Help EDR610 : The Class : Families : Experimental : Lesson2-3-1

 Experimental Research Family

Experimental Research Family: As we'll see shortly, there is also a corresponding 'design family' (to these research questions) called 'experimental.'

For the 'experimental' family we are looking to make 'somewhat stronger & definitive' statements than for the preceding two families. Remember that:

  1. for descriptive: identifying 'what is/what are;'
  2. for correlational: identifying 'what goes with what/what predicts what.'
For 'experimental,' we wish to go a step further and identify causes and effects of things!!!

Examples: (with keywords highlighted)

  1. This study is to determine the effect of a hands-on science teaching program designed for gifted and talented sixth-grade students upon their mastery of science concepts.
  2. What is the effect of an intensive aversion-therapy group counseling intervention program conducted in a hospital setting on reducing incidence of bulimia in teenaged girls?
  3. This study is to identify the effects of peer coaching on rated teaching effectiveness for first-year public school teachers.

    As you can imagine, 'strong' statements like the above require a lot of confidence that "it (cause/effect) is what I think it is -- and isn't due to something else!" That's where 'experimental designs' (which we'll see are the designs of choice for experimental-type questions/problems) come in.

    We'll see, in particular, that 'experimental designs' will be characterized by tight control!

    For instance, in the case of No. #1, above, we want to be "reasonably confident" that it's the 'hands-on teaching method' (cause) that is the "true reason" for the (effect of) the improved or higher science achievement test scores! And not something else!!!

    Such other 'hidden causes' (other than the ones we thought it would be) could include things like:

    1. Prior exposure on the part of some students to science enrichment activities after school (e.g., belonging to the astronomy club);
    2. Teacher training and experience: suppose, for instance, that the more experienced teachers also happened to be assigned in the study to teach the enriched hands-on science classes, while the less experienced did the 'baseline comparison,' or the traditional science instruction. Now, how do we know it was the teaching method itself, or if it was also 'contaminated' by 'teacher experience' (e.g., the hands-on method also got an "accidental boost" in the study from the fact that the more experienced teachers taught it)!
    3. Even something like a different mix of genders of students in the 2 classes -- suppose there happen to be, say, 30% more boys in the hands-on class than in the traditional lecture method class. If, as many suspect is the case, girls on average are 'socialized away from' and/or given less exposure to science and math as potentially fun and 'doable' activities -- could, again, the 'hands-on' class have been accidentally contaminated by the fact that there were also more boys in that group -- and boys would on average have done better in any case due to prior conditioning, socialization, expectations, etc.?!

      Such 'hidden contaminants' (which we'll learn are also known as 'threats to internal validity' of a given study design) could very well 'mess up' your logic ('this and only this caused that') unless you carefully control for them in your study!

      Therefore, in experimental designs, as we'll see, you need to do a good, thorough 'prior brainstorming' of such potential contaminants and carefully balance on, or control for, them!

      This could mean, for example:

      1. Equalizing teacher experience in both groups (e.g., have the same teacher do the lesson in both traditional and hands-on or carefully match to see that the two different teachers have equivalent prior background and experience);
      2. Doing the same type of balancing on gender (either keeping it all to one gender for the study; or seeing that there is about an equal mix of boys and girls in both the traditional and hands-on class);
      3. Doing a similar balancing if possible on determining in advance what other science enrichment activities the students participate in and making sure it's about in the same proportion or "mix" in both groups.

        Do you see how such careful prior control or balancing helps "make your case" in the sense of 'believable cause-effect statements?"

        If you do such balancing/control on as many outside 'contaminants' as possible, AND the 'hands-on group' STILL does better than the traditional lecture group, THEN you can 'safely rule out' these contaminants as the cause!

        "It can't be gender: equal number of boys & girls in both groups."

        "And it couldn't have been teacher experience either; the same teacher did both types of instruction."

        And so on and so forth -- until 'the only cause left' in the logic is the one you presumed! e.g., the (hands-on) method of instruction!

        Such is the 'stuff' of experimental questions/problems (cause/effect) and related designs (tight control and/or balancing to rule out as many other possible or competing causes as possible)!

True Experimental Designs and Their Meaning

Once you have completed this assignment, you should:

Go on to Write an Experimental Research Problem Statement
Go back to Experimental Research Family

E-mail M. Dereshiwsky at statcatmd@aol.com
Call M. Dereshiwsky at (520) 523-1892


Copyright 1998 Northern Arizona University