Home Page

Previous Chapter 12 Experimental Research: Designs, Part 1

p. 401: #1(Describe a procedure to reduce initial group differences that occur due to nonrandom assignment of subjects to experimental and control treatments

Analyses of Covariance (ANCOVA): statistically reduces the effect of initial group differences by making compensating adjustments to the posttest means of two groups. (p 403)

#2(Describe methods and statistical procedures in the static-group comparison and nonequivalent control-group designs)

Static Group Comparison: research participants are NOT randomly assigned and a posttest, but NO pretest is administered. Same steps as posttest only design except no random assignment.

1) administer the treatment to experimental group only

2) administer the posttest to both groups

Non-equivalent Control-group: participants are not randomly assigned and both groups take a pretest and posttest. Same as the pretest-posttest experimental control group design except for the random assignment.

1) administer the pretest

2) administer the treatment

3) administer the posttest to both groups

#3(Describe the major potential threats to the internal validity of quasi-experimental designs)

3. Describe the major potential threats to internal validity of quasi-experimental designs.

a. posttest differences between groups can be attributed to characteristics of the groups other than the experimental condition to which they were assigned. (p. 402)

b. Differences on the posttest may be due to pre-existing conditions not the treatment effect. (p. 403)

#6(Describe the methods, purposes, and features of single-case experimental designs)

Involves intense study of one participant or more than one treated as a group. Well suited for behavior modification studies.

Single case designs use several procedures to achieve experimental control as conceptualized within the quantitative research tradition: checks on the reliability of the experimenter’s observations of the research participants behavior, frequent observations of the behavior targeted for change, description of the treatment in sufficient detail to permit replication, and replication of the treatment effects within the experiment. (p. 416)

Most single case designs are rigorous and time-consuming, and they may involve as much data collection as a design involving experimental and control groups. (p. 416)

Factors that will help aid internal validity:

∑ Reliable observation

∑ Repeated measurement

∑ Description of experimental conditions

∑ Baseline and treatment stability

∑ Length of baseline and treatment phases

#8(State several threats to the internal validity and external validity of single-case experiments)

Internal Validity Threats (p. 418-420)

∑ Differential selection – can’t use random sample since n=1

∑ Instrumentation – use reliable observations

∑ Testing – use more frequent measures to collect data

External Validity Threats (p. 418-420)

∑ Explicit description of the treatment – provide precise description of each experimental condition

∑ Multiple treatment interference – establish good baseline

∑ Experimenter effect

∑ Interaction of time of measurement and treatment effects – should be approximately the same length of time and number of measurements in each phase.

#9(Describe problems in using gain scores to measure change, and state two statistical techniques for solving these problems)

a. ceiling effect – occurs when the range of difficulty of the test items is limited and therefore scores at the higher end of the possible score continuum are artificially restricted.

b. Regression towards the mean – participants who score either very high or very low on a measure to score nearer the mean when the measure is re-administered.

c. Assumptions of equal intervals – use of gain scores assumes equal intervals at all points of the test, yet this assumption is almost never valid.

d. Different types of ability – with the exception of factorially pure tests, a given score on a test may reflect different types and levels of ability for different students.

e. Low reliability – not very reliable.

Statistical techniques for solving these problems (p. 428-429)

∑ Part correlation

∑ Multiple Regression

∑ Analysis of Covariance and t-Tests

∑ Analysis of Variance for Repeated Measures

Home Page

Next Chapter 14 Case Study Research

Planning Page for this Wiki Check here to see who is doing what!