If you have concerns about a manuscript you want to prepare for a journal in the field, why not e-mail someone on the editorial board for advise? That is a duty the editors have.

Michael it's not about manuscript but I just want to know, if we can explain our observations then why stats?

The formal collection of data using proper randomization methods helps insure that our personal biases have minimal effect of the data, results, and interpretation. You might observe some unique event while doing an experiment, but the observation should be subjected to experimental verification to make sure that what you think you saw was actually taking place.

Statistics is not a sufficient criteria to make research good or bad. If replication is insufficient or the experimental design poor then no quality of statistics will save the study. On the other hand there are very few cases where a well designed experiment with proper replication will be relevant without a rigorous statistical analysis. That said, simply using the most complex analysis is not a good path to success (or at least I hope not). Calculating means and standard deviations is still statistics.

While difficult to learn, R is free. It is also capable of any analysis that you desire and it is available to all for the same low price.

If you only provide a description, then there is nothing that I can do with your results. I need to be able to find similar research and then combine multiple studies. It was never part of the scientific method to do one experiment and uncover universal truths. Experimental results need to be corroborated by other scientists and that is much harder if all we have is a description.

There are grey areas. For example, statistics may play a minor role in the description of a new species, or statistics may play a major role to place the new species in context with closely related species. Describing an ecosystem in a remote cave system might not include much statistical analysis. Things like case studies in psychology are also entirely (or nearly so) descriptive.

Choose your journal wisely. Why try publishing an article best suited for National Geographic in the journal Ecology? You need to do the type of research that the journal is likely to publish.

That said, sometimes things are a bit stupid. I have two means and they are clearly different by over 10 standard deviations. Will a p-value really tell you more than what you would get from a box plot of the means, median quartiles and outliers (if any)? Of course without rule like "all data must be subjected to an appropriate statistical analysis" there will be people who try to take short cuts.

In a botanical survey, it is nice to generate species lists. However, it is likely that your survey methods missed a few species here and there. Given your sampling protocol, can you estimate the number of species that you missed? If I have to make a management decision, how do I decide if I save the ten hectares in valley 1, versus the 20 hectares in valley 2 given that valley 1 had 200 species of plant and I took 50 samples and valley 2 had 127 species of plant and I took 30 samples. Which one do I save if biodiversity is my criteria? There are many things that you can do with survey data depending on how you sampled. Island biogeography problems are interesting, as are phenology issues. However, such research takes a great deal of effort. Quite a few journals will publish species lists, especially if you can do something more that list all the species that I found in valley 1, and the next paper has all the species from valley 2. You can also include a species list in supplimental information if it does not fit into a regular article.

Hi,

This question crossed my mind these days. It seems that if you do not add some statistics to your analysis, the journal's editor think that your methodology is too basic. In my opinion, the fact that so much stat is required demonstrates the shortage the exists in sound theories. When one has a sound theory, a basic methodology will suffice to explain it. This of course depends on the discipline. In several research areas statistics is needed more than in others.

Thanks Timothy for ur kind explanation

Thanks avishag