the accuracy of measurement is essential for experimental research so as to address extraneous variables

______

one can address validity in 2 ways: internal and external 

establishing cause-and effect by manipulating an independent variable and observing a dependent variable

_______

can be done in laboratory or "in the field"

internal validity

internal validity is strong if NO alternative explanations can explain the observed effects of the independent variable on the dependent variable

______

one can address the threats to internal validity by planning for and controlling extraneous factors

external validity is strong if results can be generalized which is typically more practical

_____

one can address external validity by using the best representative and random sample population available

Threats to internal validity:

history effect

life situations and events that occur during the period between a pretest and post-test that could impact dependent variable

_____

one can address this by using a control group (no treatment received) to compare with the experimental group (treatment administered)

Threats to internal validity:

maturation effect

changes in subject due to the length of time of the study; most common with children

_____

one can address this by using a control group (no treatment received) to compare with the experimental group (treatment administered)

Threats to internal validity:

testing effect

participants learn how to answer a post-test by engaging in a pretest- they become "test-wise"

_____

one can address this by using a control group that receives no pretest

Threats to internal validity:

instrumentation

inaccurate measures and unstandardized procedures; need to give consistent and accurate data and the person using the instrument must also be consistent and effective in administering the instrument

_____

one can address this threat by having well-designed measurement tools and training those who will be using the instruments

Threats to internal validity:

selection bias

differences prior to the study may account for the differences between groups; occurs when randomization is NOT utilized

____

one can address this threat by randomly assigning subjects to groups or pretesting groups to rule out pretreatment differences

Threats to internal validity:

selection maturation effect

using intact groups that already have variance in maturity within the intact group

_____

one can address this threat by pretesting and/or pre-screening intact groups based on maturity levels

Threats to internal validity:

statistical regression

selection of subjects based on high and low scores on a test; theoretically using these scores will give you the best and worst pool of subjects but their post-test scores will have a tendency to be better/worse because of the small window for imporvement (high scores) or regression (low scores); no treatment is administered to produce differing results

_____

one can address this threat by randomly selecting participants representing the full range of test scores

Threats to internal validity:

mortality/attrition

a study loses participants due to death, inability to continue, loss of contact or voluntary exit from the study; effects results because it varies # of subjects, may skew statistical data and it is problematic to replace them

_____

one can address this threat by oversampling, increasing # of subjects, introducing incentives to finish the study and collecting group demographics to compare mortality impact should that occur

Threats to internal validity:

Hawthorne effect

the attitude of subjects being studied effects results rather than implementation of treatment

______

one can address this threat by providing a control group (if one is used) with comparable treatment that has limited impact on dependent variable, or to keep participants from knowing they are being studied (which may violate consent policies) 

Threats to internal validity:

Placebo effect

expectations' of subject influences results rather than treatment

______

one can address this by using a blind study where no subject knows if they are receiving the placebo or by using a double blind study where the researchers nor the participants know who gets the placebo; lastly, as little information can be given to both groups regardless of placebo knowledge

Threats to internal validity:

Diffusion of treatment

treatment being applied to experimental group somehow gets applied involuntarily to control group

______

one can address this threat by emphasising confidentiality about treatment or to use the most similar groups with little to no interaction with each other

Threats to internal validity:

Location

differences in location of study impact results

_____

one can address this threat by using the same room or space, and if this is not possible, making the research rooms as much alike as possible

Threats to internal validity:

implementation

those overseeing research study may influence results by influencing participants via biases or inequitably

______

one can address this threat by making sure all persons involved with research implementation are equally trained, present information as a group, or have an unbiased person present the information or have a neutral observer

Threats to external validity:

selection treatment interaction

researcher tries to generalize results past what is actually appropriate

_______

one can address this by understanding the limitations of the group being studied and using randomization (again with the understanding of limitations  of the population being used to randomly select subjects)

Threats to external validity:

setting treatment interaction

the inability to duplicate studies and results in differing environmental conditions and settings

_______

one can address this threat by using participants from all possible environmental settings and conditions

Threats to external validity:

history treatment interaction

generalization is applied to past and future settings that actually may be time-sensitive

______

one can address this threat by being cognizant of the time the experiment is being administered

analysis of covariance

(ANCOVA)

statistical procedure to address difference between studied groups; this can be used in addition to randomly selecting participant from a well-defined study population, randomly assigning participant to groups and using control groups

***external validity CANNOT exist without internal validity***; ensuring internal is priority over external