Skip to main content

12 – Sampling

Why Should I Care?


A bad sample will kill your scientific power. That’s all there is to it.



Definitions


Population:

Population                    A set of observable objects, which could be individuals, groups, documents, organizations, etc. that you are interested in studying.


Sample:


Sample                          Groups of objects, taken from a population, that are studied on behalf of the population. Will be used to generalize about the population.



CaveatCaveat: a warning or proviso of specific stipulations, conditions, or limitations. Especially important in the methodology section, to provide transparency, and  discuss the strength of the findings. Acknowledges problems in method. (WIKI)

 

Deciding to Sample


1st Question: can you avoid sampling?

                                                                        If n=N, best possible situation


Is it possible? Is it feasible?

Do you have enough time, resources, tools to collect data for the whole population?



A Good Sample


Generalization:

Generalization:                          To be able to use a sample finding, and assume it is accurate upon the population.


Causality: 

Causality: Finding the variable(s) responsible for a change in the dependent variable.



SOUP EXAMPLE


Population: 2 litres of vegetable soup in a pot


Sample: 1 spoonful


Random Selection?                     No, the bottom soup may never be tasted


Representative?                          Maybe, if you stir before you sample.


Type?                                           Non-random quota

The Power of a Good, Small Sample



 

EXAMPLE: 1995 QUEBEC REFERENDUM

See p. 73 – Results from polls are shown before the results from the actual referendum.


What was the population studied?                        Quebec adult voting population. ~ 5 million


What was the sample size?                                              ~ 1000


How accurate are the survey results?                                Very


How old were you when this happened?                           17

Types of Samples


Who should be in the sample?


The type of sample is the key to producing a research finding that is robust and credible. This applies to all disciplines and all research methods. P. 76-90


  1. Random Sample                                                          Generalizable


    • Simple Random SamplemathSample                                             math rule Xth


  1. Stratified Random SamplerespectSample                                         respect the population percentage of demographics


  1. Random Assignment is not Random SamplinggoingSampling        going to caf is not random


 


 
  1. 2. Non-Random SampleNOTSample                                                  NOT GENERALIZABLE

    • Accidental SamplefindSample                                                     find documents or objects          
    • Convenience SamplegoingSample                                                 going to a location where there are people
    • Purposive SampleplaceSample                                                      place where you know you will find XYZ people
    • Quota SamplerespectSample                                                             respect the population percentage of demographics
    • Snowball SampleoneSample                                                        one case leads to more, and more, and more…



Size of the Sample


How big the sample?                                                  Big enough for most sorts of phenomena to be recorded


Accuracy is not generalization.                                 This has to do with the margin of error. See QM.


Is 10 big enough of a sample to be accurate?         Depends on the size of the population.


 

Back to top