There isn’t a straight line of demarcation between science and pseudoscience (PS), which is universally applicable in all fields categorized as scientific. A general guide for demarcating between the two is that the theory should have observable evidence, provides predictions, uses non-controversial reasoning, and is repeatable. These are simply guidelines and do not necessarily count as criteria for disqualifying a theory if all aren’t met because some are simply untenable depending on the field in which they are applied. Falsification is not necessary for a scientific theory but it does help substantiate the theory as a robust scientific theory.
When considering the criterion of observable evidence I make the distinction between observation and what is empirical. Something may be observed and qualify as evidence even though it’s not related to material causes. This is where the distinction between Duhemian science and Augustinian science must be made. I would deny the use of Duhemian science. This method, or philosophy, has a goal of stripping science from all metaphysical imports. Augustinian science is open to metaphysical presuppositions with science. In the mid 1800’s William Whewell was the first to restrict science to only mean natural science. Pierre Duhem followed this idea and constructed a methodology, which barred explanations to material causes. For instance, agent causation is completely compatible with Augustinian science but is prohibited as a scientific explanation in Duhemian science. Agent causation is something that can be observed but isn’t necessarily reductionistic in the material sense as with material causation because agent causation has metaphysical import.
Concerning the Duhemian science, this methodology is incapable of being achieved across disciplines. This would, by definition, remove psychological, sociological, and historical disciplines from being scientific since all require agent causation in explanation. Thus, the constraints of Duhemian science are untenable across the spectrum of [scientific] disciplines.
A scientific theory must be put in empirical harms way even though empirical evidence is not a necessary condition for a theory to be scientific. Scientific theories must be supported by facts and empirical evidence is certainly a sufficient condition for science and is preferred depending on the discipline. If the theory is based on facts then it should be able to withstand scrutiny when compared to with relevant facts that could potentially falsify the theory. Contrary to Karl Popper’s line of demarcation between science and PS, a theory absent of evidence is not scientific—it’s mere speculation. Popper’s model allows for a theory to be scientific prior to and supported evidence. There’s no positive case for purporting a theory under his model. There can only be a negative case to falsify it and as long as it may be potentially falsified it’s scientific. Thus, a scientific theory could have not evidence or substantiated facts to provide good reasons for why it may be true.
I agree with Lakatos when he states that falsifiability is not the solution to the problem of demarcation in science. Falsifiability is not a necessary condition for a scientific theory because it doesn’t meet it’s own requirements. Thus, falsifiability is a sufficient condition for a scientific theory and should be preferred. I believe a theory can have an anomaly in it and it’s completely acceptable for a theory to account for the anomaly. If a theory accounts for so many anomalies that the probability of the theory still being correct in light of these problems then the theory should be rendered inert. The theory needs to be abandoned as an appropriate scientific theory because at this point it’s probably a completely different theory than it had been originally. It should be reworked accounting for the, what were, anomalies as actual evidence for the theory.
Scientific theories should make predictions. This is a simple consequent to observation. If the theory is based on observed evidence, which is applicable to past observation and present observation than, if true, it should be observed in the future.
The reasoning process should be uncontroversial. Using empirical evidence is commonly accepted because it’s readily accessible to be observed by anyone. The interpretation of the data is where this is similar to the criterion of observability. This will determine if what is being observed is evidence for a theory of if it is not. Deductive reasoning may be used in scientific explanation but inductive or abductive reasoning is most preferred. Abductive reasoning is critical for the historical sciences. This serves well to the distinction between material causation and agent causation and why agent causation falls under observability. With reasoning by the inference to the best explanation the best explanation may be an agent. Whatever the best explanation is it cannot be contrived ad hoc or post hoc but must be related to the evidence and must follow from the evidence. When using non-deductive reasoning this allows for the explanans is not reached under logical control from fixed premises and that it follows from the apprehension of the data.
This follows contrary to Kant’s a priori fixed categories. Under Kantian categories one cannot know the Ding an Sich but the mind imposes its categories on to the explanandum. The explanans should explain causation as having more than just Hume’s causality where causality is deprived of being foundational in making necessary connections between actual events and leaving it with nothing more reliable than habits of mind rooted in association. This is important for historical sciences. Uniformitarianism, broadly speaking, must rely on the regularity of causality—presently known causes to phenomena may be inferred by the perceived effects are efficacious retroactively. The explanandum should be a posteriori as well as the explanans. A priori notions can be used in the reasoning process (using mathematics, semantic terms, etc.). However, the synthetic a priori should not be in the explanans and if it is contained with the explanandum the explanans cannot account for it (i.e. morality or mathematical proofs if math belongs in that category). Scientific theories must be a posteriori in its explanans because if this theory is to have any epistemic authority because epistemology follows from ontology (theory being epistemology and ontology being the observed data).
How I’m using the explanans and explanandum in relation to Kant’s categories arises the issues of rationalism and empiricism. Empiricism focuses on analytic a priori truths and synthetic a posteriori truths. I tend to be more of a rationalist in my approach. I do caution using the synthetic a priori in the explanandum.
