Chapter 4 - Designing Studies

A census would be if you asked all the employees who work for a company if they liked working there or not.

A population of group would be if we wanted to know the grades of Leland High Schools math students so we would take all the individuals needed in order to get the information we want.

It is a method for collecting data. For example, if you want to know the favorite color of the students at your school, you take a sampling of the students by asking a small group of the total population (the student body).

[This is foundational for all chapters that follow]

manufacturers and advertising agencies often use interviews at shopping malls to gather info about the habits of consumers and the effectiveness of ads. a sample of mall shoppers is fast and cheap. but people contacted at shopping malls are not representative of the entire U.S. population. they are richer, for example, and more likely to be teenagers or retired. moreover, mall interviewers tend to select neat, safe-looking individuals from the stream of customers. decisions based on mall interviews may not reflect the preferences of all consumers.

The ABC network program asks a question on domestic policy and conducts call-in polls of public opinion. Any individual who wishes to give their opinion is welcome to call and answer the question.

A group of 5 students are chosen out of a hat from a class of 30 students. In this case, the population is all 30 students, and the sample is random because each student has an equal chance of being chosen.

Table B

19223 95024 05756 28713 96409 12531 42544 82853

For example you want to know what Leland students favorite subject is. You randomly select five classes and ask all students in each class which subject they prefer.

A shampoo manufacturer wants to test how if his shampoo encourages hair growth. He tests this by measuring the amount of hair 50 Asian women have, then having the 50 women wash their hair daily with the shampoo and measuring how much hair they have by the end of the program.

There is undercoverage in this experiment, because no other ethnicities besides Asian are tested. Some hair types may differ than Asians.

For example; a single individual in the sample that you conduct about kittens refuses to cooperate because he absolutely hates kittens with a fiery passion.

Biased question: Even though McDonalds is very unhealthy, would you rather eat at McDonalds or Subway?
Unbiased question: Would you rather eat at McDonalds or Subway:?

Sometimes the sampling frame is not arranged alphabetically. One year a group chose to sample from all drivers, so we asked student services for the list of students with parking tags. These are sorted by date of application.

For example you are interested in whether karaoke singers who are more animated receive more applause from the audience. You visit a restaurant that has karaoke and observe karaoke singers during one evening. You rate each karaoke singer with respect to how animated the singer was. Then, you rate the degree of applause that each singer received. You then determine if there was an association between how animated the singers were and the applause that they received from the audience. [Note: a strong association does not prove that more animated singers cause the audience to respond with more clapping.]

An educator wants to compare the effectiveness of computer software that teaches reading with that of a traditional reading curriculum. He tests the reading ability of each student in a class of fourth graders first, and then divides them randomly into two groups. One group uses a computer regularly, the other is taught with the traditional curriculum. At the end, both groups are retested to measure any increase in reading ability.

This is an experiment, because a treatment is imposed. The explanatory variable is the teaching methods: using a computer or the traditional curriculum.
The response variable is the increase in reading ability.

A researcher uses two types of treatments for weight loss (pills and an exercise regime) for a group of volunteers to find out which works more effectively.

plants are used as experimental units in order to test out the effectiveness of a type of fertilizer.

patients in a hospital receive a certain treatment in order to test out its effectiveness.

Some experiments involve more than one factor. [Can you think of an example?]

I think that temperature has something to do with the number of chirps per minute that a cricket makes. I obtain a box of crickets and place them one at a time in a temperature-controlled environment of 40, 50 and 60 degrees. The levels of this experiment are...

6 students are assigned to 3 treatment groups to discover the better studying method. Replication would be assigning 90 students to 3 treatment groups, instead.

Experiments should compare two or more treatments in order to prevent confounding the effect of a treatment with other influences, such as lurking variables.

How unlikely does a result need to be so that you are convinced that it could not happen by chance alone? 10%? 5%? 1%?

Control: use a placebo and a treated group, OR use more than one treatment.

Randomize: Use numbers in a hat, your calculator, or use a random digits table.

Replicate: do multiple trials to get better results.

For example if a doctor is giving patients either a placebo or a real treatment but neither the doctor nor the patient know whether they have the placebo or the real treatment, it is a double blind experiment.

Can you give an example of a single-blind experiment?

"Gastric freezing" is a clever treatment for ulcers in the upper intestine. the idea is that cooling the stomach will reduce its production of acid and so relieve ulcers. thus, the design was:

Gastric freezing-->Observe pain relief

a later experiment divided ulcer patients into 2 groups. one group was treated by gastric freezing as before. the other group received a placebo treatment in which the liquid in the balloon was at body temperature rather than freezing. 34% of the 82 patients in the treatment group improved, but so did 38% of the 78 patients in the placebo group. this and other properly designed experiments showed that gastric freezing was no better than a placebo.

A researcher is carrying out a study of the effectiveness of four different skin creams for the treatment of a certain skin disease. He has 80 subjects and plans to divide them into 4 treatment groups of 20 subjects each. Using a randomized blocks design, the subjects are assessed and put in blocks of 4 according to how severe their skin condition is; the 4 most severe cases are the first block, the next 4 most severe cases are the 2nd block, and so on to the 20th block. The 4 members of each block are then randomly assigned, one to each of the 4 treatment groups.

In class we will carry out an experiment with pulse rate. See if you can figure out how to turn it into a matched pairs design.

A study compares two television advertisements by showing TV programs to student subjects. The students know it's "just an experiment." The results wouldn't apply to everyday viewers and this study doesn't provide a realistic setting as the subjects know they are being tested.