Thursday, April 19, 2018

ResearchMethods: Experiments: How to Design an Experiment (W15-P2) Sp18

If you wanted to do an experiment, there are three general ways for doing it.

First, a quick question: Based on name alone, which do you think is the worst form of experimental design?  Why?

Now let's look at each.

The diagrams in these slides explain how to do the different types of experimental designs.
Read these slides from left to right, starting with the random assignment of subjects into the treatment and control groups.  For each design, what are the steps after the random assignment?  That is, how do you do these types of experiments?

So, Post-test ONLY Control is a good design, but there is a problem.  Are we sure that the IV caused a change in the DV?  How would a pre-test help?

Again, this is a good experimental design.  However, there is a potential problem with this too.  Before the pre-test helped.  Now, the pre-test could be a problem.  How's it a problem?

How does the Solomon Four Group design help control for the possible influence of the pre-test?

The above designs are sub-types of full experiments with one independent variable.

However, what if you wanted to have more than on IV in your study?  How would you design an experiment for that?

Now, let's wrap up with a look at the "worst" designs.

If these are the "worst" designs why are we talking about them?  When would you use them?  Better than nothin'?

Which is better a quasi-experimental design or a pre-experimental design?  Why?  What standard are you using to judge whether one is "better" than another?

Speaking of experiments, catch the film reference in the last slide?

Share this post with others. See the Twitter, Facebook and other buttons below.
Please follow, add, friend or subscribe to help support this blog.
See more about me at my web site

Wednesday, April 18, 2018

MyFavMusic: Just listened to the "Country Trash" by Johnny Cash on the "American III: Solitary Man" album. Added to my "FOLK FAVS - 1000 Recordings" playlist on Spotify

Fav track from album: Country Trash
By Johnny Cash
From the album American III: Solitary Man

Added to FOLK FAVS - 1000 Recordings playlist by William Hart on April 18, 2018 at 05:58PM

See info on 1000 Recordings

Listen on Spotify

My musical interests on Tumblr

Tuesday, April 17, 2018

ResearchMethods: Experiments: What is an Experiment? (W15-P1) Sp18

Play the first minute or so.

When you think of a person doing science, doing research, doing an experiment, what images pop into your mind?

Have film and television shaped your view of science and of experiments?
Hey, I feel a research question coming on.  We could research that.

If it is not the Frankenstein movies that have shaped your image of scientists doing experiments, what movies or television shows have?

An experiment is a research methodology for determining the causal effects of one or more independent variables on a dependent variable, while controlling for all intervening variables.

Experiments are only one of the possible research methodologies you could use to answer your research questions.  Surveys and interviews, for example, are other research methodologies.  There are benefits and drawbacks when considering which research method to use to answer a research question.  What are some benefits that experiments have, but surveys do not?  What is it that experiments can do, that surveys cannot?

What is meant by causal relationship or causal effect?
Or put another way, to be able to say one thing caused another, what would need to be true?

  1. Independent variable comes before the dependent variable.
  2. Independent and dependent variables are meaningfully related.
  3. Changes in DV must be the result of changes in IV (and not anything else)
Watch the clip below.
Correlation vs. Causality: Freakonomics Movie

If you are interested, check out "Correlation and Causality (Khan Academy - YouTube),"
Also, if interested, check out, Correlation or Causation? (Some headlines) or Correlation vs. Causation (NYT)

What is experimental control?
The ability to rule out alternative explanations for the results, controlling for all intervening variables.
You want the IV to be the only thing causing a change in the DV.

In real estate it is location, location, location.
In experiments, it is control, control, control.

How can experimental control be enhanced?

Compare experimental control in the lab vs. field study.
  • In lab, can more easily manipulate IV.
  • In lab, can randomly assign participants TR and C groups
  • In lab, can control extraneous variables.
  • In field, max external validity

Share this post with others. See the Twitter, Facebook and other buttons below.
Please follow, add, friend or subscribe to help support this blog.
See more about me at my web site