Docstoc

erata2

Document Sample
erata2 Powered By Docstoc
					Rebecca Burwell Director of Graduate Studies Department of Psychology Brown University To Whom It May Concern, 11/8/04

I want to thank my committee for the opportunity to respond to the specific criticisms regarding my first year project. I appreciate the thoughtful and careful input into this process, and I have endeavored to address each specific comment as thoroughly and precisely as possible. I also want to thank my committee for patiently revealing my errors while providing me the chance to correct these errors. I have learned the value of providing more thorough explanations regarding the presentation of results, and I have also strengthened my understanding of degrees of freedom and the general reporting of data and statistical analysis. I have also familiarized myself with the APA manual (5th), and I have reformatted the document to reflect APA style. Based on the reformatting some major changes have been made to the paper including: reformatting of all Figures and Tables, truncation of test statistics and p values to two decimal places (as stipulated in the APA manual), and the addition of explanatory notes where needed. Again, I appreciate the through feedback and the opportunity to revise, as I realize that the feedback and revision process is a perpetually important aspect to research. I feel that my research skills have been enhanced by this experience and I look forward utilizing this experience as I continue to conduct research. Most of all, I look forward to benefiting from this experience, meeting the expectations of the program and then exceeding those expectations.

Sincerely,

Daniel Kauwe

cc:

Joachim Krueger Luiz Pessoa Jack Wright

1

Comments and Responses

(Please note, the revisions have altered page numbers and figure numbers, these changes are indicated where necessary by parentheses).

1) On p.23 (24): Report all three effects of the ANOVA on the unpartialed data with the appropriate degrees of freedom. State how the df were computed. All effects are now reported with the appropriate F, p, and degrees of freedom (p. 24). “2 X 2 ANOVA (Emotional Prime by Order of Ratings) found no effect of order (F (1, 48) = 0.96, p = .33), but a trend towards significance was found for the emotional prime (F (1, 48) = 2.80, p = .10). No significant interaction effect was found for the emotional prime and the order of ratings (F (1, 48) = 0.43, p = .51).” The computational derivation of the degrees of freedom is now stated and explained in several areas of the paper. Given that SPSS and Excel were used for the computational and statistical analysis, I have included a statement to that effect (p. 23), “Calculation of the r values and the transform procedure were conducted within Excel, and transformed values were then exported to SPSS for further statistical analysis.” and (p.24) “All statistical values, including degrees of freedom were derived via the univariate ANOVA module of SPSS.”

Within this document, I have also provided the mathematical derivation for the degrees of freedom appropriate for the analyses in question. The degrees of freedom are as would be expected given the calculation for degrees of freedom in a 2 X 2 ANOVA. For factorial analysis of variance in a 2 X 2 design, the calculations for df are as follows (as stipulated in Howell (2002), p. 428).  Total: o dftotal = N – 1; N = 49 o dftotal = 49 – 1 = 48 Effects: o df = number of the levels of the variable minus 1 o Order



2



 variable levels = 2  dforder = 2 - 1 = 1 o Emotional Prime  variable levels = 2  dfemotional prime = 2 - 1 = 1 Interaction o df = product of the df for the components of the interaction o dforder-emotional prime = dforder X dfemotional prime  dforder-emotional prime = (2 -1) X (2 -1) = 1 X 1 = 1

The df obtained via SPSS are as would be expected, to demonstrate: Manually Derived: Forder (dforder, dftotal) = Forder (1, 48) According to SPSS Printout: Forder (1, 48) = 0.96, p = .33 Manually Derived: Femotional prime (dfemotional prime, dftotal) = Femotional prime (1, 48) According to SPSS Printout: Femotional prime (1, 48) = 2.80, p = .10 Manually Derived: Forder-emotional prime (dforder-emotional prime, dftotal) = Forder (1, 48) According to SPSS Printout: Forder-emotional prime (1, 48) = 0.43, p = .51) 2) On p. 25 (26): Do the same for the ANOVA on the partialled data. All effects are now reported with the appropriate F, p, and degrees of freedom (p. 26). “The partialled data follows the patterns and trends seen in the unpartialled data. As before, the ordering of self-ratings before or after groupratings was not significant across conditions (F (1, 48) = 0.40, p = .53). However, the effect of emotional prime reached significance (F (1, 48) = 5.65, p = .02). Thus there was a statistically significant difference between the negatively primed conditions and the positively primed conditions. Again, there was no significant interaction effect (F (1, 48) = 0.19, p = .67).” As before, the computational derivation of the degrees of freedom is now stated and explained on (p. 25), “These individual partialled scores were once again transformed via the Fisher transform within Excel, and transformed values were then exported to SPSS for further statistical analysis.” and (p.23) “All statistical values, including degrees of freedom were derived via the univariate ANOVA module of SPSS.”

