Chapter 9

internal validity
exists if the dependent variable are real and not caused by extraneous factors. More likely to exist in laboratory research than field research
external validity
The ability to generalize study results to other groups and settings beyond those in the current study
threats to internal validity
the extraneous factors that allow for alternative explanations as to what caused a given effect on the dependent variable
history effect
threatens internal validity
when events occur outise of a reasrech study between the pretest and the posttest that could affect participants in such a way as to have an impact on the dependent variable (ex. stress, natural disaster, illness, news events)

How to adress concern:
use a control group and experimental group. Comparison between the two can be made at end of study

maturation effect
threatens internal validity.
Changes seen in subjects are a result of the time that elapsed since the study began and not any program effects (ex. participants become older, wiser, stronger)
How to address concern:
have a control and experimental group. Changes between the two groups would be a result of the independent variable.
test-wise
threatens internal validity.
In many types of research pretests are given to participants to determine a baseline level of the dependent variable. The baseline is compared with the posttest measurement to determine the effectiveness of a program. Being tested the participants may learn how to do better on the test the next time they take it. So differences from the baseline and posttest may be from a testing effect and not the independent variable.
How to address the concern:
Using a control group group that does not receive any pretesting
instrumentation
can threat internal validity when the measurements are not accurate or procedures and not standardized. Individuals using the equipment must be well trained and follow consistent procedures.

Written instruments must accurately measure what they are supposed to measure. Items on written instruments must be carefully designed to ask the correct question.
How to address this concern:
well-designed instruments

selection bias
threatens internal validity
Anytime individuals are selected in a nonrandom manner. It introduces the possibility that participants in experimental and control groups may have differences before the start of the study that could account for any differences found between groups during the study, instead of the differences being due to the effects of the independent variable.
How to adress concern:
recruit volunteers and then randomly assign them to groups, match participants on selected characteristics and then randomly assign them to groups, pretest groups on measures of the dependent variable to make sure there are no pretreatment differences between groups.
selection maturation effect
threatens internal validity
combines a selection bias with a maturation threat. Occurs when using intact groups that vary in their maturation level
how to address this concern:
pretesting and/or prescreening groups on maturity level
statistical regression
threatens internal validity
participants are selected on the basis of their extremely high or low scores. If they were administered the same instrument again the tendency would be those that scored extremely high would score lower, and those that scored extremely low would score higher. It would happen with no program in effect to impact scores, this is called regression to the mean.
How to adress concern:
choose a random sample so the full range of scores is represented.
morality effect
threatens internal validity.
study participants drop out. causes a problem in group comparison, when groups become smaller it’s harder to generalize findings.
How to address concern:
oversampling and using large groups sizes, using incentives to encourage participants to stay in the study
hawthorne effect
threatens internal validity
participants attitudes toward being involved in a study affect the way they behave.
how to address this concern:
rsearcher may try to provide the control gorup with some type of special treatment that is comparable to the experimental group, but wouldn’t have a direct impact on the dependent variable. Keep participants from knowing that they are taking part in a study
placebo effect
threatens internal validity
caused by participants’ expectations rather than by any provided treatment.
How to deal with concern:
Give both groups as little information as you can about the study. Give a placebo to the control group, the study then becomes a blind study. When the researcher doesn’t know who is taking the placebo and who is taking the real medicine it becomes a double blind study.
diffusion effect
treatment being applied to one group spills over or contaminates another group.
How to address concern:
have the control and experimental group be from different areas. Inform the experimental group of the confidentiality of the study and they can’t share it with people until the study is over
location effect
threatens internal validity
differences in the locations where interventions take place. Different atmospheres or study locations can yield different results (ex., one counseling room is more up to date than another one)
How to address the concern
make the location the same for all participants, if not possible the researcher should do everything possible to minimize location differences that could impact the dependent variable.
implementation effect
the possibility that the individuals responsible for the experimental group may inadvertently introduce inequality or bias into the study. (1)One way is there are multiple people providing the treatment program or intervention. (2) Another way this type of bias can occur is if an individual implementing the intervention inadvertently favors one group over another.
How to address this concern:
(1)make sure all persons responsible for implementing the program are equally trained and competent, and following a standardized protocol for implementation. Or all individuals involved in implementing instruction speak to the group. (2) Have someone other than the program developer present the program. The person presenting should be neutral. Or, a neutral observer should wath the presenter with specific instructions to look for ways in which the two groups are being treated differently.
selection treatment interaction
threatens external validity
concerns the ability of a researcher to generalize the results of a study beyond the groups involved in the study. When using intact groups the researchers have no ways of knowing if they are truly representative of a larger population. If a study uses a random sample of participants, the results can legitimately be generalized to the group from which the random sample was selected, or the study population. However, the results can’t be generalized to whole population.
How to adress this concern
researchers need to give equal chances for individuals in the population they want to generalize for to be chosen.
setting treatment interaction
threat to external validity
extent to which the environmental conditions or setting under which an experimental study was conducted can be duplicated in other settings.
how to address the concern:
the environment and setting needs to be documented and needs to be able to be reproduced fairly easy.
history treatment interaction
threatens external validity
researcher tries to generalize findings to past and future situations. Some experiments are time sensitive and may not produce similar results if conducted earlier or later
three important strategies that can be employed to control for most threats to inter validity
1. randomly select participants from a well defined study population
2. randomly assign selected participants to groups
3. include non treatment control groups in the research design.
2 important strategies that can be employed to control for most threats to external validity
1. carefully consider what groups one can legitimately generalize to, and do not generalize to other outside groups.
2. always duplicate setting and historical factors as nearly as possible when replicating a program with a a different population.