# *English version* ^213b94 The hypothesis is a central aspect in any type of empirical research (empirical = based on systematic observation). A hypothesis is the expected answer to the [[Research question|research question]], and specifies the relation between (at least) two variables. >*For example:* >Regular endurance training reduces elevated blood pressure. ## Characteristics of a scientific hypothesis For an hypothesis to be classified as a scientific hypothesis in the anglo-saxonian literature, it should fulfill the following criteria: - **Testability;** the respective phenomena can be measured and by principal there should be an observation that contradicts the hypothesis >*For example:* >- Hypothesis 1: the spaghetti monster exists. This is NOT testable, because we don't know how to measure the existence of a spaghetti monster. >- Hypothesis 2: all athletes perform physical exercise. This is NOT testable, because the definition of an athlete implies that the individual is proficient in a certain form of physical exercise. Therefore, by principle the observation of an athlete who dose not perform physical exercise does not exist. >- Hypothesis 3: men are stronger than women. This is testable. Strength can be measured, and by principle women can be stronger than men. * **Parsimony;** the hypothesis should be formulated as simple as possible, to gain testibility. - **Scope;** the hypothesis should not be limited without a clear need for it. Limitation can occure in for example population (men/women vs German men/women) and time. - **Fruitfullness;** the hypothesis should enable the prediction of future events. This is closely linked to the scope of the hypothesis. When the scope is limited to the present or past only, it is not fruitfull because it does not allow for a prediction of a future event. - **Conservatism;** the hypothesis should be based on the current state of knowledge. It should be stressed that the criteria which are mentioned in the anglo-saxonian literature differ from those mentioned in the German literature (discussed in the [[Hypothesis#German version|German version]] of this topic). **Relationship** between criteria according to ANglo-Saxonian ant to the German literature: **Testability** equals the combination of the German "Überprüfbarkeit" (Observability) and "Falsifizierbarkeit" (Falsifiability). For a hypothesis to be testable, the concepts which it addresses must be measurable (observable) and observations which contradict the assumption must by principle be possible because otherwise, the assumption would always be right (tautological). **Scope** corresponds strongly with the characterististic of a hypothesis being "general" according to the German Philosophy of science (see below). Nevetheless, for a hypothesis to be general, the assumption needs to be formulated to hold independently from space and time - an affordance which is typically not met e.g. in social sciences as it refers to humans. Therefore, the affordance that the scope should not be limited without necessity corresponds to the idea, that a hypothesis should be formulated as general as possible. The forth criterion that a hypothesis should express **a relation between at least two variables** is typically not mentioned explicitly in the anglo-saxonian literature but it is implied in the related concept of [[hypothesis testing]]. ## Types of hypotheses ### Directed vs Undirected. In a directed hypothesis a specific direction of the relation between the respective variables is indicated. ^843741 >*For example:* > Drinking coffee before a 10km run improves performance. In an undirected hypothesis, no specific direction of the relation between the respective variables is indicated. ^631dae >*For example:* > Drinking coffee before a 10km run changes performance. ### Specific vs Unspecific A specific hypothesis indicates not only the direction of the relation between variables but also its magnitude. >*For example:* >Drinking coffee before a 10km run improves performance by 2 minutes. An unspecific hypothesis does not indicate a magnitude of change. ### Difference, change and correlational A hypothesis can be divided in [[Statistics/Basic inferential statistics/Change hypothesis|change hypothesis]], [[Statistics/Advanced inferential statistics/Difference hypothesis|difference hypothesis]] and [[Statistics/Basic inferential statistics/Correlational hypothesis|correlational hypothesis]] hypothesis, based on the type of relation between variables that is expected. > *Example difference hypothesis:* > The maximum strength of men is larger compared to women. >*Example change hypothesis:* > Maximum strength increased after 6 weeks of strength training. >*Example correlational hypothesis:* >The larger the muscular cross-sectional area the larger is the maximal strength. ___ ## The empirical content of a hypothesis According to Popper's "The Logic of Scientific Discovery" (first published in 1935 in German language, first English translation published in 1959), the empirical content of a hypothesis corresponds to the number of possible falsificators. The empirical content of a hypothesis cannot be measured in absolute terms but it can be compared between different hypotheses. H1: A -> C meaning if A then C H2: A & B -> C meaning if A and B then C The empirical content of Hypothesis 1 is larger than the empirical content of Hypothesis 2 because for H1, any object with the property A would be a falsificator if it does not have property C. For H2, possible falsificators are only objects with both properties A and B. If A and B are distinct and independent properties, this number of objects will be smaller than the number of objects which simply have property A. H1: A -> C meaning if A then C H3: A -> B & C meaning if A then B and C The empirical content of Hypothesis 1 is smaller than the empirical content of Hypothesis 