Neuro 2001 Lecture 10

Front 1

Experimental Research Strategy

Back 1

Establish cause and effect relationship between two variables. Simplest form - changing one variable causes change in second variable. Four basic elements: manipulation, measurement, comparison, control. Two critical components: Independent vs dependent, Experimental vs control.

Front 2

Manipulation

Back 2

Manipulate the independent variable. This creates two or more conditions. Helps determine which is the cause and which is the effect.

Front 3

Measurement

Back 3

Measure the dependent variable, obtain scores for each treatment condition

Front 4

Comparison

Back 4

Scores on one condition compared with another condition.

Front 5

Control

Back 5

Control all other variables so they do not influence the two variables examined. Eliminate confounding variables. Control obscuring variables, hold variables constant. Matching and randomization.

Front 6

Independent Variable

Back 6

This is what you manipulate, Cause of your effect. At least 2 levels.

Front 7

Dependent Variable

Back 7

What you measured. The effect. Measured at each level of the independent variable.

Front 8

Experimental group

Back 8

the group exposed to the manipulation

Front 9

Control group

Back 9

The group not exposed to the manipulation. Used for comparison purposes.

Front 10

No treatment control group

Back 10

Subjects do not receive the treatment being evaluated.

Front 11

Placebo control group

Back 11

Participants get a placebo instead of actual treatment. Typically in clinical studies.

Front 12

Between subjects design

Back 12

Random assignment into groups. 2 or more treatment groups. Each group gets a different treatment. Independent groups.

Front 13

Within subjects design

Back 13

All subjects participate in both groups. only 1 treatment group. Subjects get both treatment levels. Repeated measures design.

Front 14

Between subjects design advantages

Back 14

Each measured score is independent of the other scores. Only exposed to treatment one time, measured one time. Results not influenced by other factors. Fatigue from multiple testing. Practice from exposure of other treatments.

Front 15

Between subjects design disadvantages

Back 15

Need more individuals. Large differences between individuals in each group will obscure differences between groups.

Front 16

Pretest - postest design

Back 16

Before manipulating independent variable, check if the two groups are equivalent. Goal is to compare the treatment vs control group. Pros: test equivalence of groups. Cons: time consuming, sensitize participants to procedure.

Front 17

Matched pairs design

Back 17

During random assignment to groups. Each condition uses different participants but they are matched in terms of important characteristics. Pros: minimize the effect of individual differences. Cons: Attrition affects both groups. Difficult to match people exactly.

Front 18

variability

Back 18

Variability refers to the distribution of scores. Effect of variability in between subjects design. Large differences between individuals produces large variance. Must take into account the absolute size of the difference in relation to the variance of the data points.

Front 19

Variance

Back 19

Statistical value for the difference of each value from the mean.

Front 20

Variance between subject design

Back 20

Small variance - All scores are similar to one another. Randomly assign a sample for control and sample for treatment. Difficult to see the treatment effect with large variance. large differences between individuals in each group will obscure differences between groups.

Front 21

Mitigating variance within treatments

Back 21

Standardize procedures and treatment setting: Keep all procedures the same. Treat each mouse the same.

Sample size: Statistical feature. Using a very large sample can overcome negative effects of high variance.

Group equivalency: Start by creating equal groups. Treat each group equally. Human - random assignment with matching. Animals: restrict range.

Front 22

Within subject design advantages

Back 22

Fewer participants are needed. Especially advantageous is participants/animals are rare or costs are high. This repeated measure design allows you to reuse the same individual. eliminate variability due to individual differences. Individuals are no longer confounding or obscuring variables.

Front 23

Within subject design disadvantages

Back 23

Participant attrition: when it requires extensive time and effort, participants drop out of study. Most problematic when studying changes over time, or multiple sessions. (solution: increase sample size). Cannot apply this design to all studies. Time related effects: history (environment changes over time and affects the results), Maturation (any physiological or psychological change of individuals when testing occurs over a long period of time). Instrumentation (Change to the measuring instrument lead to significant differences at testing independent of the treatment). Order (ordinal position of the manipulation can produce practice or fatigue carry over effects). Sequence (interaction between each condition in the experiment can reduce sensitivity of response).

Front 24

Counterbalancing

Back 24

Split the participants into two groups. advantages of testing the same individuals, need fewer research subjects, reduce within-subject response variability. Minimize order effects. Very routine practice for within group design. Goal: use every possible order of treatments with an equal number of individuals participating in each sequence. Purpose: Balance the order effect equally among the two treatment groups. Prevent order effect to accumulate in any one group.

Front 25

Complete counterbalancing

Back 25

Any series of treatment conditions may create a unique order effect. To completely balance all of these effects, research design must use every possible sequence. easy with 2 conditions, gets ridiculous with 3+ conditions.

Front 26

Partial Counterbalancing

Back 26

Does not use every possible sequence. Each treatment appears in one of the sequence positions. Latin square.

Front 27

Latin square

Back 27

Each treatment appears once in each ordinal position. Each sequence appears once.

Front 28

Placebo

Back 28

People experiencing improvement in their condition even though they were assigned a dummy pill without any active ingredients.