Chapter 10 complex designs 2013

COMPLEX EXPERIMENTAL

DESIGNSChapter 10

Chapter 10 Complex Designs Sisson 2013

2

PRACTICE on IV and DV

Example 1 A researcher is studying the effect of sleep on

aggression, thinking that less sleep will lead to more aggression. She has some people sleep 6 hours per night, some people sleep 3 hours per night and some people sleep as much as they want. She then monitors aggressive behavior during basketball games among participants.

IV: Levels: DV:


3

Example 2 A researcher is curious to find out what effect classical

music has on people’s level of relaxation (as measured by heart rate). He suspects that listening to classical music will make people feel more calm and relaxed. He lets one group listen to classical music for one hour. He lets another group sit in a quiet room for one hour (i.e they hear no music). After one hour, he monitors the heart rate of each participant to measure their level of relaxation.

IV: Levels: DV:


4

Example 3 A researcher conducts an experiment to assess the effects

of alcohol on people's sense of balance. He divides his subjects into three groups: in one group the participants drink one ounce of alcohol, in another they drink two ounces of alcohol and in a third group the participants drink soda. He then watches as each participant tries to walk on a straight line from one corner of the room to the next and notes how many times they stumble outside the line.

IV: Levels: DV:


5

Where we have been…

Simple design: One independent variable at 2 levels. One dependent variable.

E.g. effect of sugar on activity level Effect of television on concentration level Effect of bad smells on mood Effect of temperature on test

performance Effect of eating rotten melon or gourd on

digestion comfort.


6

Increasing the number of levels

We may want more defined amounts than No sugar vs. 3 cups of sugar50◦ temp room vs. 90◦ temp room1 slice of bacon vs. 1 pound baconE.g.


7

Why?

Remember studies on emotional arousal and performance on a task? Draw the shape.

Is it Monotonic or nonmonotonic? By increasing the number of levels, we can

find out more detailed information. E.g. Maybe some caffeine is helpful for mood,

but too much is harmful. Give an example of something that may have

a different effect based on the level.


8

Remember

Increasing the levels does not mean we are adding another independent variable.

But what if we want to add another variable?

Isn’t that more typical of LIFE? We want to look at temperature of room

(50◦ or 90◦) on test taking AS WELL AS test difficulty (easy, hard).

Test difficulty is QUALITATIVELY different from temperature! Therefore, new IV.


9

Factorial Designs (increasing # of independent variables)

A factorial design has more than one independent variable: • type of music, volume of music• Amount of calories, time of consumption• Tone of compliments, number of compliments• E.g.

All levels of IV are combined with all levels of other IV.


10

Simplest Design We want to look at temperature of room

(50◦ or 90◦) on test taking AS WELL AS test difficulty (easy, hard).

(IV - B) Room Temperature

(IV - B) Test Difficulty

(Level) 50 degrees (Level) 90 degrees

(Level) Hard Test Hard Test in 50 degrees

Hard Test in 90 degrees

(Level) Easy Test Easy Test in 50 degrees

Easy Test in 90 degrees


11

You try

It Depends….


12

Analysis of Variance (ANOVA)(More terms!)

We are looking at how scores vary between the different groups

"way" refers to the number of independent variables in an analysis.

a "one-way" ANOVA refers to an ANOVA with one independent variable,

a two-way ANOVA would be used to analyze an experiment with two independent variables.

Mathematically, we can analyze data with as many independent variables as we want. BUT:

Very complex to interpret. Rare beyond a three-way ANOVA.


13

CONDITIONS.. Conditions!

2 X 2 design has ____ conditions 2 X 3 design has ____ conditions

3 X 3 design has ____ conditions (sensing a pattern?)

B1 B2 B3

A1 A1B1 A1B2 A1B3

A2 A2B1 A2B2 A2B3

B1 B2 B3

A1 A1B1 A1B2 A1B3

A2 A2B1 A2B2 A2B3

A3 A3B1 A3B2 A3B3


14

We can do as many conditions as we want, however, it gets very complicated (and need lots of subjects). So usually it is kept at 3.


15

More terms

Participant variables (PV) are personal attributes (IQ, age, ethnicity, ________).

