Assignment 3: DATA ANALYSIS & RESULTS WRITE-UP Due in lecture Tuesday, April 15, 2008 Start with: ALL

group members names, TA name, Discussion time & date on Assignment 1. Put the heading Results (it is a first level heading see Method/Assignment 2 handout for the different levels of headings) on top of page. Then start with a new paragraph.

Follow the following steps for EACH of your Hypotheses (that is, think about writing one paragraph to address the results for each of your hypotheses). IMPORTANT: at the end of this assignment 3 handout, I am providing a lot of examples as to how to write about the statistical tests correctiy. Take the time to read them over and use them to help you write about your results more intelligently!! Writing up the result section is difficult b/c it does require some statistic speak. I provided the examples to help you with this!
2. Restate your first hypothesis. 3. State what statistical analysis you conducted to test the hypothesis 4. Then report/describe results of those analysessee points A. & B. below Use APA format to report inferential statistics! 5. Make a statement about whether the statistics/results you report (point 4.) then provide support for your hypothesis or not (that is, provide an interpretation of your statistical findings in terms of your hypothesis) 6. Start a new paragraph, restate your second hypotheses, and continue as you did for your first hypothesis. ============================================================================= A. If you are computing/reporting correlations (you get this information from your SPSS printouts) 1. Correlations: Report r, whether p < .05 or n.s.& report exact p-value (also can report something as marginally significant if p < .10) B. If you are comparing means of two (or more) groups (you get this information from your SPSS printouts) (e.g., men vs. women; experimental vs. control condition) =>Conduct one of the following analyses, depending on your design: 1. Independent Sample t test--if you have 2 independent groups 2. Paired sample t test--if you have the same (or related, non-independent) participants providing data For both types of t tests report means, standard deviations report inferential statistics: t, df, p values--whether < .05 (or < .10) or n.s. and exact p value in ( ) If you have more than two groups hopefully this will apply to none of youthats why I put it in green 3. One-Way ANOVA if 3 or more conditions 4. 2 x 2 ANOVA For both ANOVA designs create bar charts (or histograms) (or line graphs if a 2x2); attach copy report means, standard deviations report inferential statistics: a. F, df, p values--whether < .05 or n.s. and exact p value in ( ) b. probably also need to do follow-up t tests to compare individual conditions (if so, report info listed above under t testspoints B.1. or B.2.) =>Use APA format to report statistics. See book Chap 15 & examples below & class handout/notes.

--------------------------------------------------------------------------------------------------------------------------------------C. Other analyses you need to conduct although these statistics should be reported in the Method section => However, you will report the results of these analyses in the Method section rather than the Result section. Look back at the example write-up of a Method section I included in assignment 2; for example, it will tell you how to report information about your scale scores and alpha reliability analyses and participants; I do not provide a section for Outlier analyses but you all should have a final paragraph in your Method section that says what you did to check for outliers and what you found even if you say that you did not find any outliers.) => So, we suggest that you hand in a complete revised Method sectionor at the very least, provide parts of the Method section that have the following information in them: 1. Participants: report percentages of gender & ethnicity; mean & range for age; any other statistics about your Ps you measured and you feel are important to provide 2. Measures: for each major measure (construct), report means, standard deviations; if multiple items, report # of items & Alpha reliability (report in same section in which you describe the measure/items, etc.) 3. Any other reliability information (e.g. inter-rater) should go into the Method section as well. Only information that tests the hypothesized relations should be reported in the Results section. 4. Outliers: How did you check for outliers (e.g., look at descriptive statistics are min, max what you expect? Maybe compute z-scores for key variables & also look at scatterplots if you are running correlations)? Did you find any? Say if you did not; if you did, ecide what to do with them & describe in Method section).

This is what you should describe in terms of your results and what we want you to hand in for Assignment 3 (remember, some of the results will go into the Method section, so you need to hand in at least parts of your method section that have results, as well as the results section that provides information about the results you obtained with regard to testing your hypotheses!). Below I provide additional information that will be extremely useful to you in terms of writing about your results (for the Method section write-up example look at assignment 2!!)

