Scientific methodology

Using similar arguments as against knowledge some groups try to question scientific methodology. Because scientific methodology has a high reputation, various groups that are not active within such methodology may be tempted to claim that they also are active within "science".

A citation often heard is "Several types of Science exists":

We need to acknowledge that there is more than one version of science

Ross - Science Wars, Introduction, Social Text 46/47 (1996) p.12.

Such citations remind of advertising hotel rooms with a view "towards the sea" and where the view is blocked by other hotels. It sounds nice but is actually erroneous.

There are several sciences but only one kind of Science

Credible

As with all other belief systems, a scientific area is founded on a network of coupled arguments.

The credibility of these arguments forms the foundation of the development that is seen as soon a scientific methodology begins to dominate within an area.

It seems almost laughable that, while every other science makes continuous progress, metaphysics, which desires to be wisdom itself ... perpetually turns round on the same spot without coming a step further.

Kant 1783 - Prolegomena, Preface 4:256 (Hatfield 2004, Cambridge p.5-6)..

Different branches within the academic world apply scientific methodology at varying degrees.

One reason may be the problems of obtaining very credible premises as foundation for their arguments.

In USA History, Politics, Economy, Sociology, Geography and Anthropology are therefore not denoted as sciences, but as "Social studies". [Mirriam].

Fake Science

A huge and growing problem is formed by a great number of new apparently scientific journals.

They publish articles after payment and without examination of the content [Eriksson, Beall].

Unscientific

The definition of science, above, requires a close definition of the term "scientific methodology". It is not completely clear, but it yet implies that some activities may be termed as clearly unscientific:

- Areas where the argumentation starts from not verified starting points and where criticism of these hypotheses is not accepted.

- Observations that are not described in such a detail that they may be reproduced.

- Hypotheses that are created without carefully described and reproducible observations.

Science is founded on empiricism. This means that results and hypotheses are based on perception and represent probability arguments that are more or less trustworthy.

When many independent observations all verify the same hypothesis, and in addition no exceptions are reported, the credibility becomes very high.

In a mature scientific area many observations are moreover coupled, which results in that relations between them become even more trustworthy. Such relations are called natural laws.

But the results never become "absolutely certain" - no argument will ever become certain at such a fictitious level. At best, they become "very probable" where the term "very" often implies VERY.

Because scientific methodology encourages criticism, change and refinement, its accepted arguments are generally more trustworthy than arguments within dogmatic belief systems.

Several signs indicate whether a publication follows scientific methodology or not:

- Does it contain an account of earlier important contributions within the area?

- Does it describe observations in such detail that they may be reproduced?

- Does it give references to included statements so that they may be controlled?

In addition a common sign may indicate whether the author itself considers the publication as scientific:

A publication that follows scientific methodology virtually never mentions that it "is scientific". It presents in detail what was done, how it was done, the results of the investigation, and it often includes a separate discussion about the results.

As opposed to this, unscientific presentations may be saturated by:

- claims that they are scientific.

- discussions regarding opinions about scientific methodology.

- results based on observations that are not described in such a manner that they may be reproduced and then provide either supporting or disapproving results.

A survey that is not carefully described, with regard to selection, questionnaire and the experiences of the questioned during the survey, gives an uncertain result.

This have been seen during voter's surveys when parts of the population were excluded and where conservative groups got surprising influence (Sweden Riksdag election 2014, USA president election 2016).

Researchers following scientific methodology may of course commit mistakes, and it is not always that they are directly discovered and corrected by colleagues or referees.

Cause and Effect

A very common mistake is the interpretation of cause and effect.

A common interpretation is to assign a cause or reason when an effect or result is observed.

Problems with causality have been known for a long time [Hume]. They may e.g. emerge when two phenomena co-vary (are correlated), i.e. that they and that this correlation is caused by that both phenomena are influenced by a third, not studied, phenomenon.

It is then all too easy to assign one of the phenomena as the cause and the other as the effect, i.e. as a result of the assigned cause.

Such correlations are said to be collateral.

Collateral correlations

An invented (for illustration) and an actual example of the problem of cause and effect is given below:

Suppose researchers perform a cohort study covering the whole Swedish population, and among other factors note which are playing computer games.

Ten years later they analyse the frequency of deaths in cancer. They find that deaths were quite uncommon among computer players and draw the conclusion that computer games protects against cancer.

Television, radio, newspapers and blogs with advertisements report the results from this scientific study and companies rapidly begin to sell computer games that "cure cancer".

In this invented case, it is obvious that the result was formed by the normally low age of computer players together with that cancer normally becomes more frequent at higher age. "Age" was hence a third, not studied, factor.

A real example:

A cohort study based on 400 000 persons analysed whether coffee results in decreased or increased death rate. The initial result was that coffee resulted in increased death rate in almost every serious disease.

But as it is well known that coffee consumption is correlated to smoking, the influence of smoking was eliminated using statistical methods.

The result then became the opposite: Coffee gave reduced death rate and was beneficial against virtually all serious diseases! [Freedman]

But in a commentary (Letter) to the article Scott K. Aberegg discussed a plausible reason for the result:

We drink a lot of coffee at work, and the result is probably caused by a well-known correlation between coffee consumption and "healthy worker" [McMichael, Porta]. The result hence shows that persons that drink a lot of coffee are healthy enough to have a job.

The article and the commentary together demonstrate the great advantage of scientific methodology:

Publication of results and hypotheses, together with public criticism of the publications, result in that the credibility of arguments that concern our reality constantly evolves.