Hi PSPP is simmilar to SPSS just free, you could also use R programming which is also free but technical. I manually code, works best with broadcast print and online Media
Hi, Catalina. You can use Meltwater for media monitoring. Although it's not free, the company does have a university program that you can sign up for and then you can obtain free access and training.
From the student side, I use it in my strategic communications research class (audience insights and analysis) as well as in my campaigns course. I am also in the process of using it in my own research, as it has a sentiment analysis widget.
Of course, I'd suggest cross checking a sample of any computerized sentiment analysis with a sample of human-coded articles, as AI has been shown to have some difficulty with tone or sentiment. I've also heard of people using Crimson Hexagon; however, I'm not sure of the cost.
Meltwater is expensive and they dont offer content analysis only sentiment analysis, we use Meltwater in combination with other media monitoring service providers to track media, then we do the content analysis manually. This is a extreamly tedious job, espesially if you have masses of data to analyze. There is a product called Trax simmilar to Meltwater with more focus on content analysis. The trick is that neither of the tools can analyse media that you have tracked yourself, you can only analyse data that they have tracked. If its possible I can share a project that I did on 6000 + clippings, print broadcast and online. I am also busy with a paper on a facebook page that I analyzed, and did a critical discourse analysis on. I can also share this at some point if you would be interested.
As previously indicated, I use Meltwater for free as a part of my university, which is why I suggested it. There are always different problems with using any automated content OR sentiment analysis program, as noted. Deciding which automated program to use would depend on the variables of the study, I suppose, which were not provided. Regardless, I've tended to rely on human coding, which, as you know, can take an extremely long time and can certainly cause researchers to limit samples sizes. I've found that constructed week sampling is effective, and empirically supported, but again it depends on the scope of the study. I'll check out Trax as well.