Must have at least 2 levels. IV X PV (2 by 2) Caffeine intake: 0 cups 8 cups Sex: Male Female Still 4 combinations. Same idea as the IV and IV


16

How to read the designs

Other examples of factorial designs: 3 x 2 2 x 3 x 2 2 x 2 x 2 The number of digits refers to the number of

independent variables (IVs). 3 x 2 refers to 2 IVs 2 x 3 x 2 refers to 3 IVs 2 x 2 x 2 refers to ___ IVs The digit refers to the number of levels of that IV. 3 x 2 refers to 3 levels of the first IV and 2 levels

of the second IV. 2 x 3 x 2 refers to 2 levels of the first IV,

3 levels of the second IV, and 2 levels of the third IV. 2 x 2 x 2 refers to____________


17

Interpretation of effects

Statistically: two-way ANOVA has three basic types of effects that are tested:

main effect for independent variable A ( temperature)

main effect for independent variable B (test difficulty)

effect for the interaction of A and B 2 IV = ___ main effects 3 IV = ___ main effects Then have Interaction effects


18

Main Effects

Room Temperature IV A

Test DifficultyIV B

50 degrees 90 degrees Marginal means(Test difficulty)

hard 60 (Hard/50◦)

60 (Hard/90◦)

60

easy 80 (Easy/ 50◦)

80 (Easy/ 90◦)

80

Marginal means(room temperature)

70 70

Main effects only significant for IV-B. Higher scores on easy test, regardless of temp.

Look at mean differences at each level of the IV

Look at each main IV. Compare Easy/Hard Compare 50/90


19

Difference in means only with difficulty

Remember: X axis is IV, Y axis is DV


20

Main effects with interaction effects. “It depends”.

Room Temperature (IV-A)

Test Difficulty (IV-B)

50 degrees 90 degrees Marginal means(Test difficulty)

hard 60 (Hard/50◦)

70 (Hard/90◦)

65

easy 70 (Easy/ 50◦)

80 (Easy/ 90◦)

75

Marginal means(room temperature)

65 75


21

Interaction effects


22

How do you assign people to groups?

SAME people in each group (within-subjects design, aka repeated measures design).

DIFFERENT people in each group (between-subjects design, aka independent groups design)

Some groups have same people, others do not (mixed design).


23

Within- subjects design

Repeated measure design (within-subjects design) Same participants participate in all conditions 2 X 2 (1 group, 4 times) Need fewer people! YIPEE

Strengths• Increases POWER because there are more subjects

(less chance for error). E.g. 80 subjects: Use them all in each one, vs. dividing into 4 groups of 20)

• The conditions are the same for each group (IQ, age, food intake, etc.)

Weakness• Potential Carry over effects • 2 basic ones: practice and fatigue


24

Between -subjects

2) Independent group design Different participants for each condition. 2 X 2 (4 groups of people)

Strengths:No carry over effectsWeaknesses:• Need more people for same Power• Individual differences between participants (what if

one group just happens to have had more sleep, better at memorizing, more endurance, etc.)

• How do we combat this? Randomization• Environmental factors: e.g. test one in morning, one

in afternoon; one class has a better teacher, etc.


25

Or…combo platter

Mixed factorial design Some advantages and disadvantages of

other designs (e.g. need more participants but not as

many…) More complicated! Nuff said!


26

END!


27

Type of Question

Independant Variable A

Questioner typeIndependent Variable B

Unbiased Misleading Overall means (main effect of B

Knowledgeable

Unbiased/Knowledgeable 13.0

Misleading/ knowledgeable41.0

27.0

Naive Unbiased/naïve13.0

Misleading/naïve18.0

15.5

Overall meansMain effect of A

13.0 29.5


28

Simplest Design 2 X 2 (2 indep variables, each with 2 levels) Smith & Ellsworth (1987): What is the effect of

type of question and knowledge of questioner on recall of memory?

Type of Question

Independant Variable 2

Questioner typeIndependent Variable B

Unbiased Misleading

Knowledgeable Unbiased/Knowledgeable

Misleading/ knowledgeable

Naive Unbiased/naïve Misleading/naive

Entertainment & Humor

Chapter 10 complex designs 2013