3

The mathematical derivation for the degrees of freedom for the partialled data (Study 1) does not vary from the partialled data (Study 1) as the levels of the design and the number of subjects remains the same.  Total: o dftotal = N – 1; N = 49 o dftotal = 49 – 1 = 48 Effects: o df = number of the levels of the variable minus 1 o Order  variable levels = 2  dforder = 2 - 1 = 1 o Emotional Prime  variable levels = 2  dfemotional prime = 2 - 1 = 1 Interaction o df = product of the df for the components of the interaction o dforder-emotional prime = dforder X dfemotional prime  dforder-emotional prime = (2 -1) X (2 -1) = 1 X 1 = 1





The df obtained via SPSS are as would be expected, to demonstrate: Manually Derived: Forder (dforder, dftotal) = Forder (1, 48) According to SPSS Printout: F (1, 48) = 0.40, p = .53 Manually Derived: Femotional prime (dfemotional prime, dftotal) = Femotional prime (1, 48) According to SPSS Printout: F (1, 48) = 5.65, p = .02 Manually Derived: Forder-emotional prime (dforder-emotional prime, dftotal) = Forder (1, 48) According to SPSS Printout: Forder-emotional prime (1, 48) = 0.19, p = .67) 3) On p. 30: Do the same for ANOVA on the unpartialled data. All effects are now reported with the appropriate F, p, and degrees of freedom (p. 30). “The data did not indicate a significant effect by emotional prime (F (1, 323) = 0.01, p = .91), and there was also no indication of a significant effect by order of ratings (F (1, 323) = 2.02, p = .16). Also, no significant interaction effect was found between the emotional prime and the order of ratings (F (1, 323) = 0.01, p = .94).” As before, the computational derivation of the degrees of freedom is now stated and explained on (p. 29),

4

“Calculation of the r values and the transform procedure were conducted within Excel, and transformed values were then exported to SPSS for further statistical analysis.” and (p.23) “All statistical values, including degrees of freedom were derived via the univariate ANOVA module of SPSS.” Based on the manual and mathematical derivation for the degrees of freedom, the presented degrees of freedom are as would be expected given the calculation for degrees of freedom in a 2 X 2 ANOVA.  Total: o dftotal = N – 1; N = 324 o dftotal = 324 – 1 = 323 Effects: o df = number of the levels of the variable minus 1 o Order  variable levels = 2  dforder = 2 - 1 = 1 o Emotional Prime  variable levels = 2  dfemotional prime = 2 - 1 = 1 Interaction o df = product of the df for the components of the interaction o dforder-emotional prime = dforder X dfemotional prime  dforder-emotional prime = (2 -1) X (2 -1) = 1 X 1 = 1





The df obtained via SPSS are as would be expected, to demonstrate: Manually Derived: Forder (dforder, dftotal) = Forder (1, 323) SPSS Printout: Forder (1, 323) = 2.02, p = .16 Manually Derived: Femotional prime (dfemotional prime, dftotal) = Femotional prime (1, 323) SPSS Printout: Femotional prime (1, 323) = 0.01, p = .91 Manually Derived: Forder-emotional prime (dforder-emotional prime, dftotal) = Forder (1, 323) SPSS Printout: Forder-emotional prime (1, 323) = 0.01, p = .94) 4) On p. 31: Do the same for the ANOVA on the partialled data. All effects are now reported with the appropriate F, p, and degrees of freedom (p. 31).

5

“Compared to the unpartialled data, the effect for order of ratings was significant (F (1, 323) = 5.41, p = .02). As with the unpartialled data, there was no significant evidence for an effect by emotional prime (F (1, 323) = 0.10, p = .76), nor was there evidence for an interaction effect (F (1, 323) = 0.001, p = .98).” As before, the computational derivation of the degrees of freedom is now stated and explained on (p. 30), “These individual partialled scores were once again transformed via the Fisher transform within Excel, and transformed values were then exported to SPSS for further statistical analysis.” and (p.23) “All statistical values, including degrees of freedom were derived via the univariate ANOVA module of SPSS.” The mathematical derivation for the degrees of freedom for the partialled data (Study 2) does not vary from the partialled data (Study 2) as the levels of the design and the number of subjects remains the same.  Total: o dftotal = N – 1; N = 324 o dftotal = 324 – 1 = 323 Effects: o df = number of the levels of the variable minus 1 o Order  variable levels = 2  dforder = 2 - 1 = 1 o Emotional Prime  variable levels = 2  dfemotional prime = 2 - 1 = 1 Interaction o df = product of the df for the components of the interaction o dforder-emotional prime = dforder X dfemotional prime  dforder-emotional prime = (2 -1) X (2 -1) = 1 X 1 = 1





The df obtained via SPSS are as would be expected, to demonstrate: Manually Derived: Forder (dforder, dftotal) = Forder (1, 323) SPSS Printout: Forder (1, 323) = 5.41, p = .02 Manually Derived: Femotional prime (dfemotional prime, dftotal) = Femotional prime (1, 323) SPSS Printout: Femotional prime (1, 323) = 0.10, p = .76

