Experimental Design

SHOWING CAUSE & EFFECT RELATIONSHIPS

Limitations of Experiments

• Often criticized for having little to do with actual behavior because of strict laboratory conditions
• Lab environment is intended to be a simplified reality
• Purpose is not to recreate everyday life but to test theoretical principles that explain everyday behavior
• Not always Ethical to create “real life” situations
• Natural Experiments – Study natural occurring event to observe and measure the effects of something you could not create or ethically do in a lab.
• Problem is you can’t control variables in a Natural Experiment.

​

Definitions (some review)

• Hypothesis - A testable prediction of the outcome of the experiment or research
• Variables - factors that change in ways that can be observed, measured, and verified
• Operational definition - precise description of how the variables will be measured
• Experiment – A research method in which an investigator manipulates one or more factors (independent variables) to observe the effect on some behavior or mental process (the dependent variable).
• Enable researchers to isolate the effects of one or more factors by:
• Manipulating the factors of interest
• Holding constant (controlling) other factors

Operational Definitions

• How the researcher will define and measure the key variables in the experiment.
• Reduces what you are studying to an objective number
• Similar to a “recipe,” – specify exactly how to measure and/or manipulate the variables in a study.
• Good operational definitions define procedures precisely so other researchers can replicate the study.

Experimental Group

• Effect Group
• Subjects in an experiment who are exposed to the treatment (independent variable)
• The group being studied and compared to the control group
• Also called the experimental condition

​

Control Group

• Comparison Group
• Are not exposed to the independent variable
• Results are compared to those of the experimental group
• Also called the control condition

Choosing Your Groups�minimizing confounding variables/individual differences

• Randomly Select a Random Sample—every member of the population being studied should have an equal chance of being selected for the study
• Random Assignment—every subject in the study should have an equal chance of being placed in either the experimental or control group
• Randomly select a random sample then randomly assign that sample to the experimental and control groups.

​

• Randomization helps avoid false results & bias & accounts for individual differences in people.

Experimental Variables

• Independent variable (IV)
• the controlled factor in an experiment
• hypothesized to cause an effect on another variable
• Dependent variable (DV)
• the measured facts
• hypothesized to be affected

Independent Variable

• Causes something to happen
• The variable manipulated/changed by the experimenter
• The variable which should change the dependent variable
• variable is controlled by the experimenter

​

Dependent Variable

• The experimental variable which is affected by the independent variable
• The “EFFECT variable”
• At the mercy of the IV
• The outcome of the experiment
• The variable being observed and measured

​

Knowing the Difference

• Find DV first by asking:
• “What is the researcher measuring or looking for in this study?”
• Next, find IV by asking:
• “What do the researchers hope will cause the DV in this study?”

​

• Verify with an If/Then Statement:
• If this (independent variable) THEN this happens (dependent variable).
• If my subject drinks an energy drink (Ind. Variable) THEN they should get a surge in energy (Dep. Variable)

OR

• They are testing the effect of (IV) on (DV).
• Good Way to Remember: An IV in your arm causes something to happen (DV)

Potential Problems

EXPERIMENTAL FLAWS TO LOOK OUT FOR

Confounding Variables

• Variables, other than the independent variable, which could inadvertently influence the dependent variable
• “Outside factors” that could have caused your results.
• Need to be controlled/eliminated in order to draw a true, cause-effect relationship in the experiment.
• Many confounding variables can be eliminated through random assignment.

Confounding Variables: �Environmental Differences

• Any differences in the experiment’s conditions
• between the experimental and control groups
• Differences include temperature, lighting, noise levels, distractions, etc.
• Ideally, there should be a minimum of environmental differences between the two groups.

Confounding Variables:�Expectation Effects�(Participant/Researcher Bias)

• Any changes in an experiment’s results due to the subject or researcher anticipating certain outcomes to the experiment
• Change in DV produced by subject’s expectancy that change should happen
• Researcher favoring one group over another

​

Sources of Bias

• Demand characteristics—subtle cues or signals by the researcher that communicate type of responses that is expected.
• Form of Researcher Bias
• Also helps to guard against the Clever Hans Effect
• Hawthorne Effect (participant bias) - refers to a change in behavior of the subject because they have a great deal of attention focused on them.
• Usually a spurt or elevation in performance or physical phenomenon is measured.

​

Eliminating Bias: Placebo

• A non-active substance or condition administered instead of a drug or active agent
• Given to the control group
• Reduces expectancy effects
• Ever get a boo boo and have your mom or dad to kiss it and make it better?
• Doctors may use Placebos more than you think (NBC Report on Placebo 2 min.)
• “Nocebo” – Patients when told a drug won’t work can block it from working.

​

Eliminating Bias: �Single Blind Procedure

• An experimental procedure where the research participants are ignorant (blind) to the expected outcome of the experiment

Eliminating Bias:�Double Blind Procedure

• Technique in which neither the experimenter nor participant is aware of the group to which participant is assigned

​

Experiments: �Data Analysis

Are My Results Valid & Reliable?

• Validity – Does the experiment measure and predict what it is supposed to?
• Reliable – If repeated, will we get similar results?

​

Statistically Significant

• Possibility that the differences in results between the experimental and control groups could have occurred by chance is no more than 5 percent (p = .05 or less).
• Must be at least 95% certain the differences between the groups is due to the independent variable

Experiments: Replication

Replication

• Repeating the experiment to determine if similar results are found
• If so, the research is considered reliable.
• Does Vitamin C really prevent colds?

3 Types of Experiments

Experimental Method

• Play “Water, Water Everywhere” (12:20) Segment #2 from Scientific American Frontiers: Video Collection for Introductory Psychology (2^nd edition)
• Dousing Rods to find water
• An experiment is set up to see if this psychic phenomenon is true.

Examine the Concept Answers�� Confounding Variables�� i -> c� ii -> a� iii -> b�

Correlation or Experiment?�Know the Difference