Starting with authored books, we have a positive correlation (0.387), which may have been expected (while being aware that one could differentiate further, eg, according to publishers). Then, more controversially, what about book chapters and journal articles:
OK, that’s now really a surprise: book chapters have a positive correlation (0.397) whereas journal articles a negative one (-0.459). As far as I’m aware that’s against the practice in most universities which see book chapters as less reliable submissions since typically not peer-reviewed. I also checked a few universities and indeed some of the top universities have less than 50% journal articles, whereas some of the universities at the bottom have 70, 80 or 90%.
But, then, one needs to distinguish further: eg, in my previous post I had shown that long articles perform clearly better than shorter ones. Thus, the following distinguishes between journal articles with 22 pages or more (let’s call them ‘proper articles’), and those with 21 pages or less (let’s call them ‘notes’) – again as percentage of all outputs (books, book chapters, proper articles and notes).
Now, the ‘proper articles’ have a clear positive relationship (0.51) whereas the ‘notes’ a very clear negative on (-0.79). But, then, finally, we also have to distinguish between short and long book chapters:
Here too long ‘proper chapters’ have a positive relationship (0.54) whereas shorter ones a (here, slightly) negative one (-0.014). So the apparent result if one wants to rank everything is: (1) Proper (long) journal articles and book chapter perform both very well – with hardly any difference; (2) books perform well too but just a bit behind; (3) then shorter book chapters; (4) finally, the worst performers are short journal articles.