The effect of partial dependence in a binary sequence on tests for the pres
ence of a changepoint or changed segment are investigated and exemplified i
n the context of modelling noncoding deoxyribonucleic acid (DNA). For the l
evels of dependence that are commonly seen in such DNA, the null distributi
ons of the test statistics are approximately correct and so conclusions bas
ed on them are still valid. A strong dependence would, however, invalidate
the use of such procedures.