Getting started

1
Getting started

• Chapters 1, 2

2
Terminology

• Variable any characteristic you may want to
  study age, income, political party affiliation,
  view on social issues, etc
• Individuals (or elements) the people,
  households, counties, states, etc from which you
  draw your data
• Census when you ask everyone in the population
• The US government conducts a census every ten
  years. They are expensive and time-consuming but
  yield lots of information.

3

• Population the entire group of people, counties,
  households, etc you want information about
• Sample the small subgroup of the population you
  actually ask

The idea of sampling is that we ask only a small
portion of the entire group of people from which
we want information. Then we use this information
to infer about the population.
4
Observational study versus experiment

• An experiment takes individuals and imposes a
  treatment to see what happens, like randomly
  assigning students to either computer or
  traditional classrooms and then comparing their
  grades to judge the success of technology in the
  classroom.
• An observational study or sample survey, on the
  other hand, just observes individuals without
  imposing treatment, like simply comparing grades
  of students in both computer and traditional
  classrooms after theyve taken the courses.

5

• A big advantage experiments have over
  observational studies is that they give better
  evidence for cause and effect. By randomly
  assigning students to either computer or
  traditional classrooms, we can rule out other
  factors that could affect the results. By ruling
  out other factors, such as student preparedness,
  we can say what effect computers have on student
  success. By simply obtaining the students grades
  afterwards, we do not rule out other possible
  factors.

6
Terminology

• Simple random sample (SRS) the group of
  individuals you randomly choose from the
  population to sample
• Convenience sample a sample composed of people
  who were chosen because they were convenient to
  locate. Often professors will base their research
  on surveys taken from their classes these are
  convenience samples.
• Voluntary sample a sample composed of people who
  answered voluntarily. They were not selected
  specifically by the samplers. Phone-in polls are
  examples of this.

7

• Bias the design of a study is biased if it
  systematically favors certain outcomes.
• If we want to know which candidate the people
  are supporting for president, we would not want
  to conduct our sample outside the Republican
  National Convention. The sample would be slanted
  heavily toward the Republican side. It would not
  be fairly representative of the entire population
  and so would be biased.

8
Random sampling

• Convenience and voluntary samples are not good
  ways to sample. They are often biased. This means
  they often weight one outcome over others. For
  instance, phone-in polls asking for political
  opinions often attract people who feel strongly.
  Whereas, people who do not feel strongly do not
  respond. The sample only represents the vocal
  minority.
• To avoid the biased nature of these types of
  samples, we use random sampling. Every individual
  has an equal chance of being selected for the
  sample. This way, we get a sample that is
  representative of our population and gives us
  accurate data on their opinions and habits.

9
How to randomly sample

• Lets say we have 250 adults in our population
  and we want to sample ten. Well use Table A (pg
  545) to randomly select our sample. This is a
  table of random digits.
• 1.) Assign everyone in population a 3-digit
  number, 001 to 250.
• 2.) Look up 3-digit numbers on table.
• 3.) Pick out first ten that match the numbers
  001 to 250. Do not use repeats.
• 4.) Ask these ten people for information.
• 5.) Use sample data to infer about the entire
  population.

10
Reading across line 101, we have 192, 239, 503,
405, 756, 287, 139, 640, The first three in our
sample are individuals 192, 239, and 139. The
other numbers are ignored. We go in this fashion
until we have selected ten people.
11

• Ignoring the last digit of each row, we end up
  with the sample containing the ten people
  numbered 192, 239, 139, 099, 019, 170, 005, 113,
  074, and 001. We would then ask these ten people
  and they would represent the entire population.
• By the way, you might be thinking that ten
  people would be too small of a sample to do much
  good. Youd be right! The sample is kept small
  for demonstration purposes.
• Chapter 1 homework 2, 3, 4, 8, 12, 15
• Chapter 2 homework 2, 3, 5, 9, 13, 15, 16