Study Guides
(238,295)

Canada
(115,061)

Queen's University
(4,208)

Sociology
(258)

SOCY 210
(33)

Vincent F Sacco
(10)

Final

# SOCY 210 final Exam Review.docx

Unlock Document

Queen's University

Sociology

SOCY 210

Vincent F Sacco

Fall

Description

Exam Review
The Logic of sampling
The reason we collect samples is because we want to use the sample to estimate
something about the population we are sampling from
Two major types of sampling
1. Nonprobability sampling
For what ever reason, when the units doing have an equal chance of being selected
Haphazard sampling (nonprobability studies)
A.k.a reliance on available subject
Use of those avoidable at a particular time and place
Convenient and inexpensive
Useful for pretesting questionnaires or other social measurement
Not representative of a larger population
Difficult to generalize
Example: recruiting subject on a street corner
Purposive/judgmental sampling
Sample selected on the basis of your own knowledge of the population you intent to
study and your research question
Useful when studying a small subset of a larger population, where member of the
subpopulation are easily identified
You are focused on a particular subgroup of a larger population
Used in the case of deviant case study, in the sense of statistical deviance
Snowball sampling
Already recruited responded provide the researcher with assistance in locating other
members of the population under study
Useful when members of a population are particularly hard to locate
Examples: sex workers, undocumented immigrants
Often used for exploratory purposes
Since difference members of the sample may know each other and share similar
characteristics, representativeness is problematic
Quota Sampling
Like probability sampling, it helps address the issue of representativeness of a sample
Begins with a table describing the characteristics of the target population
With this you try to assign equal proportions of people who belong to different groups to
your sample
E.g. is you know that 10% of all sociology students are female and international,
the you select 10 female international student for a sample of 100 sociology students
For this to work, the quota frame (matrix) used much be accurate and up-to-date
We are usually not interested in the generalization of sampling
2. Probability sampling
The selection of sampling units in a manner that respects the fact that each unit in a
population has an equal change of being selected
This is the favorable form of sampling
The reason probability sampling is better, is because when doing an experiment, we have
to believe that all the people being selected is completely random, and that it demonstrate a
real and appropriate microcosm of the population
The theory and logic of probability sampling Conscious and unconscious sampling bias
The population of 100 has 44 white women, 44 white men, 6 aboriginal women
and 6 aboriginal men
Lets say that I enter this population and the first 10 people I meet are the white
women. I select them for my study. Because I did not select a representation of the
population (say, 2 white men, 2 white women, and 3 aboriginal men) I have a
biased sample.
Basic probability sampling jargon
Representativeness: a sample is said to be representative of a population if its aggregate
characteristics closely approximate those aggregate characteristics in the population
Basic principles: ALL member of a given population should stand an equal change
of being selected for particular sample
Population: theoretically specified aggregation of the element in a study
E.g. Canadians, sociology student
Sample: Potion of the population from which information is collected
Sampling Unit: Elements considered for selection during some state of sampling
The emblements of a population that are available and have an equal change of
being selected to form the sample
i.e. those in the sampling frame
Sampling frame: list of quasi list of elements form that a probability sample is selected
Complied list of individual sampling units
Mist be up to date
In some cases may be unrepresentative
A summary description of a given variable
E.g. Proportions of people, who smoke, mean income
In a population is called a parameter
The corresponding number of a population
In a sample is called a statistic
Simple random sampling
Most basic method of random sampling
Research participants selected at random from a sapling frame
Computer-assisted
Table of random number
th
October 30 2013
Sample types
Probability vs. Non-Probability Sampling
Probability
Most common in social research, particularly quantitative
Representative of larger population
Elements of a population stand equal change of selection
More complex
Non-probability
Simple
Common in qualitative studied
May be unrepresentative
Elements of a population do not stand equal change of selection
Methods of Probability Sampling
The theory and logic of probability sampling
Conscious and unconscious sampling bias
The population of 100 has 44 white women, 44 white men, 6 Aboriginal women, and 6
Aboriginal men Let’s say that I enter this population and the firs 10 people I meet are the white women. I
select them for my study. Because I did not select a representation of the population (say, 3
white men, 3 white women, 3 Aboriginal women, and 3 Aboriginal men), I have a biased
sample
Systematic sampling
Ever K element in the sampling frame is chosen for inclusion in the sample
To avoid bias, choose the first element from the list at random
The way it works is, you have a sampling frame (a list of all the items in the population)
and then you would select your sample by determining a sampling interval and than going
down the list and take every 10 name until you got the sample you wanted
Sampling interval: the standard
distance (k) in the list between the
elements selected for sample
Calculated as sampling
interval= population size/sample
size
Sampling ratio: the proportion of
elements in the study population
that are actually selected
Sampling ratio= sample
size/population size
