If you have any feedback for the author, E-mail me
REQUIRED TEXT: Experimental Methodology eighth edition by Larry B. Christensen, Allyn and Bacon, 2001 ISBN #0-205-30832-5 ------------------------------------------------------------ Chapter 7 Internal Validity: Independent variable produced the observed effect of dependent variable. Is there another cause of DV? Has an extraneous variable confounded experiment? Does it systematically vary with the IV? If not, ok. e.g. Does tutoring improve grades? If one group is more intelligent =>confounding. E.g. Does the teaching style of lecturing or discussion produce more profound learning? In order to control for the personality of teacher, compare an 8am section with a noon section. Must eliminate differential effects of possible extraneous variables across various levels of IV. Must control: History: Specific events occurring between the first and second measurement of DV. e.g. pre-post but a short period is ok, but an ABA design could be substituted for an AB design. Maturation: events that occur in the passage of time, such as age, learning fatigue, boredom, hunger, etc. e.g. head start program evaluation cannot just use pre-post testing, but also control participants not in a program and controls in another kind of program. Instrumentation: observations that change during the process of measurement, especially human observers become bored, fatigued and change their criteria while gaining skill at observing. :. need several trained observers or raters. Statistical Regression: If participants are chosen based on their extreme scores, these characteristics will change between pre & post testing toward the average even if there is no treatment. Selection: Ideally participants should be both randomly chosen from a population and randomly assigned to treatment (IV) groups. e.g. Testing the adequacy of a therapy on two groups of different diagnose. If one group improves faster, cannot assume that this therapy is best for that syndrome since personal characteristics t hat lead a person to be diagnosed with one disorder as opposed to another may lend themselves to the therapy. Mortality: differential participant loss Participant Effects to be Controlled: From the participant's point of view may perform the task requested but in a manner to make a positive self presentation (respond to the task in such a way that they appear most positive as defined by themselves). Same participant may respond differently to each condition depending on their perceptions of the meaning of the treatments. e.g. high scores = obsessive-compulsive personality. Different participants can perceive the personal meaning of the treatments differently. e.g. males could perceive it to be more 'manly' to endure more intense, aversive stimuli than females. Experimenter Effects to be Controlled: convey information, e.g. Clever Hans and ESP p185 Biosocial attributes: e.g. gender, race, age, religion, etc. Psychosocial attributes: e.g. friendly, authoritarian, dominant, empathy, etc. Situational factors: kinds of previous contact, experience of researcher, development of empathy. Expectancies: mistakenly record data in agreement with hypotheses. Need 'Double Blind' designs interpret data to agree with hypotheses. Nonverbal biases: Animal handling, e.g. Guinevere facial or postural signals, e.g. nods, smiles Sequencing effect 'carry-over' These biases can have a large effect. ------------------------------------------------------------ My office is in room 116C of HGH (826-4571) and I invite you to consult with me on any matter during my office hours or any time that is agreeable to you and me.Go back to the beginning
Copyright © 2002, Dr. John M. Morgan, All rights
reserved -
This page last edited Nov 25, 2002
If you have any feedback for the author, E-mail
me