Some additional Pointers for Result Section Write-ups

Writing up Test Results Remember to always remind the reader why you are doing what you are doing re-stating the hypothesis is typically enough. Then say how you tested the hypotheses (independent samples t test, paired samples t test, Pearson correlation coefficient, etc.). Then state what you found. Make sure that you make a statement about the direction of the effect (this is true no matter whether you are reporting correlations or t tests or both). Particularly for correlations that is important when it isnt necessarily clear in what direction you scored your variables. Report the actual statistics (see OHs from Chapter 15, examples below, textbook, and assignment). Make sure that somewhere here you make an explicit statement as to whether your findings are statistically significant or not. Draw some conclusion about whether these results are consistent with and supportive of your hypothesis or not. Restate your hypothesis (We hypothesized that) State what test you did to test your hypothesis (To test this hypothesis we conducted an independent samples t test State what the results were (Results suggested...) Comment briefly on the results, leaving most commentary for the discussion (These results indicate no significant relationship between x and y in this sample) Make sure that the order of your hypotheses as introduced in the introduction is consistent with your result section and, generally, with your discussion. Information to Include in Reporting Test Statistics (the numbers themselves) Mean differences (t, F, and chi-square tests). For t tests (F tests, anc chi-squares tests), include the numerical value obtained for the statistic, degrees of freedom, and probability level (state whether less than .05 or .10, marginally significant and then provide exact probability). See OHs and Assignment. You MUST report actual means and standard deviations whenever you report that means differed. These may be provided in a table but, again, as we discussed in class and see OHs for class, only use tables if you have a lot of information to summarize. Correlations. For correlations, provide the actual correlation and probability level (state whether less than .05 or .10, marginally significant and then provide exact probability). See OHs and Assignment. All results. For all results (correlations and mean differences), indicate the direction of effect; for example, the anxiety group scored significantly higher than the control group is better than the scores of the two groups were significantly different. Oftentimes it might be easier not to try and put the information about the direction of the effect and the actual statistical information in the same sentence. So, you could provide a statement about the direction of effects and then follow this info with the actual t-test results. For correlational results, b/c there is less information to provide, it might be easy enough to provide the statistical information and directional information in the same sentence. See examples below as to how to talk about your results. No matter what, ask someone else to read over what you write (a friend) and see if they know what it means. REMEMBER: do not call findings that did not reach conventional statistical significance INsignificant they are simply nonsignificant in terms of the particular level of significance you chose to evaluate your findings at (should be p < .05 for this class)

I. V.

Examples

Examples of t test Write-ups As predicted, men and women differed significantly in how many drinks they considered to constitute binge drinking, t(59) = 4.95, p < .05 (p = .012), with men reporting a higher number of drinks (M = 7.21, SD =1.75) than women (M = 4.20, SD = 1.20 ). [OPTIONAL --The size of this effect was moderate (r = .28).] Overall mean ratings of freshness did not differ as a function of age, t(49) = .03, ns (M = 6.12, SD = 1.02 for old and M = 6.50, SD = 0.93 for new). There was a tendency (t(50) = 12.37, p < .10, p = .061) for managers to rate themselves as braver (M = 4.21, SD = 1.11) than their employees rated them (M = 4.02, SD = 1.02). This difference was only marginally significant, however, and the effect size was small (r = .18). [again, effect size estimates are optional when you do a t test] Independent sample t tests showed that dogs scored significantly higher than cats on both the first test battery, t(29) = 22.42, p < .05 (p = .023), and on the second test battery, t(29) = 19.32, p < .05 (p = .04). The means and standard deviations are shown in Table 1. Examples of Correlation Write-ups Reporting correlation results can be a preposition nightmare. Here are some acceptable statements. X correlated significantly with Y X and Y were significantly correlated The correlation between X and Y was significant The correlation of X and Y was significant Correlations among X, Y, and Z were computed The correlation of X with Y was significant The hypothesis that higher levels of political conservatism would be positively related to earned income was not supported, r = .04, p >.05 (p = .18). The correlation between swimming speed and hiking speed was not significant, r = -.09 (p = .34). Our hypothesis that faster swimmers would also be faster hikers thus was not supported. Our secondary analyses showed that the association between binge drinking and GPA differed for men and women. For men, the correlation was nonsignificant (r = -.13, p = .18), whereas for women the correlation was moderate in size (r = . 37) and statistically significant (p < .05, p = .025). These results suggest that binge drinking was only significantly associated with lower grades for women but not for men.