Inverse of the sampling
interval
For SOCY 210: Sampling Interval=
193/50=3.86
Sampling Ratio= 50/193= 0.259
The problem with systematic sample is if there is a commonalty or pattern among the list
of names
Stratified sampling
You begin with a table describing the characteristic of the target population and dividing
it into homogenous subsets (strata)
E.g. Race, gender, nationality
A proportion of individual from each strata is assigned to the sample, matching the
proportion of individual from that group who belong to the population at large
More representativeness
For example if you are really concerned with the proportion of diversity in the sample is
equal to actual proportion. But we want it to be completely random
So you stratify the sample, if you have a sampling frame of all the people, you
than rearrange the list into strata (vertically ranked categories) for example, the
women at the top and the men at the bottom. So you than go in and to a simple
random sample from the first group (female) and than the second group (the male
group)
It doesn’t produce error because sampling error is produces by how
heterogeneous (or mixed) the population is, the greater the mix, the more changed
of a sample being misrepresented
Types of Sampling Design
Stratified sampling
Ex. Sampling university students Step 1
Step 2
From the Step 3
university roster, Then, select by
select only 15,000 university classes
full-time degree Sample size is set to
students be 1,000
Then, set the
sampling program for
a 1/15 sampling ratio
A random number
from 1-15 will be
generated
• The student having
that number and every
fifteenth student
Step 4
thereafter will be
selected in the sample.
Multistage cluster sampling
Cluster sampling
Used when it is either impossible impractical to complex an exhaustive list of the
elements that compose a large population
Population elements may already be grouped into subpopulation, of which list actually
exist
Researchers can randomly choose a sample of these subpopulations and then
sample individuals that belong to those selected subpopulations
The sampling is occurring at more than one level – the provincethe institutions the
people in the institutions
Example of multistage cluster sampling
Dr. C. Couture wants to study the demographics of university students who belong to
Fraternities across Canada
How would he gather a random sample of 500 individuals?
It would be impractical to create a sampling frame with the list of every single member of
a fraternity in Canada
He can use multistage cluster sampling instead
1. Make a list of all the provinces and territories (primary sampling frame) and puck 5
at random (either through SRS or systematic sampling)
These will be your primary sampling units
2. Make a list of all the fraternities in each of theses provinces. Territories (secondary
sampling frame) and pick 5 fraternities from each list
These fraternities will from part of your secondary sampling units
3. Take the list of member of each of 50 fraternities selected (tertiary sampling frame)
and select 20 individuals at random from each fraternity
This will provide you with 500 subjects Random digit dialing
The difference between telephone directories and RDD techniques
Generating a sample of telephone numbers
The increasing reliance on RDD
We would generate a list of the entire telephone prefix in the telephone code 613-555-
2232
Than we take a random number table and it generates for number suffixes and attaches
them to the prefixes that may or may not be actually telephone numbers
So this list of phone number and it is given to interviewers, who are given a very specific
way of dealing with these numbers
This will generate as good a sample of households, as any other sampling method
An Intro to probability theory (the introduction to inference)
Parameter vs. Statistic
Parameter: the summary description of a given variable in a population
Statistic: The summary description of a variable in a sample
Through statistics we try to infer the true value of a population parameter
How close is the statistic to the parameter? Can we use this as an estimate?
So how do we know if it is going to be accurate?
The sampling Distribution
If many independent random samples are selected from a population and a statistic
calculate in each of them, the sample statistic provided by those samples will be distributed
around the population parameter in a bell-shaped fashion
The bell (or normal) curve is called the sampling distribution
So if we took a very large number of samples, and calculated the statistic in each
case, those statistics would form a distribution, the from of a normal distribution, or
a bell curve
But we don’t ever take thousands of samples, but because we know that the
parameter is going to be the middle of the curve, we can make a general assumption
that all of the samples are going to be relatively accurate
Value of the statistic on the x-axis
Number of samples that yield this statistic on the y-axis
The samples will begin to group themselves around the parameter
8 Parameter ~ 5.5
7
6
5
4
3
2
1
0
$0 $1 $2 $3 $4 $5 $6 $7 $8 $9 $10 There are 10
people in the population; this is how much money they have
What is the average amount of money in a population?
If you sampled 1 person it would look like this
If we are only sampling 1 person, there is a very
good chance that our sample is going to be very off
Sampling 2 people would look like this
Narrowness is
really important
to know
Because it would be good to know how tightly the
sample is to the parameter
If we are looking at a narrow curve, it is more likely to
be close to the parameter
So we want to be able to calculate the sampling
distribution, and how accurate and close to the
parameter is
We want to know how spread out or condensed the
distributions is
Standard and sampling error
We can also calculate how closely sample statistics are clustered around the true value of
the population parameter
i.e. estimate the degree of error to be expected for one given sample design : the
sampling error
