Name______

Charlotte Mandell, Research 3

Statistics Lab 1 - Independent t-test

Data Analysis

1) Below is a hypothetical experiment with data included: Please perform a data analysis using SPSS. Include data inspection, description, and a t-test with supporting graph showing confidence intervals. Save your data file. Save your output file. Send both to me as attachments.

2) Write a brief paragraph explaining your conclusions about the data in this study. Do you think that it showed a significant difference? Why or why not.

Thirty undergraduate students were recruited for a research project in a psychology class. The students were randomly assigned to two different groups: Group A saw a moving fictional movie dealing with the death of a young man in battle. Group B saw a documentary on the war in Iraq, including the numbers of soldier killed or wounded. Empathy was then measured with the Interpersonal Empathy Test which had 25 questions scaled from 1 to 4. Thus test scores could range from 25 to 100. The researchers wanted to determine whether there was a difference in empathy between those students who saw a moving fictional account of a death; and those students who got reality-based information. The empathy scores are below:

Group A:

67 / 98 / 68 / 77 / 82 / 77 / 69 / 80 / 84 / 92 / 77 / 87 / 84 / 72 / 79

Group B:

48 / 69 / 82 / 55 / 59 / 52 / 49 / 75 / 78 / 66 / 48 / 57 / 61 / 59 / 70

I. Enter Data - Be sure to end the dependent and independent (labeled numeric) variable for each participant. The dependent variable is scalar; the independent variable is nominal.

I. Inspect: Using Explore, inspect box-plots and look for outliers and Clinkers. If any exist, change them. Using box plots describe the shape of your distributions - symmetrical? Skewed?

II. Describe: Using data from Explore , complete the following table

Variable A / Variable B
Mean
S.D.
Minimum
Maximum

III. Estimate:

A. Confidence intervals

Using explore what are confidence intervals for mean of Variable A and Variable B?

Confidence interval for A ______

Confidence interval for B ______

Using error bar graphs - create a graph of considence intervals for variables A and B

B. Independent Samples t-test

Conduct t-test - Fill in the following information:

df = ______

T = ______

P (significance) = ______

Effect size (g) = (M1-M2) / spooled = ______

IV. Announce - here is a sample results section for another study - using this as a sample, please announce the results of the analysis that you just completed. Send it to me with copy of output. (I have been vague about treatments A and B and the scores -- you would be more specific.)

Sample results

This experiment tested the hypothesis that treatment A will produce lower scores than treatment B.

Inspection of the data showed that neither group contained any suspicious data points or significant skewness. The scores for group A seemed lower than those for group B, as predicted. The A scores ranged from 4 to 7 and the B scores from 6 to 9 out of a possible 10 points. The confidence intervals for group A ranged from 5.5 to 6.7 and for group B from 6.8 to 9.2. The mean for group A was 6.1(SD = 1.2); the mean for group B was 7.8 (SD = 1.5). The means and confidence intervals for both groups are shown in Figure 1. There is no overlap between the mean of each group and the error bars of the other, suggesting that there is a significant difference between them.

A two-tailed independent samples t-test indicated that the scores for group A were significantly lower than those for group B, t(24) = 3.68, p = .040, g = .45. Thus we can reject the null hypothesis, that our treatment had no effect on scores.