Scientific theories should not have religious import. This may certain have tensions with Augustinian science but religious import is much more than mere metaphysical import. Religious belief imports something that is considered to be internally authoritative (as in within that system of belief). The applicability of some of the beliefs may be universal but using religious belief as a grid for interpreting what is and what is not science is methodologically irresponsible. Religious belief is not itself scientific but may have scientific beliefs and in sync with science. There’s a categorical difference.
For instance, using Scripture to interpret science or empirical data is circular in its reasoning. Scripture would already have the conclusion and then uses the reasoning process to conclude with that Scripture may be advocating. Since I’m coming from a religious perspective I would argue that science and Scripture are harmonious and congruent. It’s necessary to have a scientific understanding of nature and agency prior to interpreting Scripture. In order to know a miracle has happened one must know that sea water is less dense than the human body, or that water doesn’t normally undergo chemical reactions to become fermented wine, or that dead bodies don’t normally undergo a natural biological resuscitation or resurrection. If creation science is an actual science then the antecedent conditions must be subject to scientific scrutiny. In other words, in order for creation science to be considered a scientific theory it must meet all the appropriate conditions I mentioned earlier. However, it’s not that creation science does or does not meet all that criteria it’s that it’s dependent on an antecedent interpretation of Scripture. Though I’m not as militant as Michael Ruse when classifying creation science as pseudoscience I do believe it falls short of being a scientific theory. Additionally, what makes creation science so unattractive is that it is completely void of the possibility of being falsifiable unless the antecedent conditions (the interpretation) have been falsified. This makes the issue of accounting for anomalies so absurd that creation science doesn’t really account for anomalies; rather, it produces extreme ad hoc explanations to account for contradictions to it’s theory. Lakatos includes this distinction between anomalies and refutations. Refutations are falsifiers. Additionally, scientific theories are true regardless of any religious understanding. Religious belief, like I mentioned earlier, begs the question on certain scientific matters. Religious belief, when used as a hermeneutic for interpreting scientific data and developing scientific theories, is also a controversial methodology. Its appeal to method isn’t necessarily objective (as close to objectivity can be) and is not commonly accepted (though not to be used as an argumentum ad populum). There’s no such thing as creation science–there’s just science.
Repeatability is only applicable to certain categories of [scientific] disciplines. Repeatability is no applicable to historical sciences. The explanandum cannot necessarily always be repeated. Certainly, facets and contributions that bring about the explanandum could be repeated (i.e. what would happen in sedimentary formation in geology). Astrophysics cannot be repeated but can be continually observed. We can understand that stars uses lighter elements for fuel and then proceed to consume up to iron but this is a process that cannot be repeated. Also, galactic formation cannot be repeated. Repeatability is appropriate in laboratory sciences like biology and chemistry and may be used in social sciences on repeating certain experiments on subjects.
With regards to a litmus test to identify science from PS the line of demarcation is straight and strong. The logical positivists made contributions to the methodology of science with their litmus test. Kurt Gödel’s proofs required that the positivist use a priori truths to explain away a priori truths. Their attempt was certainly admirable for science but their demarcation was too rigid and untenable. I would argue that something should be classified as pseudoscience if it contradicts any of the above criteria. There’s a difference between inconsistent or incongruent and contradiction. A theory may have anomalies or unknown ways of falsifiability but if the theory is utterly incapable of being falsified at any point it’s not scientific. A theory can be scientific if it is based on little evidence or observation but if there is none then it’s just as bad as Popperian science and pure speculation. Also, as an example, if a theory uses illogical reasoning then it should not be scientific (it can have controversial reasoning, which is not preferred, but it cannot be fallacious). I don’t think there’s such thing as a pseudo-question because science is not the sole source and means of epistemology. Any other non-scientific questions or theories may be legitimate in the field in which it applies (i.e. theology or ethics).
 See Imre Lakatos, “Science and Pseudoscience” in Philosophy of Science, Eds. Martin Curd and J.A. Cover (New York: Norton, 1998), 22-23.
 Lakatos, 23.
 Michael Ruse, along similar lines, suggests that creationists use the data from non-creationists and warp the data to have a conclusion fitted to the preconceived idea of what it should say. Michael Ruse in “Creation-Science is Not Science” in Philosophy of Science, Eds. Martin Curd and J.A. Cover (New York: Norton, 1998), 43. This circular reasoning is worse than Kant’s use of having fixed premises from his categories by which he used deductive means to arrive at a conclusion. Creation science does not allow for free invention of hypotheses derived from the evidence. This is different from mere invention where science is a human institution, something we invented to organize our experiences and enhance our technological control of nature. Alex Rosenberg, Philosophy of Science ed. 3 (New York: Routledge, 2012), 53. Free invention is more in sync with discovery contra Kant. This free invention could be likened to an exegesis of nature.
 Following the idea presented by Ruse. 43.
 Lakatos, 23.