6

Manually Derived: Forder-emotional prime (dforder-emotional prime, dftotal) = Forder (1, 323) SPSS Printout: Forder-emotional prime (1, 323) = 0.001, p = .98) 5) On p. 29: Describe how error bars were computed for data in Figure 14 (Figure 13). Additional description is now given on p. 29, “Mean projection scores and standard errors are based on r values derived from the Fisher transform. Mean projection scores and standard error bars were computed and graphed in Excel, with standard error based on sample standard deviation divided by the square root of n. Error bars depict one standard error above the mean, and one standard error below the mean” In computing the standard error, the calculations were based on the equation

SE  SD/ n , yielding the following results:
Condition Group+ GroupSelf+ Selfmean 0.34 0.35 0.41 0.41 SD 0.44 0.45 0.40 0.42 n 86 59 82 97 SE 0.05 0.06 0.04 0.04



6) On p. 31: Describe how error bars were computed for Figure 15 (Figure 14). Additional description is now given on p. 31, “Mean projection scores and standard errors are based on r values derived from the Fisher transform (mean projection scores and standard error bars were computed and graphed in Excel, with error bars depicting one standard error above and below the mean).” In computing the standard error, the calculations were based on the equation

SE  SD/ n , yielding the following results:
Condition Group+ GroupSelf+ Selfmean 0.30 0.28 0.41 0.40 SD 0.45 0.50 0.41 0.40 n 86 59 82 97 SE 0.05 0.06 0.05 0.04



7) On p. 33: Describe how error bars were computed for Figure 16 (15).

7

In the previous paper, the standard error bars presented in Figure 15 were calculated incorrectly ( SD / n was used instead of SD/ n ). The correct calculation has been used to correct the standard error bars, and the appropriate standard error bars are now displayed in Figure 15. The computation of the standard error bars is described p. 33,   “Mean projection scores and standard error bars computed and graphed in Excel, with error bars depicting one standard error above and below the mean.” In computing the standard error, the calculations were based on the equation

SE  SD/ n , yielding the following results:
Condition Group+ High Group+ Low GroupSelf+ SelfMean 0.18 0.40 0.28 0.41 0.40 SD 0.39 0.48 0.50 0.41 0.40 n 40 46 59 82 97 SE 0.06 0.07 0.06 0.05 0.04



8) Explain whether the data in Figure 16 (Figure 15) correspond to the data also shown in Figure 14 (Figure 13) or Figure 15 (Figure 14), explain the differences in means and standard errors across figures. The data in Figure 15 corresponds to the data in Figure 14. This connection has been made more clear in the text. p. 32 A median split was performed via AIM scores upon the partialled projection scores from Study 2 (presented in Figure 14) p. 33 “The graphical presentation of the data is identical between Figures 14 and 15, except for pairs of sub-groups that demonstrated significant mean differences on the High and Low split. For pairs of High and Low sub-groups with a significant difference, separate and means and standard errors are presented. Conditions with no significant difference on the High-Low split, are presented with the mean projection scores and standard errors first presented for the Study 2 unpartialled data (Figure 14).”

8

p. 33 “As shown in Figure 15, only the Group Positive condition showed a significant difference between High and Low sub-groups (t (84) = 2.28, p = .025, two-tailed), and this difference is represented by separate and distinct bars for Group Positive High and Group Positive Low. The other three conditions (Group Negative, Self Positive, Self Negative) did show a significant between High and Low sub-groups (complete descriptive data located in Table 6), and therefore we continue to present the original condition mean projection scores and standard errors (as first presented in Figure 14).” A table of descriptive data for the AIM analysis has also been included (p. 34).

Condition N
Group+ High 40 Group+ Low 46 Group- High 30 Group- Low 29 Self+ High Self+ Low Self- High Self- Low 42 40 45 52

Mean Projection Score
0.18 0.40 0.40 0.16 0.35 0.47 0.41 0.39

Projection Score SE
0.06 0.07 0.10 0.07 0.06 0.07 0.06 0.06

Mean Group Difference
0.22 -0.23 0.12 -0.02

Significance (Two tailed)
t (84) = 2.28 p = .025 t (57)= -1.85 p = .07 t (80) = 1.36 p = .18 t (95) = -0.21 p = .83

9) Please format the manuscript according to the publication manual of the American Psychological Association. This paper is now presented in APA format A. Where appropriate, p values were truncated from three decimal places to two decimal places as shown in the APA manual. B. Where appropriate, F values were also consistently truncated to two decimal places. C. Leading zeroes were added where appropriate. D. Changed page numbers from footer to header positions. E. Added abstract. F. Added running header to title page. G. Added running header through out document. H. Moved captions to top of Tables/Figures I. Figures and Tables were retained in text as the APA manual suggests that this is more appropriate for the submission of student theses.

9

References
Howell, D. C. (2002). Statistical Methods for Psychology (Fifth ed.): Duxbury, Thomson Learning.

10


				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:3
posted:10/16/2009
language:English
pages:10