Most research is retrospective in nature - that is, data on troubled or ended relationships is based on memory of the relationship over time. There is some prospective research, such as Bradbury & Fincham, but these may be open to demand characteristics.

The majority of research is carried out on Western couples who have come in to seek therapy. There is a sampling bias of the research.

Some research is “hypothetical” in nature - that is, asking participants to anticipate how they would react to a variable. Thus, the research may lack ecological validity.

Peak-end rule will influence perceptions of the relationship - that is, they will tend to remember the last part of the relationship, which is negative, rather than the number of positive experiences experienced over the duration of the relationship.

Research is correlational in nature and does not establish a cause and effect relationship. This also means that there is often bidirectional ambiguity - did a lack of disclosure lead to the end of the relationship or did feelings of disengagement from the relationship lead to a lack of disclosure?

Relationships are complex and there is not usually a single variable that leads to the dissolution of the relationship. Many theories are reductionist in nature and do not address the interaction of factors that may negatively influence a relationship .