If you refer to statistical inference in their more classic form (i.e. the frequentist parametric paradigm), then you can think of several alternatives such as non-parametric statistics, Bayesian statistics, resampling, heuristics and some others. I prefer to see them as complementary approaches.

Statistical Inference is based on sound scientific reasoning.Alternative?Probably not.

Following

I do not see any, but many statistical tools are there and you can discuss with someone to select what suits your work

In social and human sciences , statistical inferencing has long been regarded as a legitimate tool in analyzing and interpreting the results. As things stand right now, nothing can replace the objectivity and utility of statistics yet.

Question is not clear to me.

If you refer to statistical inference in their more classic form (i.e. the frequentist parametric paradigm), then you can think of several alternatives such as non-parametric statistics, Bayesian statistics, resampling, heuristics and some others. I prefer to see them as complementary approaches.

Any consistent way of making decisions based on a set of data is equivalent to Bayesian statistical inference. This is the content of the Dutch Book Theorem and several related theorems. The alternative to Bayesian inference is therefore inconsistent inference. Obviously there are some assumptions that should be fulfulled in order to make this conclusion, but these assumptions are surprisingly mild.

Although I don't understand the exact thing what you want to say, I guess something important and fundamental issue you are posing.

While some argue statistics offers formal and genuine "scientific" inferences, it is, in its nature, inductive; comes with many limitations.

There might be cases I don't need a statistical methodology to verify some claim. Clarifying the scope and limitations of statistics would bring me deep insights about my inferences.

the selection of statistical tool depends on the nature of your problem. Baysian analysis, Stochastic process, Morkov analysis are some tools to find the appropriate solution of your problem.

You can use Bayesian methods (instead of frequentist hypothesis testing) to find the distribution of the unknown parameters.

Instead of trying to reject the notion that μ=0, you can actually construct a distribution for mu if you're willing to subscribe to Bayesian ideology and say that an unknown parameter can have a distribution that reflects one's uncertainty about the value.

I am not sure I understand the question. From my experience one has data and uses appropriate procedures -- statistics, OR, basic math, whatever -- to answer the question, solve the problem, understand what the situation...

some people make stuff, politicians seem to prefer this method

Statistics has no alternatives when there is uncertainty and chance in the phenomena that you study.

I do not see alternatives. If used wisely and correctly statistics is a good tool. However as pointed out politicians and others can abuse it to their own advantage.

The best alternative to statistics is statistics itself.There is nothing better than statistics!

guessing? divination? reading tea leaves?

Within Statistics itself, the major alternatives to "Statistical Inference" are:

(a) complete enumeration of the population, so that no "inference" is necessary. By its nature, this rules out making any inferences about future, changed versions of the population.

(b) "Descriptive Statistics" and "Statistical Summaries", so that you just report the facts, or a clear, easily understood summary of them. This can include "Statistical Graphics".

Outside of Statistics, you may find things like "Big Data", "Automatic Learning", "Fuzzy Logic", etc., where some things are based on ideas developed/accepted within Statistics but other things may not be.

bro, alternatives are nature based newly developed heuristics. Mudassar Rauf & Jabir Mumtaz what do you say?

Dear Saad

Read the paper " Bayesian statistics in medical research: an intuitive alternative to conventional data analysis "

Statistics is essential and unavoidable analytical tool of scientific research and investigation.

I do strongly agree with the comment

"Statistics has no alternatives when there is uncertainty and chance in the phenomena that you study."

provided by Mohamed I. Riffi.

My view on statistics is also same with that due to Florence Nightingale whose opinion regarding statistics is as follows:

" Statistics is

the most important science in the whole world:

for upon it depends the practical application of every other

science and every art: the one science essential to all political

and social administration, all education, all organization based on

experience, for it only gives results of our experience.

Statistics exists everywhere, everyday and with everybody. Everybody uses statistics everyday knowingly or unknowingly.

Concept of Homogeneity in Statistics is equivalent to the concept of Equality in Mathematics.

Thanks for the replies. HOW MANY of you are, by the way, familiar with Fuzzy Theory, Grey System Theory and Rough Set Theory and their ability to analyse data when traditional statistical methods doesn't sound appropriate?

"Traditional statistical methods" are always appropriate.The traditional statistical method is to start by thinking about one's problem and one's objectives and to not expect the standard statistical software packages to do everything for one.

If faced by a "too difficult" problem, appropriate things to do are:

(a) find the time and money to involve someone with better and more relevant statistical expertise;

(b) seek a pragmatic solution that takes reasonable account of what one does know about the problem .... this might sensibly include some of these "grey" methodologies.

The "alternative" approaches you mention are essentially attempts to deal with a lack of information and resources to deal with things. Given that, if one can fully describe all the uncertainties involved, a "Bayesian" solution is always best, the questions about using alternatives are:

(i) how much is lost by not undertaking all the work in implementing a fully Bayesian solution, compared with just implementing something ad-hoc;

(ii) how can any given "alternative" approach be made to better align to the Bayesian "ideal", in order to improve its performance.

The theory of conventional / traditional statistics is based on the vital philosophy of statistics. This philosophy is found more or less weak in the other allied disciplines. Accordingly, no other discipline can be treated as genuine alternative of conventional / traditional statistics.

Statistics is the science which deals with the collection, analysis and interpretation of numerical data.

However, statisticians have no field of their own from which to harvest their data while the data of other sciences come chiefly from their own disciplines.

Statistics can be interpreted as reality handling mathematics.

Thank you so much for your valuable and interesting answers.

new question: why does dhritikesh chakrabarty - whose RG score consists to 23.37% of answers, btw. - split his extremely insight- and helpful contribution into 8 different answers? please don't misuse the forum to boost your scores! respect the principles of scientific integrity also on researchgate. and please refrain from posting spam, i.e. don't post anything if you haven't got anything important to say. thank you!

Here is a link to a brief open-access article which introduces the ONLY statistical paradigm which EXPLICITLY maximizes (weighted) overall predictive accuracy, as well as (weighted) predictive accuracy normed against chance.

Article What is Optimal Data Analysis?

Thomas Grischott

With respect to dhritikesh artificially boosting his RG score, read the 'Other' section of this article lol

https://en.wikipedia.org/wiki/List_of_scientific_misconduct_incidents#Other