Textbook Notes (290,000)
CA (170,000)
U of G (10,000)
MCS (700)
Chapter 2

# MCS 3030 Chapter Notes - Chapter 2: Nonprobability Sampling, Sampling Fraction, Quota Sampling

Department
Marketing and Consumer Studies
Course Code
MCS 3030
Professor
Hetherington
Chapter
2

Page:
of 6
Chapter 2: Sampling
Population- group you want to generalize to and the
group you sample from in a study
Theoretical Population- former population who you want to generalize to
Sample-actual units you select to participate in your study
Sampling- process of selecting units (ie. People, organizations) from a population of interest so that
studying a sample you can generalize results to population from which units were chosen
Sampling frame-list from which you draw sample (sometimes no list and samples drawn from explicit
rule (ie. When population is people who walk by then sampling frame is population of people who pass
by in time frame and rules you use to decide whom to select)
External Validity
External validity- degree to which conclusions in study would hold for other persons in other places and
at other times
-related to generalizing a sample to population
-refers to approximate truth of conclusions that involve generalization
generalizability- degree to which study conclusions are valid for members of population no included in
study sample
-two approaches to provide evidence for generalization:
1.Sampling model- model for generalizing in which you identify your population, draw a fair sample,
conduct your research and finally generalize results to other population groups ( Population-Draw
Sample-Generalize back to Population)
Problems:
-at time of study you may not know what part of population you will want to generalize to
-may not be able to draw a fair or representative sample easily
-impossible to sample across all times that you may like to generalize to (like next year)
2. Proximal similarity model model for generalizing from your study to another context based upon the
degree to which the other context is similar to your
study
-think about different generalizability contexts and
develop theory about which contexts are like your
study and which are less so (place different
contexts in terms of relative similarities and then can generalize results of study to other people, places
or times)
Proximal- means nearby
Gradient of similarity- dimension along which your study context can be related to other potential
contexts to which you might wish to generalize. Contexts that are closer to yours along the gradient of
similarity of place, time, people, and so on can be generalized to with more confidence than ones that
are further away.
Threats to External Validity
-explanation of how you may be wrong in making generalization (people, places, times)
Improving External Validity
Based on sampling model:
-do a good job drawing sample from population
-use random sample rather than nonrandom procedure
-random selection-process or procedure that assures that the different units in your population
are selected by chance
-once selected, try to ensure respondents participate in study (keep dropout rates low)
Based on proximal similarity model:
-could do better job describing ways contexts differ from others (provide data about similarity
between various groups of people, places, and times)
-map degree of proximal similarity among various contexts with concept mapping
Concept mapping-two dimensional graphs of a groups ideas where ideas that are more similar
are located closer together and those judged less similar are more distant (often used by group
to develop conceptual framework for research project)
-show critics they’re wrong (do study in variety of places, with different people at different times)
Sampling Terminology
Census- kind of survey that involves a complete enumeration of the entire population of interest
Statistical Terms in Sampling
Response- specific measurement value that a sampling unit
supplies
Statistic- process of estimating various features from data,
often using probability theory
-used when you look across the responses for your
entire sample
Population parameter- mean (average) you would obtain if you were able to sample entire population
The Sampling Distribution
Sampling distribution- theoretical distribution of an infinite number of samples of the population of
Bell curve- smoothed histogram (bar graph) describing expected frequency for each value of a variable
Standard deviation (SD) - spread of variability of the scores around their average in a single sample
-square root of variance
Standard error- spread of averages around the average of averages in a sampling distribution
Sampling Error
-in sampling, standard error is called sampling error
Sampling error- error in measurement associated with sampling
-gives you some idea of the precision of your statistical estimate
-low sampling error means that you had relatively less variability or range in sampling
distribution
-calculate sampling error by basing calculation on standard deviation of your sample
-greater sample’s standard deviation, greater standard error (and sampling error)
-greater sample size the smaller the standard error (because greater sample size the closer your
sample is to actual population itself)
-if take sample that consists of entire population than there is no sampling error because you
don’t have a sample, you have the entire population (consensus) and therefore the mean you estimate
is parameter
-estimate standard error (sampling error) based on standard deviation
Systematic error- random errors in measurements are caused by unknown and unpredictable changes in
the study (changes may occur in measuring instruments or environmental conditions)
Systematic error- systematic errors in measurements usually come from the instrument or the person
conducting the study (may occur because there’s something wrong with instrument or in data handling
system or because instrument is wrongly used by the experimenter)
-types: sample frame, non response
The 68, 95, 99 Percent Rule (Empirical Rule)
-applies to standard deviation and standard error
-symmetrical distributions =bell curve=normal distribution
-one standard unit -68% of cases in distribution
-two standard units- 95% of cases in distribution
-three standard units-99% of cases in distribution
-if you have sampling distribution, you should be able to
predict 68, 95, 99 percent confidence intervals for where
the population parameter should be