Internal External Validity of Experimental Research Essays

Internal External Validity of Experimental Research Essays Internal External Validity of Experimental Research Paper Internal External Validity of Experimental Research Paper History Effect When events occur between the pretest post-test that could affect participants in such a way as to impact the dependent variable; e.g. stress, natural disasters, news events, illness, etc. Strategy: Use a control groups that is unable to be exposed to such an unplanned for event; i.e. track those who dont receive any treatment/alteration of independent variable. Maturation Effect EWhen changes are seen in subjects b/c of the time that has elapsed since the study began may not necessarily be due to program effects; i.e. over time participants may have become wiser, stronger, more experienced therefore not affected by indep. variable; most commonly seen in children. Strategy: Establish a control group. This would account for any developmental changes as both groups would experience such, so further changes could most likely be due to indep. variable. Testing Effect By being pretested, participants may learn how to do better on the test or post test, whenever they take the test again, that is supposed to measure the indep. variable. Strategy: Use a control group that does not receive pre-testing, but only administration of the indep. variable, and post-test, then compare the results. Instrumentation Effect When the instrument is inaccurate or procedures are not standardized the correlation or causation of a phenomena can be misrepresented. Strategy: Use well-designed instruments that have been pilot tested and reviewed by advisors, colleagues, and other researchers who have performed similar studies. Selection Bias When individuals are selected in a non random manner and participants separated into different groups w/in the experiment (experimental vs. control) they may have difference before even starting the study, instead of differences being dues to effects of the indep. variable. Four Strategies: 1. Dont use intact groups (such as Dr. Rosss freshman psychology class at XXX University) 2. Recruit volunteersthen randomly assign to groups, rather than letting volunteers self-select. 3. Matching subjects on selected characteristics.. then randomly assigning them to groups 5. Pretesting groups on measures of the dependent variable to make sure there are no pretreatment differences between groups Selection Maturation Effect Combination of selection bias w/ maturation; When using intact groups that vary in their maturation level. Strategy: Pretesting or Prescreening groups to rule out differences in maturity levels with each intact group Statistical Regression When participants are selected on the basis of their extremely high or low scores Strategy: Use a random sample representing the full range of the dep. variable being measured rather than the entire intact group and placing them into high low scores Statistical Regression When participants are selected on the basis of their extremely high or low scores Strategy: Use a random sample representing the full range of the dep. variable being measured rather than the entire intact group and placing them into high low scores Mortality/Attrition Effect Losing participants over the course of the study dues to death, illness, or relocation, or drop outs (least motivated individuals, etc). Strategy: 1. Oversampling, and large N#s is one way to overcome small group size. 2. Use incentives to encourage participant to stay in the study. 3. Obtaining good demographic info about study participants at the beginning and then determining how much group makeup has changed at the end will help account for mortality Hawthorne Effect When participants attitude(s) toward being involved in the study affect the way they behave; i.e. when a study group receives special attention over the control group, any noted differences in dep. variable may simply be a result of the attention given, not the effect of the indep. variable. Strategy: Provide the control group with some type of special treatment that is comparable to the experimental group but wold not have a direct impact on the dep. variable. 2. Keep participants from knowing that they are taking part in a study or being observed (however, due to IRB, this is difficult nowadays) Placebo Effect When participant expectations of what the indep. variable will do, affect the dep. variable. Strategies: 1. Use a control group (e.g. blind or double-blind study) 2. Give participants the same info. whether or not they are in the control or experimental group (give them as little info as possible to reduce any expectations, therefore similar small expectations any differences between groups are more likely from the RX effects than placebo effects. Diffusion of Treatment When the Rx being applied to one group pills over or contaminates another group Strategy: Use a separate intact group for the control group that is similar to but away from/unable to be influenced by the experimental group Location Effect When there are differences in the locations where interventions take place Strategy: Make the locations equitable / same for all participants Implementation Effect When the individual(s) responsible for implementing the indep. variable inadvertently introduce inequality or bias into the study. (e.g. researcher favors one group or another) Strategy: 1. Ensure the knowledge levels, understanding of the program personality traits, and presentation skills are equitable. (i.e. make sure all research assistants are equally trained and competent and follow a standardize protocol for implementation. 2. Have all researchers involved present to/interact with all groups. 3. Have someone other than the program developer present the program. (e.g. a. the person presenting should not have a vested interest in one group or the other; b. have a neutral observer watch the presenter with specific instructions to look for ways in which the two groups are being treated differently) Selection Treatment Interaction When the ability of a researcher to generalize the results of the study beyond the groups involved is compromised in some manner. Strategy; Use a random sample, no intact groups. Train the same presenters to employ the study across multiple groups involving a diverse study population. (external validity) Setting Treatment Interaction The extent to which the environmental conditions or setting under which an experimental study is conducted can be duplicated in other settings. Strategy: Attempt to use similar or relatively universal environmental settings. History Treatment Interaction When the researcher tries to generalize findings to past and future situations. ( most studies are time-sensitve until repeated. Strategy: Make participation in the study as convenient as possible.