SPSS Rozelle Campbell
Locked
 
|
Hi everyone!
I am writing on my master thesis and I have got a few questions. I will compare the data of twitter and online newspapers within 14 days. My variable is "sentiment" (how positive/negative is an article or tweet). Now, I have way more tweets than articles. I am wondering now, how to compare it best. Shall I take the mean value of each day and work out a regression? Is there a more specific way? Also, I want to measure, if twitter and online newspapers influence each other. Now for this, I thought I will use the Rozelle Campbell baseline formula, but now my question is: how can I execute that in SPSS? Is that even possible? If not, can I do it in Excel? Thank you so much for any help, and happy Easter! |
|
|
Hi Patrizia, I've got some questions for you to help clarify what the nature of the data is that you're dealing with and what you're trying to accomplish. First, what exactly is your research question(s)? This/ these should guide your analysis plan. Next, when you say you have
"way more" tweets than newspaper articles, what is the exact quantity you have of each a) in total, and b) for each of your 14 days?Next, you said you are measuring "sentiment." How are you determining the extent to which each record positive or negative? What type/form of "score" are you giving them, and what is the lower and upper bounds of the possible range of scores (e.g., 0-9, unbounded). On Sun, Apr 5, 2015 at 2:18 AM, Ineedhelp <[hidden email]> wrote: Hi everyone! -- Brandi L. Smith Good Lighting Practices Fellow PhD Candidate Parks, Recreation and Tourism Management Clemson University (205) 266-1258 [hidden email] |
|
|
Hi Brandi,
Thanks you for your reply. I have two research questions: 1. Is the sentiment on Twitter and Online-News similar? 2. Can the sentiment on one media-type be predicted by the other media-type? I used the LIWC dictionary to measure the sentiment. So I have a value between 0 and 0.5 (unbounded) for each content. This value refers always to either being positiv OR negative. To create my variable, I subtracted the negative value from the positiv value. So I have only one value for each content. I have 3916 Tweets and 926 articles of 14 days and they are spread evenly over the 14 days. Kind Regards Patrizia |
|
|
Hi Patrizia newspaper articles (i.e. the printed pages) are published on a daily basis, whereas tweets have a timestamp of minutes. Your unit of analysis would be day (N=14), aggregated newspaper sentiment compared to aggregated Twitter sentiment. To further refine your analysis you could do the analysis on a title by title analysis.Still, N=14 is too small for statistical analysis. That could solve the small n. However, particularly in the case of the minute by minute analysis this would result in a newspaper article varialke which would become enormously skewed. So, aggregating at the level of hour would, I think, be the best option. On Mon, Apr 6, 2015 at 11:50 AM, Ineedhelp <[hidden email]> wrote: Hi Brandi, -- ________________________________________________ Maurice Vergeer To contact me, see http://mauricevergeer.nl/node/5 To see my publications, see http://mauricevergeer.nl/node/1 ________________________________________________ |
|
|
Hi,
Thanks for that hint. I am working now with a bigger sample and the regression is significant. Now, as a next step, I want to measure, if one media-type influences the other. So I created two periods in time. I can correlate the data like that: time1 twitter/time2 online-news time 1 online-news/time2 twitter and then see what correlation is higher. To see if it's significant, I want to use the Rozelle Campbell baseline. But as I have never learned how to run this formula, I am a bit lost. Does anybody know, if it works on SPSS, OR how to work it in Excel? Thanks so much for any help. Kind Regards |
Administrator
|
See CCF command.
I'm too lazy to look up 'Rozelle Campbell baseline'. Maybe you should provide the formula or an online free link to the substance of the matter? Pretty much anything you need to compute can be done in SPSS. If you want to know how to do it in Excel, ask in an Excel forum? --
Please reply to the list and not to my personal email.
Those desiring my consulting or training services please feel free to email me. --- "Nolite dare sanctum canibus neque mittatis margaritas vestras ante porcos ne forte conculcent eas pedibus suis." Cum es damnatorum possederunt porcos iens ut salire off sanguinum cliff in abyssum?" |
|
|
In reply to this post by Ineedhelp
Having never seen this sort of analysis before I'm curious how you are going to work the computations for the correlations. Are you averaging each days twitter sentiment scores and each days online media scores so that you get 14 records with a twitter and a media score for each day? Then how are you defining time 1 and time 2? Time 1 = days 1-7; time 2 = days 8=14, so that the N is 7 for the correlations?
Assuming you have only four correlations, the actual computation is most easily done by hand. You have the formula; you have a calculator. If you had 'many' sets of four correlations, the thing that I would do would be to edit each correlation matrix, i.e., correlations twitter1 twitter2 media1 media1, down to a single record consisting of the six correlations (4*3/2) and reading them back in using Data list (of course, you could put the correlations excel and read them back in from there). Once in: Compute RC=(PX1Y1+ PX2Y2)/2)*sqrt(((PX1X2)**2+ (PY1Y2)**2)/2). Gene Maguin -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Ineedhelp Sent: Tuesday, April 07, 2015 4:22 AM To: [hidden email] Subject: Re: SPSS Rozelle Campbell Hi, Thanks for that hint. I am working now with a bigger sample and the regression is significant. Now, as a next step, I want to measure, if one media-type influences the other. So I created two periods in time. I can correlate the data like that: time1 twitter/time2 online-news time 1 online-news/time2 twitter and then see what correlation is higher. To see if it's significant, I want to use the Rozelle Campbell baseline. But as I have never learned how to run this formula, I am a bit lost. Does anybody know, if it works on SPSS, OR how to work it in Excel? Thanks so much for any help. Kind Regards -- View this message in context: http://spssx-discussion.1045642.n5.nabble.com/SPSS-Rozelle-Campbell-tp5729152p5729159.html Sent from the SPSSX Discussion mailing list archive at Nabble.com. ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD |
|
|
Hi Gene,
This sort of analysis is often used for measuring the Agenda Setting effects in media. In case you are interested, here is an example: Lim, J. (2011). First-level and second-level intermedia agenda-setting among major news websites. Asian Journal of Communication, 21(2), 167–185. doi:10.1080/01292986.2010.539300 Yes that's how I will do it, but with more data, I have now data from over three month, so N is much bigger. You are right, I should simply edit each of the RC formula and correlations. Thanks for that, Kind Regards |
«
Return to SPSSX Discussion
|
1 view|%1 views
| Free forum by Nabble | Edit this page |