population in each stratum, otherwise increases error

2. If stratified lists are not available, may be costly to prepare them; possibility of faulty classification hence increase in variability

2. Optimum allocation

Same as before except sample is proportionate to variability within strata as well as their size

Less variability for the same sample then the previous

Requires knowledge of variability of pertinent characteristic within the strata

3. Disproportionate

Same as previous except that size of sample is not proportionate to size of sampling unit but is dictated by analytical considerations or convenience

More efficient than previous for comparison of strata or where different errors are optimum for different strata

Less efficient than previous for determining population characteristics, i.e. more variability for same sample size


Select sampling units by some form of random sampling; ultimate units are groups: select these at random and take a complete count of each

1. If clusters are geographically defined yields lowest field costs

2. Requires listing only individuals in selected clusters

3. Characteristics of clusters as well as those of population can be estimated. 4. Can be used for subsequent samples, since clusters, not individuals, are selected, and substitution of individuals may be permissible

1. Larger errors for comparable size than other probability samples

2. Requires ability to assign each member of population uniquely to a cluster; inability to do so may result in duplication or omission of individuals

Stratified cluster

Select clusters at random from every sampling unit

1. Reduces variability of plain cluster sampling

1. Disadvantages of stratified sampling added to those of cluster sampling

2. Since cluster properties may change, advantage of stratification may be reduced and make sample unusable for later research

Repetitive multiple or sequential

Two or more samples of any of those above types are taken, using results from earlier samples to design later ones, or determine if they are necessary

1. Provides estimates of population characteristics which facilitate efficient planning of succeeding sample, therefore reduces

2. In the long run reduces number of observations required

1. Complicates administration of field work,

2. More computation and analysis required than in no repetitive sampling. 3. Sequential sampling can only be used where a very small sample can approximate representative ness and where the number of observations can be increased conveniently at any stage of the research.



Select a subgroup of the population which, on the basis of available information, can be judged to be representative of the total population; take a complete count of this sub sample of this group

Classify population by pertinent properties determined desired proportion of sample from each class; fix quotas for each observer

1. Reduces cost of preparing sample and field work, since ultimate units can be selected so that they are close together

1. Same as above

2. Introduce some stratification effect

1. Variability and bias of estimates cannot be measured or controlled

2. Requires strong assumptions or considerable knowledge of population and subgroup selected

1. Introduces bias of observerís classification of subjects and nonrandom selection within classes

Source: nfm

Places of use docs

*non parametric statistics *

*invariants *