And You Believed That?!

A solid knowledge of statistical procedures will help you be an educated consumer of information. Every day, we are confronted by a news report citing a new relationship researchers have discovered or a claim being made in advertising. Beginning with this task, we will examine the questions you should keep in mind when analyzing such claims.

Read the article below.

Facebook use linked to less textbook time
By Mary Beth Marklein, USA TODAY (April 13, 2009)
Does Facebook lead to lower grades? Or do college students with lower grades use Facebook more than their higher-achieving peers?
A study of 219 students at Ohio State University being presented at a conference this week doesn't answer those questions definitively. But it suggests a link between the social networking site and academic performance.
Students who said they used Facebook reported grade-point averages between 3.0 and 3.5; those who don't use it said they average 3.5 to 4.0. Also, Facebook users said they studied one to five hours a week, vs. non-users' 11 hours or more.
Ohio State doctoral student Aryn Karpinski, who conducted the research with graduate student Adam Duberstein, says the study is too narrow to conclude that Facebook and academics don't mix.
"It cannot be stated (that) Facebook use causes a student to study less" or get lower grades, she says. "I'm just saying that they're related somehow, and we need to look into it further." Of the 68% of students who said they used Facebook, 65% accessed the site daily or multiple times daily.
Karpinski says 79% of Facebook users believe it has no impact on their academics; some say it helps them form study groups.
She says faculty ought to consider harnessing it as a learning tool. Yet a preliminary peek at a second survey suggests "a lot of faculty … didn't even know what Facebook is," she says.

1. What claim is being made? What can we NOT conclude from the study?

Bias in Research and Studies

The first question raised when evaluating the believability of a claim is whether or not the questions and procedures when designed in such a way as to eliminate bias. It is critical for statisticians and researchers to avoid leading questions and questions that are vague or contain confusing wording. For example, asking someone each of the following questions may illicit different responses even though all three questions address the same topic.

  1. “Is it really possible for a person to still believe that wearing a seat belt is not

completely necessary?”

  1. “Is wearing a seat belt necessary for the complete safety of all passengers?”
  2. “Wearing a seat belt is currently required by state law. Do you agree with this law?”

2. How would you answer each of these questions? Did the wording of the questions influence

your responses?

3. Refer to the article at the beginning of the task. What questions could the researchers have asked? Can you write two unbiased questions related to the article that researchers might have asked the subjects of the study?

Another possible source of bias in studies is in the sampling technique. Remember that a sample is a subgroup of the population. It is important that researchers use unbiased samples. In order to have an unbiased sample, the sample must be selected at random. There are many types of random samples. The most common are:

- A simple random sample, in which every possible sample of the same size has the same chance of being selected. This can be accomplished by assigning every member of the population a distinct number and then using a random number generator or table to select members of the sample.

- A systematic sample, in which every member of population is assigned a number or put in order and then members of the sample are selected at set intervals, for example every tenth member is selected for the sample.

- A stratified random sample, in which members of the population are grouped by a specific characteristic and then members from each group, or strata, are selected using a simple random sample procedure.

- A cluster sample, in which the researcher identifies pre-existing groups, or clusters, within the population and then randomly selects a set numbers of these clusters as the sample. In this case, every member of the selected cluster is a part of the sample.

There are also sampling methods that create bias in the study. Examples of these methods are convenience sampling (asking the first ten people who walk by) and voluntary responseorself-selected(asking radio listeners to call in to share responses or vote on a particular issue or asking subjects to return a survey by mail or email). Both convenience sampling and voluntary response lack the critical element of randomization.

4. Create your own example of each type of sampling method.

Simple:

Systematic:

Stratified:

Cluster:

Convenience:

Voluntary or self-selected:

EQ: What is the margin of error when sampling?

Margin of Error:

This gives a limit on how much the responses of a sample would differ from the responses of a population. When a random sample of size, n, is taken from a large population, the margin of error is approximated by this formula:

Margin of Error =

In other words, if the percent of the sample responding a certain way is p, (expressed as a decimal), then the percent of the population that would respond the same way is likely to be between:

and

Planning a Prom

You are on the prom committee and need a quick idea of how many students will be attending. You randomly select 5 students from each junior and senior homeroom (80 total) and ask whether or not they plan to attend the prom. The survey results say that 52% of students responding plan to attend.

1. What sampling method BEST describes your poll? Why?

2. If the junior and senior classes combined have 1050 students, about how many students

will attend the prom based on your survey?

3. What is the margin of error for the survey?

4. If all juniors and seniors were surveyed, what is the interval that is likely to contain the exact

percent that will attend the prom?

5. What is the interval that represents the number of students that are likely to attend the prom?

Examples:

  1. Find the margin of error for a survey that has a sample size of a. 125; b. 4000.
  1. Find the sample size required to achieve the given margin of error a. ±3%; b. ±0.8%.
  1. What do you notice about the relationship between the margin of error and the sample size?
  1. In a survey of 800 home buyers, 90% said they used a real-estate agent to research home listings. What is the margin of error? Give an interval that is likely to contain the exact percent of all home buyers who used a real estate-agent to research home listings.

5. In a recent telephone poll, a major news agency found that, including margin of error, between 46% and 54% of voters plan to the vote for the Democratic presidential candidate. What sample size did the news agency use in conducting the poll?

EQ: 1. What types of studies are carried out?

2. What sort of flaws in experiments do we need to be aware of?

Vocabulary:

  1. Experimental group: a group of individuals who undergo a procedure or treatment.
  2. Control group: a group that does not undergo the procedure or treatment.
  3. Observational study: individuals are observed and variables of interest are measured, but there is no attempt to influence the responses. The assignment of individuals is outside the control of the investigator.
  4. Experimental study: some treatment is deliberately imposed on the experimental group in order to observe their responses. The investigator assigns the individuals to the experimental or control group.

Examples:

For questions 1 and 2, determine whether the study is an experimental study or an observational study. Explain your reasoning.

  1. A scientist wants to study the effects that a nutritional supplement has on the growth of mice. The weight of each mouse is recorded daily. The control group consists of mice that do not receive the supplement. The experimental group consists of mice that receive a safe amount of the supplement.
  1. You want to study the effects that using a calculator has on the time it takes to complete a math test. You record how long it takes a student to complete the test. The control group is students that choose not to use calculators. The experimental group is students that choose to use a calculator.

For questions 3 and 4, identify any flaws in the experiment given and describe how they can be corrected.

  1. A researcher conducts an experiment to see if a new medication is effective in preventing strokes. An experimental group of lawyers suffers more strokes than a control group of professional tennis players.
  1. You conduct an experiment to see if parents make their children wear seatbelts. You have a police officer ask the question to a large group of parents.

Assignment:P. 231: 1-3 all, 5-19 odd, 20-23 all; P. 235: 1-6 all; P. 236:1-6.