Resampling Assignment for PSYC 7431, Spring, 2018:

Earlier you completed an assignment evaluating the effectiveness of Floating Anxiety Reduction Therapy for treatment of gastrointestinal anxiety disorder (GAD). For that assignment, you conducted a nonparametric analysis, the Wilcoxon Signed Ranks Test. For this assignment you will use the same data but will use resampling analysis.

Conduct a permutation/randomization test using David Howell’s resampling software, which is available on Howell’s Resampling Statistics page and in BlackBoard (Documents, Nonparametric and Resampling Statistics). You will first need to create a data file (*.dat). In that data file the first line should contain the number of pairs of scores that follow. Each of the remaining 15 lines should contain the two scores for one pair of patients, separated by a blank space. You can use any text editor (such as Notepad) to edit the dowloaded data file to conform with the required format. To the right is an example of what the data file will look like: /

Run 10,000 replications. Paste into your Word document the screen that presents the results. To do so, select the window with the output, hold down the Alt key while you hit the Print Screen key, put the cursor at the insertion point in the Word document, and then hold down the Ctrl key while you hit the V key.

Here is an example:

A permutation/randomization test using David Howell’s resampling software indicated that the patients treated with the experimental therapy experienced significantly less gastric distress than did those on the waiting list, p < .001.

If you are unable to use David Howell’s software, try using R. Paste into your document the two lines that give the results, like this:

The probability from the sampling statistics is = 4e-04 That is, p = .0004

The t value from a standard matched-pairs t test is= 6.092172

Now try a different resampling approach. Construct a 95% confidence interval for the median difference in GAS for a waitlist patient versus for a treated patient. If that CI excludes zero, then the groups differ significantly. First create a data file which has on the first line the number of difference scores that follow. On the next line type the difference scores, with each one separated from the next by a blank space. Here is an example of such a data file:

Use 10,000 replications. Paste the output screen into your Word document and then type a brief interpretation of the output, like this:

David Howell’s resampling software was employed to estimate the median amount by which placebo patients’ gastric distress exceeded that of treated patients. The 95% bootstrapped confidence interval ran from 15 to 29.

If you are unable to use David Howell’s software, try using SPSS or SAS. Paste in the output like this:

Statistics
Diff
Statistic / Bootstrapa
Bias / Std. Error / 95% Confidence Interval
Lower / Upper
N / Valid / 15 / 0 / 0 / 15 / 15
Missing / 0 / 0 / 0 / 0 / 0
Mean / 20.3333 / .0018 / 3.3319 / 13.0701 / 26.3333
Median / 23.0000 / .4860 / 4.0829 / 17.0000 / 32.0000
a. Unless otherwise noted, bootstrap results are based on 1000 bootstrap samples
Bias Reduced upper and lower confidence intervals for Mean and Median using Bootstrap
Data = Diffs Variable=Diff CI=95
Obs / Mean Diff / Median
Diff / Lower Mean / Upper Mean / Lower Median / Upper Median
1 / 20.3333 / 23 / 14.3170 / 26.3497 / 15.7749 / 30.2251

Lastly, construct a 95% confidence interval for the mean difference in GAS. Use the same data file you used for the median differences. Paste in the output screen and provide a brief interpretation, like this:

A 95% bootstrapped confidence interval for the mean amount by which GAS was less among the treated patients than among the waitlisted patients runs from 8.984 to 26.311.

Attach your Word document (named “Nnnn_Resampling”, where “Nnnn” is your last name) to email (with subject line “PSYC 7431: WSRT”) and deliver it to Karl by noon on Friday the 13th of April, 2018 /