An index compares data under two hypotheses
A kinship index usually expresses a likelihood ratio: how much more likely the observed genetic data are under one specified relationship than under an alternative hypothesis. For paternity, the hypotheses may be that the tested man is the father and that the father is another person unrelated to him. The meaning of the number is defined by this particular pair of hypotheses.
An index does not show the proportion of shared genome and is not a kinship percentage. It cannot be read as the probability of laboratory error or sample substitution. A value greater than one supports the first stated hypothesis relative to the second; a value below one supports the second relative to the first. Whether this is sufficient for a particular conclusion is assessed together with the task, data quality and limitations.
Where the percentage in a report comes from
Converting a likelihood ratio to the probability of a hypothesis requires an initial, or prior, probability. In a simple calculation for two hypotheses, the final probability equals LR × q / (LR × q + 1 − q), where LR is the likelihood ratio and q is the assumed prior probability of the first hypothesis. The formula does not create missing family history information.
If q = 0.5 is assumed, the formula reduces to LR / (LR + 1). Equal prior probabilities are an assumption that must be stated; they are not derived from genetic markers. A percentage without the index, hypotheses compared and assumptions is therefore difficult to assess substantively. Extra decimal places likewise do not replace a description of the objects examined.
Why the participants change the task
Examining a child and alleged father differs in the available information from an examination including the mother. For indirect kinship, the exact participants matter: full or half-siblings, a grandmother, grandfather or other relatives. Different family structures may explain some of the same observations. Substituting another relationship name into an existing percentage without new justification is not valid.
If a significant alternative concerns a close relative of the tested person, comparison with a random unrelated person answers a different question. Tell the appropriate specialist about this before calculation. For preliminary discussion, a diagram using coded identifiers is helpful, showing which relationships are considered known, which are being tested and which remain possible.
Which assumptions underlie a combined index
The combined indicator is calculated from the genetic systems examined using an appropriate model. The marker list, population frequencies applied, their sources and significant adjustments matter. Simply multiplying any figures provided is not always permissible: marker properties and the conditions under which the calculation assumes independence must be considered. A forensic genetics specialist checks these questions.
A mismatch at a locus, possible mutation, incomplete profile or characteristics of complex material may require separate explanation. A difficulty does not automatically justify either excluding kinship or declaring it insignificant. The report must show how the observation was considered and whether it changes the strength of support for the hypothesis. Checking the calculation requires input information rather than the final paragraph alone.
What to request for a clear analysis
Start with the full report and appendices. Check whether participants and samples, relationship hypotheses, methods, marker results, statistical indicator and its assumptions are stated. If the document contains only a percentage, ask the provider specifically which hypotheses it concerns and which prior probability was used.
Preparation and review help define a task for the appropriate specialist; laboratory examination is conducted separately. A genetic kinship conclusion is not a medical diagnosis and does not by itself determine legal consequences. For an initial enquiry, describe the relationship type and available documents using coded identifiers. Do not send identifying genetic tables or biological material through the general form; transfer arrangements are agreed separately.