Class Project

1
Class Project

• Emotion vs. No Emotion
• Auditory vs. Visual
• Picture vs. Words
• Persuasion True vs. False Memories

2
Class Project Emotion vs. No Emotion
3
Class Project Auditory vs. Visual
4
Class Project Words vs. Pictures
5
Assignment 4

• Find 6 empirical journal articles that are
  relevant to our study of false memory
• One will be the Roediger McDermott (1995) study
• You need a copy of each article (which you will
  submit with your assignment)
• No foreign languages or dissertations
• Only one article may be a review article

6
Assignment 4

• For each article, write a 2-3 page summary
• Like a very detailed journal entry
• Authors
• Question
• Method (including the design)
• Results
• Implications
• Include any problems or concerns you have about
  the study, and any new questions that arise as a
  result of your reading
• Think especially about the implications for our
  understanding of how memory works, and less about
  real world implications

7
Experimental Design

• Avoiding the Pitfalls

8
Between-Participants

• The participants in each group are different
• Advantages
• Minimizes order effects
• Minimizes carryover effects
• Minimizes demand characteristics

9
Between-Participants Advantages

• Minimizes order effects
• Practice, fatigue
• Minimizes carryover effects
• Participants carry something from one condition
  to another (learning, fatigue, sensitization,
  etc.)
• Minimizes demand characteristics
• Information the participant infers about the
  study
• They know how the instructions vary from one
  condition to another and thus may determine the
  purpose of the investigation

10
Between-Participants Weaknesses

• Requires more participants
• Requires more testing time
• Groups may not be equal
• Less powerful statistically

11
Between-Participants

• Random sampling
• Random assignment
• Assigning participants randomly to the groups so
  that each person has an equal chance of appearing
  in any group
• Equalization of groups on individual differences
  (IQ, gender, etc.)
• How do you do this?

12
Within-Participants

• The participants are the same in all conditions
• Each person is exposed to all levels of the IV
• Advantages
• Requires less participants and testing time
• Equivalency of groups is ensured
• Increases statistical power (more likely to
  detect an effect of the IV)
• Performance differences across groups is not
  because of error variance due to individual
  differences

13
Within-Participants

• Weaknesses
• Demanding on the subjects
• Fatigue and boredom
• High probability of order effects
• High probability of carryover effects
• High probability of demand characteristics

14
Within-Participants

• Random sample
• Breaks between blocks
• Counterbalancing
• Systematically varying the order of conditions
  for participants
• E.g., Half of the participants receive the
  shallow processing condition first, the other
  half receive the deep processing condition first
• Carryover effects are evenly distributed across
  conditions so that it does not produce
  differences that could be mistaken for effects of
  the IV

15
Statistical Power

• Type I Error Saying that there is a difference
  between groups when there really is no difference
  (false alarm)
• Type II Error Saying there is no difference
  between groups when there really is (miss)
• Statistical Power The ability of the
  statistical test to detect group differences in
  the data
• To increase power you need to reduce the risk of
  a Type I error

16
Statistical Power

• How do you reduce the risk of a Type I error?
• Reduce the alpha level (.05 ? .01)
• Warning This will reduce your power as a larger
  difference is needed to obtain statistical
  significance
• Warning This will increase your risk of a Type
  II error
• How can we reduce the risk of Type I error and
  not increase the risk of Type II error?

17
Statistical Power

• Use a One-Tailed Test
• Only if their is reason to believe that one
  sample will perform above or below the mean
• Increase the number of trials
• Increase the number of participants
• Larger samples provide more stable estimates of
  the population means (the sample mean will be
  closer to the population mean with fewer people
  in the tails of the distribution) ? less
  variability

18
Statistical vs. Practical Significance

• To say results are statically significant
  indicates that the observed differences between
  the means are probably reliable, not the result
  of chance
• A difference is either reliable or not, if your
  results are significant at the .0001 level, they
  are not any more significant than if they were
  significant at the .05 level
• This is not to say that statistically significant
  results are important
• The results were significant, Bayer Aspirin was
  better

19
Preparing Stimuli

• Previous literature
• Do not reinvent the wheel
• Databases
• Word Stimuli http//www.psy.uwa.edu.au/mrcdatabas
  e/uwa_mrc.htm (search for words of specific
  length, class, frequency, imageability etc.)

20
See you Thursday!

• Assignment 3B
• (Due Thursday)
• NIH Human Subject Tutorial
• (Due Next Tuesday)