Data Collection

1
Data Collection Sampling

• Dr. Guerette

2
Gathering Data

• Three ways a researcher collects data
• By asking questions
• By direct observation
• By using written records
• The fundamental issue in data collection is its
  representativeness.

3
Logic of Probability Sampling

• Allows researchers to generalize to unobserved
  cases.
• Conscious and Unconscious Sampling Bias
• Bias in this case means that selections are not
  typical or representative of the population from
  which they are drawn. This could be result of
  personal beliefs or many other reasons.

4
Logic of Probability Sampling

• Representativeness and Probability of Selection
• All members of a population have an equal chance
  of being selected for a sample in probability
  selections.

5
Probability Theory Sampling Distribution

• Probability theory permits inferences about how
  sampled data are distributed around the value
  found in a larger population.
• Sample Element
• The unit about which information is collected and
  it provides the basis of analysis. It is the
  grouping of study elements.

6
Probability Theory Sampling Distribution

• Population parameter
• The summary description of a given variable in
  the population.
• Sample statistic
• The summary description of a given variable in
  the sample.
• Sampling distribution
• The range of sample statistics that would be
  obtained when many samples are selected.

7
From Sampling Distribution to Parameter Estimate

• Sampling frame
• A list of elements in the population
• Binomial variable
• A variable that has only two values
• Estimating sampling error
• When independent random samples are selected from
  a population and sample statistics are calculated
  from those samples they will be distributed
  around the population parameter in a known way.

8
From Sampling Distribution to Parameter Estimate

• Standard error
• Is the way of estimating how closely the sample
  statistics are clustered around the true value.
• Confidence levels and Confidence Intervals
• Use probability theory to indicate sample
  estimates that fall within one, two or three
  standard errors of the parameter.

9
(No Transcript)
10
Standard Error Diagram
Source Jeremy Kemp (2005)
11

• SE vp(1-p)/n
• Or
• SE vp x q/n

12
In Class Exercise Probability Sampling

• You have selected a probability sample and want
  to determine the sampling error in order to
  estimate the population parameter. In your sample
  you compute that 70 percent of your sample
  opposes establishing a hurricane relief fund
  derived from Miami-Dade tax dollars, while 30
  percent favor such a fund. You had a sample of
  400 (N 400).
• Report your confidence level and confidence
  interval.

13
Probability Sampling

• Simple random sampling
• All elements in the population have an equal
  chance of being selected for the sample.
• Systematic sampling
• Drawing a sample of every Nth element in the
  population.
• Stratified sampling
• Based upon prior knowledge of a population, a
  sample is drawn that will offer a greater degree
  of representativeness.

14
Probability Sampling

• Disproportionate Stratified Sampling
• Specifically produce samples that are not
  representative of a population on some variable.
• Multistage Cluster Sampling
• Used when populations cannot easily be listed for
  sampling purposes. Generally involves geographic
  dispersion. While this technique increases
  efficiency it decreases accuracy.

15
Probability Sampling

• Multistage Cluster Sampling with Stratification
• Used to increase the homogeneity of the sample.

16
Non-Probability Sampling

• Used when the likelihood of any given element
  will be selected is not known (e.g. when random
  probability sampling is not possible).
• Purposive or Judgemental Sampling
• Selection based upon prior knowledge of the
  population.
• Quota sampling
• Uses a matrix or table to describe the
  characteristics of the population and the sample
  is drawn to reflect the cells of the matrix.

17
Non-Probability Sampling

• Reliance on available subjects
• Using people that are readily available seldom
  produces data that have great value.
• Snowball Sampling
• Begins by identifying a single member of a
  population and then having that subject identify
  others like him/her.