exclusive and exhaustive. A null hypothesis is a statement that “No differ-
ence exists between groups A and B” or “There is no correlation between
variables A and B”, whereas the alternative hypothesis is an opposite state-
ment that “The mean for group A is higher than that for group B” or “There
is a positive correlation between variables A and B”. The statistical analysis
of research results is frequently designed to determine whether or not a null
hypothesis should be rejected, thus providing support for an alternative
hypothesis.
see also hypothesis testinghypothesis formationn
(in language learning) the formation of ideas (“hypotheses”) by a learner
about the language he or she is learning. These hypotheses may be con-
scious or unconscious. Most people would agree that at least some of
these ideas come from the language we see and hear around us, but scholars
holding the innatist hypothesisclaim that our most important and
basic ideas about language in general are present at birth.
hypothesis testingn
a procedure to test a statistical hypothesis. A five-step version of hypothesis
testing proceeds as follows:
1 State a null hypothesis (H 0 ) and an alternative hypothesis (Ha).
2 Set a level of statistical significance (a) (see alpha).
3 Select and calculate an appropriate test statistic, a numerical value
calculated from the data sampled from a population and used to deter-
mine whether or not H 0 should be rejected, which results in a calculated
value.
4 Compare the sample evidence from Step 3 against a criterion (i.e. a
calculated value against a critical value or a p-value against a).
5 Make a decision regarding the null hypothesis (i.e. either to reject H 0 in
favour of Haor fail to reject H 0 ).
see also statistical significance
hypothesis testing