The application of forensic science depends on a variety of skilled scientists who apply scientific thinking and methodology to a wide range of evidence associated with a crime scene.
Scientific inquiry relies on being able to ensure that the scientific question being investigated can be answered through observational and experimental approaches, and conclusions drawn need to be based on evidence that is unaffected by personal values, opinions, or expectations.
How can bias affect a scientific investigation?
When conducting research, scientists try to design our studies to ensure that they give a fair and accurate result. For example, this might involve making sure that materials are handled carefully to avoid contamination, and that measurements are taken and recorded precisely and consistently. Although there are sometimes cases of scientific fraud – a serious ethical violation – these are thankfully rare.
However, scientists are, of course, people and we all have our own biases. Some of these biases can be subtle and could affect the results of our research if we are not careful to control them. There are also other potential sources of error beyond biases, which scientists need to be aware of so that we can control for them. We’ll start with an example of a non-bias potential source of error: the ‘observer effect’.
If you drive a car, you will be familiar with checking tyre pressure. Imagine you are responsible for determining the exact pressure of your tyre as the car stands on the road. Is this possible with the tools you have? You may have answered ‘Yes’ to this, as you have access to an accurate tyre pressure gauge.
Consider what happens when you take the measurement though – it’s impossible to do this without letting a small amount of air escape, which changes the pressure of the tyre very slightly. This is an example of observer effect: that is when the action of observing (or measuring) something changes the thing being observed.
In forensic investigations, a wide range of evidence is collected from the scene and processed in laboratories. Some samples are direct samples of fluid or fabric which can then be tested themselves. Other evidence can be more indirect. An example of this is fingerprint evidence, which is initially a transfer from an individual’s finger to a surface, and crime scene investigators then take an impression of this for analysis.
For this kind of evidence, it is vital that investigators are aware of how any processes may affect the samples taken and the protocols and procedures are designed to minimise or remove observer effects. Some forensic processes are ‘destructive’, in that they damage or destroy the trace such that other tests cannot be done later.
For example, swabbing a fingerprint for DNA will disrupt the print itself, as the swab will smudge it. To control for this, prints are photographed before swabbing – a simple, but important, procedural safeguard to maximise the value of the fingerprint evidence.
Whilst observer effects are often easy to identify and control for, it is much harder to eliminate biases. First, a point of terminology: here we are not talking about biases in the sense of opinions or viewpoints (e.g. favouring a particular political party or policy). Rather we are talking about what psychologists call ‘cognitive biases’, and in the specific context of scientific research ‘experimenter bias’. This is the term used to describe occurrences where scientists inadvertently influence an experiment, observation or analysis in such a way as to influence the result of their conclusion (if it was intentional it would be fraud, rather than bias).
This unintentional bias can manifest in the scientist unwittingly undertaking an experiment in a particular way which would influence the result in a particular direction, or only looking at the results of an experiment that confirm what they were originally expecting and ignoring other results which did not confirm their expectations.
It can also lead to an unconscious tendency to preferentially interpret ambiguous evidence as supportive of an outcome they are invested in. This sort of bias is also called confirmation bias because it involves being biased towards interpreting evidence in a way that confirms our expectations. It may surprise you to know that we can all unintentionally fall foul of it, and it is vital that all scientists are aware of the possibility of unintentional experimenter bias in order to avoid it.
This is important when related to the scientific tests carried out in criminal investigations, where there may be preconceptions of the guilt or innocence of a suspect.
For this reason, scientists working in these laboratories have stringent procedures and protocols to avoid introducing confirmation bias. This includes only receiving information that is necessary for the processing of samples rather than large amounts of additional information about the case. This minimises expectation effects, in this case referring to the likelihood that the interpretation and perception of the reliability of a result may change as a direct result of the expectations of the scientist interpreting that result.
In simple terms, if a forensic scientist has the expectation that the suspect is guilty and they are presented with ambiguous evidence (e.g. a slightly unclear partial fingerprint or grainy CCTV footage), they might interpret the evidence as suggesting the suspect’s guilt (e.g. focus on similarities in the fingerprint or CCTV image).
In contrast, if they think that the suspect is innocent then they might interpret the same ambiguous evidence as consistent with the suspect’s innocence (e.g. focus on differences in the fingerprint or CCTV image). It is, therefore, necessary for forensic scientists to – as much as is possible – have no expectations about the guilt or innocence of the suspect.
Acknowledging the potential for bias in science isn’t a weakness – it enables us to strengthen protocols and procedures to ensure it doesn’t affect the validity of the final conclusions.