This is the difference between ONE sample statistic and the population
parameter
So on average, how close are we to the center, and to know this it will be very helpful for
us to now how wide or narrow the distribution is
Calculating (estimating) the S.E from a sample binomial
s.e- √ (P x Q / n)
P is the population parameter we want to calculate
E.g. the proportion of sociology students who were born in Canada
Q= 1-P
n is the number of cases in each sample The standard error indicates how tightly sample estimates will be distributed around a
population parameter
If small, the samples statistic most closely resemble the population parameter
The formula assumes we do no the population parameter
Standard error
Suppose in Socy 210, 70 percent support the use of a study guide and 30 percent oppose
it, what in the standard error of the sampling distribution?
√.7 X .3/50 = .064 or 6 percent
Our sample yields a figure of 67 percent in favour
We know that 95 percent of the samples of size 50 would fall between 58 and 82
A property of normal curve is that even
though the standard error of the normal
curve will differ from one to another, it is
the case in every normal distribution; the
cases will be distributed in a particular
proportionate way, given the sizes of a
standard sample
Each line down, is one more error
away, and so the first line ever cases will
be within plus or minus 1 error away, the
second line will be plus or minus 2 curves
away, and the 3 line is plus or minus 3
errors away
So we would be able to get the interval of where the parameter lies, because it is in plus
or minus of where our errors is
S.E in the normal distribution
Certain proportions of the sample statistics will often fall within a specified number of
standard errors from the population parameter
Approximately 68% of sample statistics will fall within 1 and -1 s.e. of the
population parameter
95% of sample statistics will fall within 2 and -2 s.e. of the population parameter
99% of sample statistics will fall within 3 and -3 s.e. of the population parameter
Rule 1
First, if many independent random samples are selected from a population, the sample
statistics provided by those samples will be distributed around the population parameter in
a known way.
Rule 2
Probability theory gives us a formula for estimating how closely the sample statistics are
clustered around the true value. In other words, probability theory enables us to estimate
the sampling error—the degree of error to be expected for a given sample design. This
formula contains three factors: the parameter, the sample size, and the standard error (a
measure of sampling error).
Rule 3
The standard error is also a function of the sample size—an inverse function. As the
sample size increases, the standard error decreases.
Rule 4
Another general guideline is evident in the formula: Because of the square root in the
formula, the standard error is reduced by half if the sample size is quadrupled.
Level of confidence
We can express the accuracy of a sample statistic in terms of the level of confidence that
the statistic falls within a specified interval from the parameter The more confident we wish to be, the larger the error is
Confidence interval: A range in which we expect a sample statistic to lie for a given
percentage of the time (confidence level)
E.g. 95% or 68% of the time we expect a sample statistic to fall between a ranges
of number (the confidence level
Calculating confidence intervals
Prof. J. Benhaim wants to estimate the proportion of McGill students who voted in the
past Canadian Federal election. To do so (instead of asking every McGill student if they
voted or not) she collects a random sample of 100 students. She finds that 35% of those
students voted in the past federal election.
How would she go about getting a 95% confidence interval of McGill students who voted
in the past federal election?
Calculating Confidence Intervals
1. Calculate the standard error
s.e. = √ (P x Q/ n)
P= 0.35; Q= (1-0.35) = 0.65
n= 100
s.e.= √ (0.35 x 0.65/100) = √0.002275 = 0.048
2. Calculate the confidence interval
95 % c.i. = P 2 x s.e.
= 0.35 2 x 0.048
= [0.254, 0.446] or between 25.4% and 44.6%
The sampling error is greatest the population is heterogeneous, or when the division is
50/50 we will have the largest sampling error
1x9=9; 2x8=16; 3x7=21, 4x6=24; 5x5=25
So we calculate the confidence error at 95%, which means that we are 95% confident
that the real population parameter is between 25.4% and 44.6%
Example
We want to undertake a study of Queen’s student to determine how man drank to a point
of inebriation on the weekend
We decided to interview 500 people and to employ some kind of random design
Now… we know that out random sample will come from all possible samples of 500
Refer to note Qualitative field research
Why Qualitative field research
Social research ‘right where it happens’
No armchair philosophy’ but direct involvement
No Artificial setting – eg. Experiments
Deeper understanding of social phenomena
Probing social processes over time as they happen
Good to study social processes over time as they happen
Data collecting AND theory generating
If you really want to understand human actions, you have to try and understand
human action within its own framework
In order to really make sense of what people are doing in their worlds, you have
to get back inside their worlds
Because it is unique, it comes with a lot of traps and problems
This is no necessarily the easiest thing to do
Because it is research that takes place over time, it has a remarkable type of
flexibility those others to do not have
Historical context
The influence of the University of Chicago and in particular the writing of Robert Park
Because he has lived the life of a newspaper reporter, he has a real sense for the
complexity of the city, specifically what happens in isolated dark communities
So, as an academic he encouraged his students to just go out and explore and
study
Types of fields of studies
Case Study Design
Common in qualitative fieldwork
