(ii)       Was the comparison between conditions statistically significant? Write a sentence reporting

the outcome of the test.

(iii)     The SPSS output also shows the correlation between the two conditions. Use what you have

learned in PS105 to perform a (bivariate) correlation between Neutral_N170 and Angry_N170. You should be able to confirm that you get the same value for the correlation coefficient.

(iv)      How would you describe this correlation?

(c) We can work out Cohen’s d as a measure of the effect size for the difference between our two N170 samples.

(i)        Calculate the effect size by taking the difference between the means in Neutral_N170 and Angry_N170 and then dividing by the standard deviation from Neutral_N170. Write down your value here.

(ii)

SPSS can also calculate Cohen’s d. Repeat the t-test and make sure that the option marked   Estimate effect sizes is ticked. Select Average of the variances as the standard deviation to divide by. The estimated effect size is given in the Point Estimate column (see below). It will be different from the one you calculated in part (i).

(iii)      Blechert et al. (2012) also compared N170s to neutral and angry faces. Part of the table that

they report in their results is shown here. Use these results to calculate Cohen’s d for their data (again, use the standard deviation for the neutral sample). How does our effect size    compare to theirs?

For this example, we will first open the data in Excel.  Open the file PS106week2_testosterone.xlsx. This file shows the change in heart rate a few seconds after seeing an angry face.

(b) We need to re-organise this data so that the conditions are together and so it is in a suitable format for a paired-samples t-test.

First, we want to make two new columns which will contain data for each participant      separately in the two conditions. At the top of an empty column (such as column G), type “Placebo” and then in the next column “Testosterone” .

The data in the spreadsheet is in a randomized order, but we can sort it so that the conditions are together. In Excel, click somewhere in the main table and then click >Data > Sort… In the column next to where it says “Sort by”, select the “Condition” column and click OK.

You should now have all the Placebo scores together, and all of the Testosterone scores together. Copy the scores from the first condition (the first 16 rows of the data) and paste         them into your new column. Copy the scores underneath and paste them into the second new column.

We can also use Excel to calculate the difference between conditions. In a third new column, type a formula for the first participant by typing “=” and then click on the Placebo score, type a  minus sign, and then click on the Testosterone score and press enter. Your spreadsheet should  now look like this.

Copy the formula down for all participants (by clicking and dragging the small square in the bottom right corner of the formula cell).

Finally, create a new data sheet in SPSS. Copy the two columns and paste them from Excel     into SPSS. (You do not need to include the headings or the difference scores). Use the Variable view to name the two columns something sensible.

(c) Carry out a paired-samples t-test to examine the difference between Placebo and Testosterone groups. (You can check the full results in the answers at the back of this handout)

(i)        What do you conclude from this experiment?

(ii)       If you look back at Van Honk et al’s Method, you will see that they lost some participants

because of “apparatus failure” . How many participants did they lose? (Hint: you can tell by looking at the degrees of freedom as well as looking back at the raw data).

(iii)      The SPSS output displays the mean of the paired differences (as well as the standard

deviation and standard error of these differences). We can also work these out in Excel.

Go back to Excel and write an “=AVERAGE()” formula under the difference scores. You will need to delete any error cells which appear because of missing participant data. Confirm    that the average difference score here is the same as that in the SPSS table.

(iv)      Which participant showed the greatest effect of testosterone?

Example 3. Moral dilemmas and fMRI

The last example in Lecture 5 (Example B) came from the study by Greene et al. (2001) in which       participants responded to different types of moral dilemma while in an fMRI scanner. Download the data from moodle (PS106week2_morality.sav) and replicate the analysis from the lecture handout. Perform additional t-tests including for the NonMoral condition. What do you find?

References

Blechert, J., Sheppes, G., Di Tella, C., Williams, H., & Gross, J. J. (2012). See what you think: Reappraisal modulates behavioral and neural responses to social stimuli. Psychological Science, 23(4), 346-353.

Greene, J. D., Sommerville, R. B., Nystrom, L. E., Darley, J. M., & Cohen, J. D. (2001). An fMRI investigation of emotional engagement in moral judgment. Science, 293(5537), 2105-2108.

Lakens, D. (2013). Calculating and reporting effect sizes to facilitate cumulative science: a practical primer for t-tests and ANOVAs. Frontiers in psychology, 4, 863.

van Honk, J., Tuiten, A., Hermans, E., Putnam, P., Koppeschaar, H., Thijssen, J., ... & van Doornen, L. (2001). A single  administration of testosterone induces cardiac accelerative responses to angry faces in healthy young women. Behavioral Neuroscience, 115(1), 238.