Lab Report Writing
Donald Trump
lab_report-results-_change_blindness.ppt2007PSY – Mt Gravatt
*
2007PSY Biological Psychology 2017
Week 6 Tutorial
Visual Change Blindness Experiment: How to Write up the Practical Report II
Results
2007PSY – Mt Gravatt
2007PSY – Mt Gravatt
*
For your information:
All Information and files required to complete the Laboratory Report Assignment are now posted on the Learning@Griffith website.
2007PSY – Mt Gravatt
2007PSY – Mt Gravatt
*
Data Screening
2007PSY – Mt Gravatt
2007PSY – Mt Gravatt
*
Let’s have a look at the data file
Change Blindness Data File final.sav
There are six variables to look at: Age, Gender, NoChange, Addition, Subtraction and Other
2007PSY – Mt Gravatt
2007PSY – Mt Gravatt
*
Variable View
Age: The age in years of the participants. All the data of those who attended the Week 3 tutorials and submitted data using the google document
Gender: Whether the participant was Male or Female
NoChange: How many of the pairs in which there was no change did the participant get correct, max = 8, min = 0
Addition: How many of the pairs where something was added did the participant get correct, max = 8, min = 0
Subtraction: How many of the pairs where something was taken away the did the participant get correct, max = 8, min = 0
Other: How many of the pairs where the colour change or something moved did the participant get correct, max = 8, min = 0
2007PSY – Mt Gravatt
Let’s Screen the Data
- Are there any outliers or data entry errors?
- Are there any empty data cells? If so, what should you do about these cells?
- Make sure you report that the data has been screened.
- This doesn’t require evidence in your SPSS output
2007PSY – Mt Gravatt
*
2007PSY – Mt Gravatt
2007PSY – Mt Gravatt
*
Descriptive: are the variables normally distributed?
What is the key assumption of ANOVA?
Are the distributions of the scores in the change blindness variables normally distributed.
That is: are the variables skewed?
To calculate skewness:
Skewness = skewness statistic / SE
YOU ARE NOT EXPECTED TO TRANSFORM THE DATA! Rather, simply state whether a variable is skewed or normally distributed.
2007PSY – Mt Gravatt
*
2007PSY – Mt Gravatt
*
Just a reminder that you have already looked at Descriptives in relation to the demographics of the participants in the Method section
2007PSY – Mt Gravatt
2007PSY – Mt Gravatt
*
It is important that you graph the data. This should be means as well as error bars of some form (see screen capture on left for SPSS functionality). Think about what you are doing – clearly the pie chart on the right is not the way to go.
2007PSY – Mt Gravatt
2007PSY – Mt Gravatt
*
Statistics
2007PSY – Mt Gravatt
2007PSY – Mt Gravatt
*
Information on What Statistics to Do
from 2000 PSY – Research Methods and Statistics II
- Witte, R.S. & Witte, R.S. Statistics. (9th Ed). John Wiley & Sons, 2010.
- Chapter 17 – Analysis of Variance (Repeated Measures)
- Paired Samples t-tests. Remember for multiple comparisons you must correct for inflated Type I error in your α-level (Bonferroni correction).
2007PSY – Mt Gravatt
Repeated Measures ANOVA
- Report F statistic.
- What about sphericity?
- Is the assumption of homogeneity of variance violated?
- What do we do?
2007PSY – Mt Gravatt
*
2007PSY – Mt Gravatt
2007PSY – Mt Gravatt
*
- Clearly the design is a repeated measures design. Each and every participant did each and every type of change condition.
- So there are four repeats, and first we have to see if there are any differences at all among those four repeated conditions
- But based on the hypotheses about what condition(s) would yield a greater number of correct responses you will need to do comparisons – possible many times – among scores on each condition.
- Report Effect Size (partial eta squared)
2007PSY – Mt Gravatt
2007PSY – Mt Gravatt
*
Comparisons between the different conditions
- There are a number of different ways of doing this:
- We recommend PAIRED-SAMPLES T-Tests
- Analyse compare means paired samples t-tests
- If you are making multiple comparisons, we need to correct for inflated Type I error rate
- Bonferroni (a = .05 / no. of comparisons .05/ 6 = .008
- We check significance by seeing if p < .008
- Remember, there is not just one and one only correct way of doing this
- REPORT YOUR t-test RESULTS IN A TABLE
2007PSY – Mt Gravatt
2007PSY – Mt Gravatt
*
What about the variables Age and Gender?
- Do not include these in the statistical analysis of differences among conditions unless you have a hypothesis backed up with relevant literature relating to age or gender, e.g. women are better at seeing change than men.
2007PSY – Mt Gravatt
2007PSY – Mt Gravatt
*
Statistical Analysis Output
Provide a copy of the SPSS output (the .spv file) in an Appendix
2007PSY – Mt Gravatt
