• Know the three Factors in Identifying a Population for a Research Project
• Advantages of Sampling.
• Learn the three factors influencing the representativeness of a sample.
• Population - the group that we wish to generalize about.
• Sample - can’t study everyone in the population sometimes; therefore, we
select a smaller group (sample) that is representative of the population
under study and from the statistical analysis on this sample, we can make
generalizations about the population as a whole.
3 Factors in Identifying a Population for a Research Project
• Unit of Analysis - individual MPs
• Geographic Location - Canada
• Time Period - those serving from1993-2000
• Example: Instead of studying “Members of Parliament,” you would state that
you are studying “Canadian MPs between 1993-2000.”
Advantages of Sampling
• Less Expensive
• Restricted to a Certain Time Frame
• Less Data Collection & Entry
• Sampling can Provide Accurate Estimates of the Population Parameters.
• *** Note: We are ultimately interested in the population & the population
parameters; the sample & the sample statistics are merely a means to these
Representativeness of the Sample
• Three factors influencing the representativeness of a sample:
• (1) the accuracy of the sampling frame
• (2) the sample size
• (3) the method by which the sample is selected
• All three factors are important - a weakness with respect to one cannot be
compensated by strength with respect to another.
• This is simply a list of all the units in the target population. If our target
population is Canadian MPs serving from 1993-2000, our sample frame
would include all MP who were in Parliament during this time period.
• For this type of population it is not as hard to get everyone compared to a
national opinion research population. • Some problems even with this small MP population: Not all MPs would be
willing, not all alive, might not find some if they were defeated or resigned,
might forget to include MPs who won during by-elections, etc.
• The challenge is to find a sampling fame that minimizes inaccuracies in the
sample frame - one way is random sampling.
• Not all target populations have a population with every person listed with
• Rule of Thumb: Sample statistics are more likely to be closer to the
population parameter when the sample size is larger than when the sample
• Our goal is to reduce error, therefore we prefer larger samples.
• To determine the appropriate sample size, we need to consider a number of
• (1) the homogeneity of the sample
• (2) the number of variables under study
• (3) the desired degree of accuracy
• (4) the method of random sampling used
1. Homogeneity of the Sample
- This refers to how similar a population is with respect to the variable of
interest. (If all our Canadian MPs who served from 1993-2000 had the exact
same opinions on a topic we would not need a large sample.)
- Heterogeneity refers to how dissimilar a population is with respect to the
variable of interest.
- We want to estimate how homogeneous or heterogeneous our population is -
a highly homogeneous population allows us to use a smaller sample, whereas
a highly heterogeneous population requires a larger sample.
- The appropriate sample size increases as we move along the continuum from
homogeneity to heterogeneity
2. Number of Variables Under Study
• The more complex our study, the more variables and relationships that we
include, the more cases we need in our sample.
• The need for a larger sample stems from the desire to look at the subgroups
within the sample and to impose statistical controls.
• If we want to look at visible minority MPs, then our sample would have to be
larger in order to include more non-white MPs.
3. Desired Degree of Accuracy
• Researcher can state the margin of error that they are willing to accept.
• Knowing the margin of error allows researchers to state their sample
statistics as a confidence interval. 4. Method of Sample Selection in Mostly Quantitative Statistics
• Probability sampling can be conducted in several ways, the three most
(1) Simple Random Sample
(2) Stratified Sample
(3) Cluster Sample.
• Error varies with the different probability sampling approaches.
• Stratified sampling is more precise than simple random sampling.
• Cluster sampling is less precise than simple random sampling.
Simple Random Sampling
• All the cases are listed and assigned numbers. Through computer selection
or by use of a table of random numbers, cases are selected until the desired
sample size is