Research Final: Internal and External Validity

Internal Validity

* The change in the IV caused the change in the DV
* Ability to make (implied) cause and effect statements

History

* Pre-test: Participants smoke, on average, 2 packs of cigarettes/day


* Treatment: Anti-smoking Hypnosis Program
* Post-test: Participants smoke, on average, 1 pack of cigarettes/day
* Internal Validity: Our hypnosis program caused the decrease in smoking…..
* Possible alternative explanation: The surgeon general came out with a warning between your pre-test and post-test report indicating that smoking causes cancer; and this may have caused the decrease in smoking

Instrumentation

* Pre-test: Assessing Language Skills (CELF-R)
* Treatment: Intensive treatment with FastForward Program for 6 months
* Post-test: Re-Assess Language Skills (CELF-III)
* Used different pre-test from post-test to deal with testing threat:
* Alternative Explanation: Two different tests are not equivalent… so results due to tests, not to tx

Maturation

* Pre-test: Assessing Language Skills


* Treatment: Intensive treatment with Fast-Forward Program for 6 months
* Post-test: Re-Assess Language SKills
* Internal Validity: Fast-forward treatment caused the improved language skills
* Possible alternative explanation: MATURATION-maybe the children would have had the same exact growth in language skills without the tx, you are just measuring maturation and growth

Testing

* Pre-test: Assessing Language Skills
* Treatment: Intensive treatment with Fast-Forward Program for 6 months
* Post-test: Re-Assess Language Skills
* Internal validity: Treatment caused growth in language skills
* Alternative Explanation: Taking the pretest made the children more aware, so did better 2nd time

Drop out (or mortality even works here too)

* Pre-test: Assessing Language Skills (CELF-5)
* Treatment: Intensive treatment with FastForward Program for 6 months
* Post-test: Re-Assess Language Skills (CELF-5)
* Systematic Drop-Out--- The ones who found fastforward a boring program that was too hard dropped out, influenced your results

Groups not comparable before study began

Outcomes differ not because of treatment but because of prior group differences

Placebo Effect

A beneficial health outcome resulting from a person's anticipation that an intervention will help

Hawthorne Effect

Improved performance because of the researcher's presence

Pygmalian Effect

Self-fulfilling prophecies of investigator’s expectancies

Placebo Treatment

Any treatment that has no active properties, such as a sugar pill

Single blind

The participant is blind to the purpose of the study

Double blind

* Both the data collector and the research participant are "blind" to the purpose of the study.
* They don't know what treatment the participant is getting or what the treatment is about

Control Group

The group that isn’t being tested

Ensuring Groups are Comparable beforehand---how do we this?

???

External Validity

Can the findings of the study be generalized?

* From sample to population/your client(s)
* From setting to other settings
* From these stimuli to other stimuli
* From this treatment to slight variations of treatment?
* From these measures to other measures which assess roughly the same constructs?

Generalizability from sample to population

* Random sampling
* Random assignment

Generalizability from stimuli to stimuli

* Observe real life


* Think about it

Generalizability from treatment to slightly modified versions of treatment

???

Generalizability from measure to measure

???

Generalizability beyond the lab (or from setting to setting)

???