Awesome analysis, Thanks! Paul
On Nov 1, 2012, at 1:48 AM, Jim Clark wrote: > Hi > > Paul's observations were intriguing. One thing to note is that the p > value for the Message main effect is .1128 for F(1, 34) = 2.65. So very > close to significance by a directional test (i.e., one assuming the > priming effect mentioned by Paul). It was simply reported as ns in the > paper, without a p value. > > So as to analyze something like the original paper's data more fully, I > generated some values very close to those in the paper, except that I > did not have unequal Ns. I used n = 7 per cell to keep fairly close to > the df error from the original study. Here is the hypothetical data and > the descriptive statistics. I hope the formatting does not get too > messed up below (like it just did when I unwisely switched between text > and html mode). > > data list free / mess hurr help. > begin data > 1 1 1 1 1 5 1 1 5 1 1 4 1 1 2 1 1 5 1 1 5 > 1 2 1 1 2 3 1 2 3 1 2 0 1 2 0 1 2 5 1 2 2 > 1 3 0 1 3 0 1 3 0 1 3 1 1 3 4 1 3 0 1 3 2 > 2 1 0 2 1 4 2 1 4 2 1 2 2 1 0 2 1 0 2 1 2 > 2 2 3 2 2 2 2 2 0 2 2 5 2 2 0 2 2 0 2 2 2 > 2 3 0 2 3 0 2 3 0 2 3 0 2 3 0 2 3 0 2 3 4 > end data. > > glm help by mess hurr /print = descr. > > mess hurr Mean Std. Deviation N > 1 1 3.86 1.676 7 > 2 2.00 1.826 7 > 3 1.00 1.528 7 > Total 2.29 2.004 21 > > 2 1 1.71 1.799 7 > 2 1.71 1.890 7 > 3 .57 1.512 7 > Total 1.33 1.742 21 > > Total 1 2.79 2.007 14 > 2 1.86 1.791 14 > 3 .79 1.477 14 > Total 1.81 1.916 42 > > The cell means (above) and MSE (below) are very close to those in the > published paper, although the column means in particular are off because > of the unequal Ns in the paper. The following analysis shows more robust > effects than the analysis in the paper, presumably because of the > unequal Ns analysis they did. Note the message effect now has p = .08, > which would be significant by a directional test. > > Source Type III Sum of df Mean Square F Sig. > Squares > Corrected Model 45.048(a) 5 9.010 3.076 .021 > Intercept 137.524 1 137.524 46.959 .000 > mess 9.524 1 9.524 3.252 .080 > hurr 28.048 2 14.024 4.789 .014 > mess * hurr 7.476 2 3.738 1.276 .291 > Error 105.429 36 2.929 > Total 288.000 42 > Corrected Total150.476 41 > > I then wondered what a simple effects analysis would show even though > the interaction does not appear to be significant, p = .291, given > interaction terms are notoriously insensitive to non-cross-over > interactions. The simple effect of message at different levels of hurry > produced a significant difference for the No Hurry condition, p = .025. > > manova help by mess(1 2) hurr(1 3) > /design hurr mess w hurr(1) mess w hurr(2) mess w hurr(3). > > Source of Variation SS DF MS F Sig of F > WITHIN+RESIDUAL 105.43 36 2.93 > HURR 28.05 2 14.02 4.79 .014 > MESS W HURR(1) 16.07 1 16.07 5.49 .025 > MESS W HURR(2) .29 1 .29 .10 .757 > MESS W HURR(3) .64 1 .64 .22 .642 > > And the simple effect of hurry at each level of message produced only a > significant simple effect for the helping message condition, which was > due to the linear decrease in helping, but no significant effect for the > no hurry condition. > > manova help by mess(1 2) hurr(1 3) /print = sign(single) > /contrast(hurr) = polynomial > /design mess hurr w mess(1) hurr w mess(2). > > Source of Variation SS DF MS F Sig of F > WITHIN+RESIDUAL 105.43 36 2.93 > MESS 9.52 1 9.52 3.25 .080 > HURR W MESS(1) 29.43 2 14.71 5.02 .012 > 1ST Parameter 28.57 1 28.57 9.76 .004 > 2ND Parameter .86 1 .86 .29 .592 > HURR W MESS(2) 6.10 2 3.05 1.04 .364 > 1ST Parameter 4.57 1 4.57 1.56 .220 > 2ND Parameter 1.52 1 1.52 .52 .475 > > These (hypothetical) results are probably more interesting than the > original conclusions. The interaction demonstrates the power of the > situation (urge to hurry) to nullify the priming effect of having > thought about the Good Samaritan. > > I now have another good example to illustrate the analysis of factorial > studies, but the more fundamental question, as Paul notes, is whether > any researchers have replicated this study and what they found. > > Take care > Jim > > > > > James M. Clark > Professor & Chair of Psychology > 204-786-9757 > 204-774-4134 Fax > [email protected] > >>>> Paul C Bernhardt <[email protected]> 31-Oct-12 9:58:13 PM >>>> > So, back to work finally. I'm preparing my lecture for Social > Psychology tomorrow. Revising my old Helping lecture to fit only one day > because of lost days to the storm. I'm reviewing in detail on the Darley > and Batson study from 1973 that showed seminary students became unlikely > to help an apparently ill stranger when put under time pressure, even if > they'd just written a sermon based on the parable of the Good Samaritan. > I wanted some details so I found the published study and wow... wow* so > much to question to my eye: erroneously applied randomization of > procedures making for unequal Ns and no indication of what the Ns were > in the conditions, weak statistical methods (though possibly state of > the art for the time). > > One of the main findings is that preparing the sermon on the parable > did not increase helping compared to creating a sermon on another topic. > But, looking at the results, it looked to me like there was evidence for > helping having increased for those who prepared the Good Samaritan > sermon. And, I found a later critique suggesting the analysis looked > wrong. This is a very popular study to teach in undergraduate classes > and it doesn't fit well, IMO, with the large body of work supporting > priming a cognitive schema, in this case for helping (the type of sermon > written should have done that). > > Given the level of importance this is given in some textbooks, has > there been any replications that would support the lack of finding an > effect for activation of a helping schema? > > You might enjoy reading the study. > > http://www.ssc.wisc.edu/~omirosa/357/Readings/13-Darley_and_Batson.pdf > > > Paul > --- > You are currently subscribed to tips as: [email protected] . > To unsubscribe click here: > http://fsulist.frostburg.edu/u?id=13251.645f86b5cec4da0a56ffea7a891720c9&n=T&l=tips&o=21410 > > or send a blank email to > leave-21410-13251.645f86b5cec4da0a56ffea7a89172...@fsulist.frostburg.edu > > > --- > You are currently subscribed to tips as: [email protected]. > To unsubscribe click here: > http://fsulist.frostburg.edu/u?id=13441.4e79e96ebb5671bdb50111f18f263003&n=T&l=tips&o=21418 > or send a blank email to > leave-21418-13441.4e79e96ebb5671bdb50111f18f263...@fsulist.frostburg.edu<Jim > Clark.vcf> --- You are currently subscribed to tips as: [email protected]. To unsubscribe click here: http://fsulist.frostburg.edu/u?id=13090.68da6e6e5325aa33287ff385b70df5d5&n=T&l=tips&o=21435 or send a blank email to leave-21435-13090.68da6e6e5325aa33287ff385b70df...@fsulist.frostburg.edu
