Lec 5: Research Design II

True Experimental Design

• Random selection of participants

• Maybe conducted w or w/o pretest

• Test group & control group

Quasi-experimental Design

• Participants are not randomly selected

• Involves pretest & posttest

Experimental Research

• Determine causal relationships

• Intervention

• Variable manipulation

• Controlled setting

Experimental Design

• Way to carefully plan experiments in advance to obtain valid results

• Design of Experiments

• Systematic data collection

• Focus on design itself

• Planning changes to independent variables and observe effect on dependent variables

Randomization

assigning study participants to experimental groups

Blocking

controlling other sources of variation (than treatments) in experimental results

Replication

repeating experiment more than once

True Experimental Design: Post-test Only Equal (2) Groups

R x O₂ (Treatment Group)

R x O₂ (Control Group)

• Randomization

  • each group randomly chosen & assigned to treatment or control group

• Comparison of control & treatment group

• Posttest to determine the difference between 2 groups

True Experimental Design: Post-test Only Equal (>2) Groups

R x O₂ (Treatment Group₁)

R x O₂ (Treatment Group₂)

R x O₂ (Treatment Group₃)

R x O₂ (Control Group)

• Randomization

  • each group randomly chosen & assigned to treatment or control group

• Comparison of control & treatment group

• Posttest to determine the difference between 2 groups

True Experimental Design: Pre- & Post-tests Equal (2) Groups

R O₁ x O₂ (Treatment Group)

R O₁ x O₂ (Control Group)

• Randomization

• Comparison of control & treatment groups

• Measure change, pretest to assess difference between groups prior to study taking place

Pre-test Post-tests (Control) Group Design Example

If one wants to gauge if a new way of teaching ‘Time Complexity’ is effective

• randomly assign participants to a treatment/control group

• pre-test to treatment & control group

• new teaching method (ILO) on treatment group and standard on control group, to ensure treatment method is only differing condition

• administer post-test to both groups

• assess differences between groups

Quasi-experimental Design: Pre- & Post-tests 2 Groups

O₁ X O₂ (Treatment Group)

O₁ X O₂ (Control Group)

• No randomization

  • assigned to treatment/control group for convenience

• Compare control & treatment groups

Quasi-experimental Design: Pre- & Post-tests 1 Group

O₁ X O₂ (Treatment Group)

• No randomization

• Treatment group only

• Measure change in dependent variable (outcome)