This is the final step. This page contains a list of hypotheses and analyses to test them. What you should turn in for the final project/paper are the following three things:

1. A print-out of your data file. If you'd like you can remove the "code" and "code2" columns before printing, removing all the initials you used. (If you turned in such a print out earlier, you can turn the same pages in again.)

2. The output from the analyses indicated below. NOTE: You should not turn in print-outs of all outputs you generate (this could waste paper) -- only those for the analyses indicated below, which are relatively few.

3. A page or two (of your own words, typed/in print) that functions as a "Results" summary. In it, for each hypothesis you should state the hypothesis (e.g., "The hypothesis was that..."), state the analysis you ran to test it (e.g., "To test this hypothesis, I conducted...."), and describe whether the hypothesis was supported or not and how you arrived at this conclusion.

At the end of this Results summary, you are free to add a "Discussion" in which you offer any comments about why you think the results came out the way they did -- you can be eligible for some extra credit if your comments are especially useful or insightful; such additional comments are not required however to get full credit.

Bear in mind that the population from which your subjects/participants was drawn is probably "People whom I personally know best" rather than "People in general" -- it is a very good sample of people in your world, but not necessarily people in the world. Thus, your results need not agree with (and should not be perfectly identical to) those of any other student in the class.

In order to test hypotheses 2 through 4, you will need to use Transform-->Compute to create two new variables. Name one Intvext (target variable) using the formula (numeric expression) Internal - External. (Intvext stands for internalizing versus externalizing; a high score on this variable would indicate high internalizing and low externalizing tendencies, a low score would indicate low internalizing and high externalizing tendencies -- under ordinary conditions). Name the other Intvext2 using the formula Interna2 - Externa2. (This is internalizing versus externalizing but under stressful conditions, when things aren't going well.)

Here are the hypotheses and the analyses:

Hypothesis 1: Internalizing scores will be negatively correlated with Externalizing scores, for both "how the person is in ordinary life" and "how the person is under stress, etc."

Analysis: Compute bivariate correlations among the two internalizing and two externalizing variables, and determine whether Internal and External are significantly correlated and whether Interna2 and Externa2 are significantly correlated. Please also select under "Options" for this analysis, Mean and Standard Deviations for these variables, and include these descriptive statistics with the correlations print out.

Hypothesis 2: Overall, being under stress causes people (that I know) to become different with respect to internalizing versus externalizing tendencies, compared to how they are in ordinary life. Note: this is a nondirectional hypothesis.

Analysis: Run a paired samples t-test (i.e., t test for dependent means), with Intvext2 and Intvext as the pair of variables. Determine whether there is a significant difference for condition (ordinary life vs. stress); if there is you should determine and report the direction of effect.

Hypothesis 3: Women are more prone to internalizing tendencies and men to externalizing tendencies, whether in ordinary life or under stress. Note: This hypothesis is based on the observation that men are more prone to externalizing disorders (e.g., antisociality, substance abuse) whereas women are more prone to internalizing disorders (e.g., depression). An alternative research hypothesis you might wish to take into account: One could argue that hypothesis 3 might need work for the "under stress" condition only, on the assumption that under stress (and only then) men externalize more and women internalize more.

Analysis: run independent samples t-tests (i.e., t test for independent means) with Female as grouping/independent variable and Intvext and then Intvext2 as test/dependent variables. For "Define Variables" set Female=0 as Group 1 and Female=1 as Group 2. Determine whether there is a significant effect for the gender (Female) variable; if there is you should determine and report the direction of effect.

Hypothesis 4: The effect of gender on internalizing and externalizing tendencies interacts with (or is moderated by or affected by) how well-known the person is (to me). Note: The rationale for this hypothesis is that rating people who are not well-known would obscure the real effects that would be evident if we only rated the people we know really well.

Analysis: under General Linear Model --> univariate on SPSS, run an ANOVA (this is a two-way ANOVA) with Intvext as dependent variable and as "fixed factors" both Female and Well_kno variables. Under Options select "estimates of effect size", and also "estimates of marginal means" for Female, Well_kno, and their interaction Female*Well_kno. After completing this analysis, repeat it substituting Intvext2 as the dependent variable. If you find a significant interaction, you should check the estimates of marginal means to help you make sense of it. The main effect for gender in this ANOVA analysis will usually but not always correspond to the results of the t tests for hypothesis 3; if one is significant and the other not, it may be due to the effects of "controlling for how well-known the person is" in the ANOVA analysis.

Those are the four hypotheses and analyses. If you think there are other hypotheses or analyses that should be tried, you are encouraged to describe them and why you think they would be advisable (no need to carry them out though), in an added Discussion section. As noted above, the Discussion section is not required, but if it is especially useful or insightful it can make you eligible for some extra credit.

Final paper/project is due Friday, March 14 by 4 pm.

Characterizations of the overall pattern of results found by all members of this class in their final paper/project -- functioning as many dozen replications of the same study! -- will be reported in an e-mail sent out via Blackboard at the end of the term. This characterization of the overall pattern of results will not mention any student's name but will simply concern how frequently each of the hypotheses was supported (or not) overall (e.g., that Hypothesis 3 was supported half the time, if it was).