Statistical Sciences 2035 Study Guide - Quiz Guide: Standard Score, Null Hypothesis, Statistical Hypothesis Testing
7 Oct 2018
School
Department
Professor
HYPOTHESIS FORMATION, TYPES OF ERROR AND ESTIMATION
A hypothesis is a tentative assumption that is capable of being tested. The scientific
process never leads to certainty in explanation; only the rejection of existing
hypotheses and the construction of new ones, which may stand up better to the test of
empirical evidence. Thus, they are not proved by producing evidence that support
them; they are simply not disproved. There are two types of hypothesis - relationship
hypothesis (sth. causes sth.) and a comparison hypothesis (sth. is more/ less than
sth.).
1. The proposal of a hypothesis or tentative assumption to account for a phenomenon
or test the validity of some situation.
2. The deduction from the hypothesis that certain phenomena should be observed in
given circumstances.
3. The checking of this deduction by observation and testing.
Criteria of hypothesis:
- stated clearly, in correct terminology, and operationally defined
- testable so that it is capable of being either confirmed or refuted
- state expected relationships between variables explicitly
- be limited in scope
- grounded in past knowledge, consistent with most known facts
For a research, the researcher states two different hypothesis H0 and H1.
H0 is the null hypothesis, assuming that any difference or relationship that exists
between populations is due to chance.
H1 is the alternate hypothesis, a restatement of the hypothesis in the same terms as
those used in the null hypothesis except that the differences or relationships are
assumed to be real and not due to chance.
These two competing hypotheses are then tested by disproving H0. The null hypothesis
is assumed to be true unless we find evidence to the contrary and we can assume that
H1 is more likely to be correct. If the calculated significance level is (usually) 5% or less
(p < .05) then we can reject the null hypothesis - if it is more than 5% (p > .05) then we
retain the null hypothesis. If a researcher does not find significance, he can assume that
there is no significant difference. Statistics can never ‘prove’ anything. All a statistical
Document Summary
A hypothesis is a tentative assumption that is capable of being tested. The scientific process never leads to certainty in explanation; only the rejection of existing hypotheses and the construction of new ones, which may stand up better to the test of empirical evidence. Thus, they are not proved by producing evidence that support them; they are simply not disproved. Stated clearly, in correct terminology, and operationally defined testable so that it is capable of being either confirmed or refuted. Grounded in past knowledge, consistent with most known facts. For a research, the researcher states two different hypothesis h0 and h1. H0 is the null hypothesis, assuming that any difference or relationship that exists between populations is due to chance. H1 is the alternate hypothesis, a restatement of the hypothesis in the same terms as those used in the null hypothesis except that the differences or relationships are assumed to be real and not due to chance.
