Appraisal & common methodological issues

A number of important research methodological issues were frequently identified in research investigating potential risk factors for human-directed dog aggression.

The appraisal process involved examining each asset of a study outlined below in order to determine the quality of evidence, for risk factors of human-directed dog aggression, provided by the work.

Several important methodological issues were identified in the studies; these are outlined below. Within the database some of the limitations of each paper have been identified using the same headings as below. Neither list is exhaustive.

Design

  • Absence of comparator group.
  • Absence of sample size calculations.
  • Small sample size, leading to high chance of insufficient power to detect effect.

Selection

  • Convenience sampling.
  • Selection of controls from different population to cases.
  • Use of registries as comparator.
  • Lack of randomisation.
  • High risk of bias.

Measurement

  • Lack of clear definition of outcome - for example what is meant by "bite"?
  • Non-reproducible or subjective measurement methods.
  • Lack of standardisation.
  • Absence of blinding.

Lack of control of confounding through design and/or analysis

 Analysis

  • Inappropriate methods.
  • Univariable analyses where multivariable would have been appropriate.
  • Multiple statistical tests - increasing the likelihood of chance findings.
  •  Lack of a clear hypothesis.
  • Causal relationships identified where direction of association unclear.