Transcript RM Chapter 4- Samples & Populations

Research Methods
Winter 2006
Chapter 4 – Samples & PopulationsChoosing or what to Study
Instructor: Dr. Harry Webster
1
Where Do Data Come From?
Anything that we can put into a number is data.
Good data requires valid measurements; an
appropriate way to investigate the topic.
Types of investigations in Social Science and
Commerce that usually have data are:
2
1. Observational Study: Uses predetermined
categories (target behaviors/events) and observes
frequency. Seeks to describe behavior or event.
Ex., observe children for helpful behavior. Have target
behaviors identified before beginning observation.
Define precisely helpful behavior.
Ex., p. 7 textbook. Compared proximity of residence
to power lines for children with leukemia to those
without leukemia.
3
2. Survey Studies:
Offer a series of questions usually to a large number
of people.
Ex., ability and attitude/opinion questionnaires.
Looks at how many people gave a certain answer
(frequency & percents).
2a) Sample surveys: A survey where much attention is
given to securing the sample in a random manner
where, as much as possible, everyone has an equal
chance of being picked.
Ex., public opinion polls, marketing research.
4
3. Census
Tries to gather data from everyone in a country.
Underestimates the homeless and some minority
groups.
Governments do this to establish voting districts,
economic and social trends.
For Canadian Census
4. Experiments
Researcher changes one thing (independent variable,
IV; called treatment in textbook); keeps everything
else the same, and determines if behavior (dependent
variable, DV) is affected.
5
Basic design is two groups where one group gets one
level of the IV and the other gets another level of the
IV (this includes its absence).
Since everything else kept constant, then any change
in behavior (DV) is due to the change in the IV.
Can make cause and effect (what causes what)
statements with experiments.
Cause and effect statements can only be made about
groups; not individuals. They are probable causes not
certain ones (remember: statistics is about probability
not certainty).
6
7
Ex., Does a tutor impact academic performance?
Group A
No tutor in RM
Group B
A tutor in RM
Both groups are otherwise the same.
Compare their grades in RM.
8
Vocabulary
Individual: anything that is measured; includes
people, objects, animals. The ‘thing’ that
provides the measurement.
Ex., person
Variable: any characteristic of an individual;
anything that we actually measure.
Ex., height, weight, schooling, singing ability.
9
Population: The entire group of individuals
about which we want information.
Ex, students, Canadians, Montrealers, workers,
dads.
Sample: The part of the population that we
actually measure on a variable. Too timeconsuming, often too costly, & difficult to
measure the whole population.
Ex., samples of 100 students, 1000 Canadians,
300 Montrealers, 500 dads are more easily
10
measured than the entire populations.
Exercises Chapter 1
1. Fewer women than men vote for the PQ. A
political scientist interviewed 400 male and
female voters and asked how they voted in the
last election.
a) Is this an experiment?
b) What is the population?
c) What is the sample?
d) What is the individual?
e) What is the variable under study?
11
2. Identify the individual, variable, population, sample
and type of study:
Researchers investigated how many homeless
people have chronic diseases by interviewing 100
homeless in the downtown Montreal area.
Individual:
Population:
Sample:
Type of Study:
(Stop Mon. 25 feb 08)
12
3. As a researcher you are interested in
whether students who sleep one hour
more per night would improve their
academic performance.
Design an experiment to investigate this.
Be sure to identify the independent
variable, the dependent variable, the
population and the sample.
13
Chapter 4 - continued
Samples, Good and Bad
Probability Sample
Any sample that is chosen by chance. Includes
Simple Random Sample (SRS) and Stratified
Samples.
1. Simple Random Sample
Gives each individual in the population an equal
chance of being chosen.
14
Use table of random digits (handout) or computer
program to choose individuals or samples.
Random: every individual or sample has an equal
chance of being picked.
This ensures the sample is representative of the
population.
Therefore, we can generalize the results back to the
population.
15
Choose a Simple Random Sample (SRS)
Step 1. Assign a number to every individual in the
population (or to every sample).
Assign numbers beginning with 0 and keep going until
everyone has a number. Go down columns of lists.
Step 2. Use random digits to select the numbers and
choose those people.
16
The first two lines of the Random Digit Table p 550:
Line
101
102
19223
95034
05756
28713
96409
12531
42544
82853
73676
47150
99400
01927
27754
42648
82425
36290
17
Ex., Choose a Simple Random Sample of 4 from the
following past players of the Canadiens. (number
down beginning with 0). Begin at line 101.
Ribeiro
Brisebois
Markov
Souray
Sundstrom
Hainsey
Ryder
Rivet
Hossa
Zednik
Dackell
Gratton
Koivu
Bouillon
Langdon
Perreault
Kilger
Theodore
Bulis
Quintal
Dwyer
Dagenais
Ward
Higgins
Juneau
Begin
Plekanec
Garon
Komisarek
18
The sample of 4 is: Komisarek (19); Hossa (22);
Perreault (05); Dackell (13).
19
3. 1200 college students, evenly divided per gender,
were asked about part-time work. 850 said they
worked at least 12 hours per week.
a) What is the sample?
b) What is the population?
20
2. A newspaper presents a survey where 500 of 1100
people coming out of the cinema said they went to see
an action movie.
a) What is the sample?
b) What is the population?
21
Chapter 4
Sample Surveys in the Real World
How sample surveys go wrong:
1. Sampling errors: errors that occur as we secure the
sample causing the sample to be different from the
population. Biased samples.
a) convenience sampling: grab them where you can.
ex., Malls, stores.
b) voluntary/self initiated response sample: interested
people decide to participate given a public request.
ex., Call in votes; internet polls.
22
2. Random sampling errors: difference between the
sample result (statistic) and the population result
(parameter) caused by chance in selecting the random
sample. Ex., margin of error mentioned in Box 4.1 on
page 80 and pp 96 to 99 in the text.
23