Not a mode of observation, but a type of research design where attention is paid to a
single instance of some social phenomenon
Se in both qualitative and quantitative research
Sometimes used as preliminary to a more elaborate study—exploratory purpose
Any study that we focus a particular fashion on a small and specific group is a
case study
Christie Pitz baseball riot
Extended case method
Developed by Michael Burawoy and colleagues
Used to discover flaws and modify exisiting social theories
1 step: researcher enters the field with a clear expectation of what to find
Become familiarized with existent literature
Step: notes how observations conflict with extant theory and whether they support or
reject what already exist
Theory Observations=largely deductive
Grounded theory approach
Derives theory from analyzing patterns and these discovered in observational data
Researcher does NOT begin with preconceived ideas—but instead allows theory to
emerge from the data
Balance subjectivity/objectivity by thinking comparative, gaining multiple points of view,
fathering data in a variety of ways, checking with respondents, maintain skepticism
Suppose you know absolutely nothing about catholic religious observant, so what
you are doing is going in a Catholic Mast, your first pass at this would be to
determine the major pieces of Church So you come up with categories, you conceptualize them, you relate them and
than you begin to understand and put together the pieces of what is going on
Shortcoming of not having structures theoretical framework are overcome by rigorous
data analysis
Methodological choices guided by an alternating process of fata collections and
analysis
Coding procedures assist the researcher in both systematically and creatively
identifying, developing, and relating concepts
Observation Theory= largely inductive
Approaching the field without preconceptions
Ethnographies
Rooted in naturalism- observing events and people in a natural setting; this is important
scientific because it considers what people do not do what they say they do
Aim for deep understanding of someone’s community’s way of life
Involve
Direct first-hand observation
Note taking, photographing, drawing
Interviewing
Archival analyses
In some cases, concealing one’s real identity
Preparing to hit the field: Getting In
Gather the relevant literature
Discuss your field with those who have studied it or know something about it
Gaining access to the field
Identify the Gatekeeper: formal access and consent by contacting the person in
charge
Seek out an Informant: establishing contact with a member who introduces you
to the field
Joining the group: securing membership status being hired within an
organization
In some cases deception is used
Ethical concerns-balance between harm to subjects and usefulness or
research
Deception requires you to play a certain role, that you are not prepared to play—you
cannot slip
Establishing Initial Contacts
Develop a rapport with your subjects in order to gain their trust and make them feel
conformable
Your initial contact may influence you subsequent observations, the data you are able to
collect, and the people you are able to reach
Deciding how much to disclose of your study to your subjects should depend on the
nature and purpose of your research
As well as ethical considerations!
Voluntary participation, informed consent
The various roles of the researcher
You initial contact depends on your role as a researcher
Complete observer: observes the field but does not participate in the activities that take
place
Participant observer: reveals that she/he is researcher but still participates in the
activities that take place in the field
Complete participant: engages fully in activities with other members Note-taking
Take full and accurate notes of your observations
If possible, take them as your make them, or as soon as you can
Do no trust your memory more than needed
Take quick notes at first, more elaborate ones later
Process is made easier by having standardized recording forms, or codes that can be
easily jotted down
Notes include empirical observations AND a researcher’s interpretation of them
Should be distinguishable
You not only want to write down what happens, you also want to write what you
this what happens means
Included as much detail as possible—later on decide what is relevant
Leaving the field
Fieldwork should end when theory building reaches a closure
Leaving can be disruptive for both researcher and subjects
Types of exit (quick exit, slow withdraw)
Exit strategy should depend on the field and the relationships developed
You would do this if:
Your position becomes intangible (people find out, are upset with you, it’s dangerous,
your research is over)
Strengths and weaknesses of field research
Advantages of field research
In depth understanding of social phenomena and subtle nuances in attitudes and
behaviours
Flexibility
May or may no be expensive
Validity: Comprehensive measure which tap into great depth of meaning, instead
of relying on concepts, detailed researchers often give detailed illustrations
Disadvantages of field research
Reliability: Measurements depend on the researcher making observations, the
researcher may have an impact on the field and the data she/he collects
It involves a humanistic aspect of research that other research doesn’t have
Institutional Ethnography
Developed by Dorothy Smith in conducting research from the standpoint of women
Pays close attention to the voices of subordinated groups and how their experiences are
shaped by institutional practices
While looking at personal experiences, the goal is to uncover deeper institutional and
structural power relations that given them
Uncover forms of oppression that are often overlooked by more traditional types
of research
Participatory actions research (PAR)
Critiques the distinction between researcher and subject
Challenges the usefulness of classical objectivity
The researcher becomes a resource to those being studied (typically disadvantages
g

More
Less
Related notes for SOCY 210