Schuldig Oder Unschuldig?

Sampling Process

Sampling is about taking a small portion of the larger part and drawing conclusions from it by using our experience of the smaller part. Put simply, sampling is finding out a lot from a little, in a scientific way.

From a practical marketing application, sampling asks the questions:

Who do we want to talk to? Who has the needed info?

The aim of sampling is to make inferences about a population within acceptable margins of error.

Sampling Terminology, Basic sampling concepts

Population

Total set of people, shops, items, members and/or objects of interest.

The term �population’ refers to every, and all members from whom information is needed. A pop. member is one individual case from the total…

пÑ" Students UL

The Pop. Is what we make inferences about, based upon info from the sample.

Symbol N denotes the size of a population.

Sample

The aim of a sample is to enable the researcher to draw inferences or concl. from the sample about the pop (number of elements in a sample = n)

Census

Where all members or cases of a given pop. are observed/measured.

Parameter

Numerical dimension, measure, value or characteristics of the pop. we are interested in

пÑ" pop. Parameter / Sampling is used to estimate a population parameter

Probability laws

Mathematical laws governing the likelihood of a certain event occurring.

Estimates can be made to determine how close the sample mirrors or reflects the pop.

Thus, a correctly selected sample will be, within measurable limits, representative of the pop. of interest.

Sampling error

Sampling error is the probable measurement of difference between the actual pop. parameter and the observed sample estimate of it.

CENSUS OR SAMPLE?

Census is a complete canvass of all pop. members, whereas a sample is a selection of a smaller number of members from the pop.

Why do we use a sample, not a census?

Time Requirements

General rule: Samples take less time than a census, except the pop. is very small, finite, contained, and accessible.

For us, it’s not poss. to do a questionnaire on all 10k UL students, because of the time limit and the fact that we won’t be able to ask the students in a row.

Financial costs

By considering that printing one black/white page costs 6c, it is too expensive to ask all students

Required Information, Related Decision

пÑ" goes with the time requirements:

A sample can allow the researcher to control other dimensions of the project.

пÑ" goes with costs: a more efficient use of money and time can be made by undertaking a sample…

If properly conducted, a sample can be nearly as accurate as a census

Pragmatic Reasons

By using a census, however, it can happen that people become sensitised to the topic. Therefore, the results of the data collection process may not be representative of the population’s true opinion on the topic.

Accuracy and Confidence

пÑ" The final issue to be considered in choosing between a sample or census

1. sampling error can be calculated and so is subject to statistical estimation, thus the accuracy of the results can be defined;

2. a census is more prone to non-sampling errors, e.g. interviewer bias, analysis error;

3. a sampling error can be reduced by increasing the sample size (sometimes impossible)

пÑ" samples are the preferred choice of most researchers (more convenient, easier and more economical than a census)

The sampling plan

Stage 1 вЂ" define the population (element: UL Students, unit: - , time dimension: now, geographic boundary: UL)

Stage 2 вЂ" select sample frame (not in our case)

…3 - choose a sampling method (non-probability)