internal and external validity

internal validity

is assured when the changes in the dependent variable result only from the controlled, manipulation of the independent variables

external validity

Is assured when the researcher can generalize results obtained from a sample to the population of interest

threats to internal validity

-History
-Maturation
-Attrition
-Testing
-Instrumentation
-Statistical Regression
-Selection effects
-Placebo effect or nonspecific treatment effect

History

this threat occurs when measurements of the dependent variable are recorded before and after some uncontrolled event, that is not a part of the study design

uncontrolled events occurring between the pretest and posttest might influence the information obtained at the posttest and thus confound the relationship between the independent and dependent variables

maturation

a time related difficulty when measurements are taken over long intervals of time. the effects are directly related to the element of time rather than to the events

subjects grow, develop, or change over time and the research design must control for these factors

attrition

this threat can affect the results if subjects who are less or more likely to benefit from treatment drop out in disproportional numbers

treatment

this treat occurs when the subjects responses are altered as a result of mere exposure to the evaluation tools

instrumentation

the state of the instrument being used can influence scores. may not be caliprated properly; age, sensitivity or other technical performance characteristics may influence recorded values

statisticcal regression

this threat is only possible when subjects are selected on the basis of extreme scores (very high or very low) and it can account for observed gains or losses in performance during post-test measurements

selection bias

threat: many subjects volunteering for research studies do so for ulterior motives, and these subjects will generally seek to meet their needs rather than respecting the design of the study

placebo effect or non specific treatment 

threat: the subjects expectations about benefit from the treatment may influence the outcome

threats to external validity

-Sample to Population Effect
-Description of the Experimental Treatment
-Exposure of Multitreatments
-Hawthorne Effect
-Researcher Effect
-Pretest Effects

sample to population effect

threat: this concerns the extent to which the researcher can generalize the results of a study to the defined population

description of the experimental treatmemt

threat: In reporting research, the author must describe the experimental treatment in sufficient detail so that other investigators can exactly replicate the study if they choose.

exposure of multitreatments

threat: sometimes subjects are exposed to more (X1, X2, X3,…) than one manipulated independent variable because of this circumstance the researcher cannot safely generalize his results to a situation in which variable X1 is administered alone.

Hawthorne effect

whenever subjects participate in research they receive special attention. the special attention may alter behavior enough so that subjects perform better during the experimental measurements than they otherwise would

researcher effect

subjects may react differently to the particular operator administered the experimental procedure

pretest effects

the initial test or measurement session can serve as part of the treatment effect causing different results than if the experiment was repeated without a pretest

dependent variable

is measured to determine the effects of the independent variable

independent variable

is manipulated or controlled by the researcher