3 because for H1, any object with the property A would be a falsificator if it does not have property C. For H3, possible falsificators are all objects with property A which do not simultaneously have property B and C. If B and C are distinct and independent properties, this number of objects will be larger than the number of objects which simply have property C, irrespective of their B. H1: A -> C meaning if A then C H4: A | B -> C meaning if A or B then C The empirical content of Hypothesis 1 is smaller than the empirical content of Hypothesis 4 because for H1, only object with the property A would be falsificators if they do not have property C. For H4, possible falsificators are both objects with property A and objects with porperty B do not have property C. If A and B are distinct and independent properties, this number of objects will be larger than solely the number of objects which have property A. >Please be aware that the words "or" and "and" are used in logic and in everyday colloquial language with different meanings. H1: A -> C meaning if A then C H5: A -> B | C meaning if A then B or C The empirical content of Hypothesis 1 is larger than the empirical content of Hypothesis 5 because for H1, any object with the property A would be a falsificators if it does do not have property C. For H5, possible falsificators are only objects with property A which do neither have poperty B nor C. If B and C are distinct and independent properties, this number of objects will be smaller than the number of objects which simply have property B, irrespective of their C. In theory, researchers should try to falsify their hypotheses. The researcher's assumption is built on his/her already available knowledge. In the case, the hypothesis holds when tested, the researcher does not learn anything new about the world. Only in the case of falsification, the researcher learns that his/her assumptions do not appropriately apply to the world and will develop new ones. In real life, researchers are often justificators of their theories and hypotheses instead of searchers for falsifikation. Therefore, researchers sometimes add an "or" clause to the "then" part of their hypothesis or an "and" clause to its "if"-part. The latter sometimes is done when a hypothesis does not hold for a sample under study but holds for a subgroup wihtin this sample. In this case, A & B in a subsequent hypothesis then is a subgroup of A in a former, already falsified hypothesis. When these techniques of extending the "if"- and the "then"-part of a hypothesis are used subsequently on a set of related hypotheses in order to save then from being falsified, this is named "exhaustion" of a hypothesis. ## How to test if a hypothesis holds The task of empirical research is to test the validity of the hypothesis through appropriate methods. In [[Hypothesis testing|hypothesis testing]], researchers detect if an empirical study outcome supports or contradicts the hypothesis. _________________________ ____________________________ --- # *German version* # Hypothese Die Hypothese ist der zentrale Bestandteil jeder Form der empirischen Forschung (empirisch = auf systematischer Beobachtung beruhend). Hypothesen sind vermutete Antworten auf Forschungsfragen. > Beispiel: > *Regelmäßiges Ausdauertraining reduziert Bluthochdruck* ## Hypothesenmerkmale Das grundlegende Kennzeichen von Hypothesen ist, dass sie Aussagen über Beziehungen zwischen mindestens zwei [[Variable |Variablen]] darstellen. Weitere Merkmale sind: - präzise und widerspruchsfreie Formulierung - Falsifizierbarkeit, also die prinzipielle Widerlegbarkeit - [[Operationalization |Operationalisierbarkeit]], was bedeutet, dass die in der Hypothese benannten Konzepte mit den jeweils (historisch und räumlich) gegebenen Mitteln beobachtbar sind und dass damit die Beziehung zwischen den beiden Variablen überprüfbar ist. - universelle Gültigkeit Es gilt zu berücksichtigen, dass sich in der wissenschaftlichen Literatur verschiedene Angaben über die Hypothesenmerkmale finden lassen. Insbesondere zwischen der deutschen und angelsächsischen Literatur gibt es Unterschiede. Die Hypothesenmerkmale der englischsprachigen Literatur können dem englischen Teil dieses Artikel entnommen werden. ![[Arbeitsblatt Hypothesen.pdf]] ## Hypothesenarten Es wird zwischen gerichteten und ungerichteten Hypothesen unterschieden. > **ungerichtete Hypothese** > *Regelmäßiges Ausdauertraining verändert den Blutdruck* > **gerichtetete Hypothese** > *Regelmäßiges Ausdauertraining senkt den Blutdruck* Der Informationsgehalt einer gerichteten Hypothese ist höher und basiert auf Vorwissen über den entsprechenden Forschungsgegenstand. ___ Weiterhin gibt es spezifische und unspezifische Hypothesen. > **spezifische Hypothese** > Regelmäßiges Ausdauertraining reduziert den Blutdruck um 10 mmHg Spezifische Hypothesen haben einen höheren Informationsgehalt, da sie auch Aussagen über die erwartete Effektgröße geben. ___ Wie oben bereits erwähnt wurde, geben Hypothesen Auskunft über die vermutete Beziehung von Variablen. Anhand dieser Beziehung lassen sich Hypothesen in einer der folgenden Kategorien einordnen. > **Unterschiedshypothese**: > Zwischen zwei oder mehreren Variablen besteht ein Unterschied > *Die Maximalkraft bei Männern ist höher als bei Frauen* > **Zusammenhangshypothese** > Zwischen zwei oder mehreren Variablen besteht ein Zusammenhang > *Je größer der Muskeldurchschnitt, desto höher die Maximalkraft* > **Veränderungshypothese** > Die Ausprägung einer Variable verändert sich über die Zeit > *Mit zunehmendem Alter sinkt die Maximalkraft* ___ Die Aufgabe der empirischen Forschung ist es mittels geeigneter Verfahren die Gültigkeit von Hypothesen zu untersuchen. Mittels [[Hypothesis testing | Hypothesentests]] lässt sich überprüfen, ob die Forschungsergebnisse die Annahme(n) der Hypothese unterstützen oder ihr widersprechen.