Sampling Design

Sample decisions when canvasing locals about information of even geographical areas require a solid sample selection. Judgements must be made in an effort to minimize costs, time, and acceptance of a project.What much of this tells me is that most project managers wouldn't understand the statistical choices and results provided, therefore seeking out peer review for this work should be done to confirm the project makes sense. Alternative energy solutions is a complex and statistical based industry and must be properly understood for the benefits to be realized.

Depends on a number of factors...

Similarity among certain traits.

True values are never truely known

Determine how close to the actual population size should be measured.

Relationship recognition.

Resource constraints.

Percentage of the population that possesses the trait being measured

The measure of participation by respondants

Considerations can be countered by screening respondents.

The complete set of unit analysis under investigation. May be one of two scopes.

Also known as the study sample, it is the subset chosen from the population to study. A good sample allows for accurate estimates of the population. No sample is without some limitations.

Non overlapping elements of a population, ie. individuals, households, companies. Can be an individual or a set of elements.

Forms the basis of analysis from which conclusions are drawn and problems are solved. It is the basic unit from which data is collected. It is a descriptor ie. "A mid-level executive"

A physical representation of objects of the study. The actual list of sampling units. Important phase, often suffers from inadequate identification of elements in the population.

Formal specifications for methods and procedures of the study

Biases in the procedure

Innate when statisitcs are used. Can be estimated.

Inference based on a sample is more cost effieicient but in some cases a full census is required.

Closed

Exhaustive

Estimates infered from the sample.

Determines which design will produce the fewest number of errors.

Compares the costs of precision

Non overlapping groups

Method used to select the units of study.

Allows for estimation of the sampling errors and express confidence in a study.

Assign unique numbers to each population number then randomly draw numbers.

Use natural ordering. A mathematical process of selection.

Use random sampling for each stage

From every sampling unit a random sample proportionate to size of smapling unit

As above but also proportionate to variability in strata

Not proportionate to sampling size

Randomly select some sampling units.

Select clusters at random from every sampling unit

Utilizes 2 or more of the above listed types and determines if further designs are necessary

Chance selection procedures are not used. LEads to high variable error and lacks characteristics to estimate error.

Select subgroup of population based on qualitative information

Classifies populaiton based on important qualities. Desired porportion for each sample determined and quotas suggest equal representation from each class.

What is most convenient to the researcher

Rare characteristics selected first.