Scientific Report references I'm taking a course about "Scientific Report", where I need to write reports about statistics. 
I have many difficulties to elaborate scientific texts, I can not develop the theory and presentation of results is horrible. I understand that writing a scientific text may be like, write any other kind of text, where you will honing your writing as you practice, and that reading is also key to the development of writing.
Is there any reference that deals with writing scientific texts? As you improved your reports?
 A: There's no better way than reading good papers. Go to Google Scholar or arXiv where you can find any kind of scientific paper. The problem is that you may not know which ones are good ones. So, look up the most cited papers, they're usually good. For instance, if you search for "linear regression" in Google Scholar one of the top hits is "Leggetter, Christopher J., and Philip C. Woodland. "Maximum likelihood linear regression for speaker adaptation of continuous density hidden Markov models." Computer Speech & Language 9.2 (1995): 171-185." with currently 2690 citations. I haven't read the paper but it must be good if 2690 papers cited it.
Once you get a feel of how good papers look like, you may look for resources which suggest you how to write and format papers. There's a ton of them out there. I'd recommend reading this paper: Mathematical Writing by Donald E. Knuth, Tracy Larrabee, and Paul M. Roberts. It's short and covers a lot of ground. Knuth is the author of TeX system, which is the most popular scientific typesetting software. He wrote a lot of great papers and books.
A: Last year a colleague of mine suggested:
The Craft of Scientific Writing by Michael Alley
http://www.amazon.com/The-Craft-Scientific-Writing-Edition/dp/0387947663
Moreover, I think you should start reading scientific papers on a specific topic, so that you can understand how to write a good manuscript.
A: The other answers are very helpful; just a couple of things to add that are a bit more than a comment.
First, if part of the problem is that you are having trouble getting free (unpaid) access to papers, peer-reviewed journal articles based on work supported by the US National Institutes of Health (NIH) must now be made freely available within 12 months of publication. NIH supports some biostatistics research and a lot of applied work, so you can find such papers via the PubMed website. I just did a search there for "regression methods," sorted by "Relevance," and the top hit was a comparison of penalized regression methods that is available as a "Free PMC Article." Haven't read the paper so I won't comment on quality, but you do need to read some lower-quality papers in any event to appreciate the differences. So add PubMed to Google Scholar or arXiv as recommended by @Aksakal.
Second, consider trying to answer more questions on this site (if you have a strong interest in statistics) or on another StackExchange site if another site is more aligned with your interests. I've already written a fair number of scientific papers, yet I've found that the discipline developed by trying to write brief, intelligible answers here still helps me a lot.
