correction for guessingn
a mathematical adjustment to correct for the effects of random guessing
by test takers. It is not generally recommended in scoring a teacher-
constructed test, but when it is used, test takers should be informed that
their scores will be corrected for guessing or penalized.
corrective recastn
see recast
correlationn
a measure of the strength of the relationship or association between two or
more sets of data. For example, we may wish to determine the relationship
between the scores of a group of students on a mathematics test and on
a language test. Different types of correlation are reported in the applied
linguistics literature, whose use is determined by the types of variables that
are correlated. The Pearson product-moment correlation (r) is a measure of
association between two continuous variables. The point-biserial correla-
tion(rpbi) is a measure of association between a continuous variable and
a dichotomous or binary variable (e.g., gender – male versus female). The
biserial correlation (rb) is a measure of association between a continuous
variable and an artificially dichotomized variable (i.e. a variable that is
continuously measurable has been reduced to two categories (e.g. age – old
versus young or test score – pass versus fail)) but is rarely used nowadays.
The tetrachoric correlation is a measure of association between two
artificially dichotomized variables (i.e. both variables that are continuously
measurable have been reduced to two categories each). The phi correlation
(f) is a measure of association between two genuinely dichotomous
variables. The Spearman rank-order correlation or Spearman’s rho(r) is
a measure of association between two ordinal variables. The Kendall rank-
order correlation or Kendall’s tau(t) is another measure of association
between two ordinal variables but better deals with tied ranks (i.e. when
two or more test takers have the same score and thus occupy the same rank)
than the Spearman’s rho does.
correlational researchn
research carried out to examine the nature of the relationship between two
naturally occurring variables.
correlation coefficientn
also coefficient of correlation
a numerical index of the degree of relationship between two variables that
ranges in value from -1.00 (i.e. a perfect negative relationship) through
correlation coefficient