Quantitative Forensic Science Research Topics: 12 Feasible Student Projects

Published · References and presentation revised

Save

A good research title starts with a measurable question and data you can actually obtain. For a student project, a small, reproducible test on known samples is usually more workable than a promise to evaluate an entire forensic laboratory or police department.

The ideas below are proposed study designs, not completed studies or findings. Check your department's requirements, available equipment and supervision before choosing one. NIJ's overview is a useful starting point for the range of forensic disciplines; a review of quantitative evaluation in digital forensics explains why context and the meaning of a result matter.

Choose the data before polishing the title

  1. Write down what you can measure: elapsed time, recovered known files, agreement between repeated classifications, or completeness of a report.
  2. Identify an accessible source: files you create, deliberately prepared training samples, a suitable public teaching dataset, or records you have explicit permission to use.
  3. Define the comparison. Change one factor at a time where possible, and document what stays fixed.
  4. Decide what counts as success before seeing the results. A known-answer file set gives a clearer baseline than an unknown collection of real case files.
  5. Keep the claim within the design: success on your training samples does not establish a tool's accuracy in all investigations.

Twelve projects with an observable outcome

Working titleWhat to measureFeasible starting materialMain design check
Repeatability of file-hash results across two documented toolsAgreement for unchanged files and deliberately modified copiesA small set of files you createUse the same hash algorithm and preserve the exact inputs.
Effect of image resolution on recovery of known metadata fieldsPresence of predefined fields after controlled resizing/exportYour own images with recorded original metadataRecord the export software and settings; do not assume all resizing strips metadata.
Recovery of known file types from a controlled test imageCorrectly recovered known files and false identificationsA permitted teaching disk image or a synthetic training imageKeep a ground-truth file inventory and a fixed tool version.
Effect of examiner instructions on evidence-log completenessMissing predefined documentation fieldsA fictional case exerciseUse a written scoring rubric; obtain required approval before involving participants.
Agreement between students classifying simulated trace imagesAgreement on a defined set of observable featuresNon-sensitive teaching imagesSeparate agreement from correctness, and avoid claims about real suspect identification.
Reproducibility of a documented digital-evidence workflowMatching outputs and differences across repeated runsThe same known-answer training filesLog software versions, settings and environment.
Effect of report structure on completion of a forensic reporting checklistChecklist coverage and omitted limitationsReports for a fictional exerciseDefine the rubric before scoring and disclose who scored the reports.
Time needed to locate known artifacts using two documented workflowsTask time and missed target artifactsA synthetic set with a fixed answer keyControl order and practice effects; speed alone is not accuracy.
Consistency of timestamp interpretation across documented display settingsDifferences in displayed times for known eventsFiles created at recorded timesRecord time zone, daylight-saving rules and which timestamp field is being read.
Effect of file renaming on extension-based and content-based identificationCorrect identification against known original typesHarmless files you create and renameKeep originals; do not treat an extension as ground truth.
Completeness of chain-of-custody forms in a simulated transfer exerciseMissing fields and inconsistent transfer entriesFictional items and formsMeasure documentation only; do not infer legal admissibility.
Agreement when two readers apply a predefined evidence-description rubricAgreement and categories of disagreementThe same permitted teaching recordsReport ambiguous categories rather than silently changing the rubric.

Turn one idea into a short proposal

For the file-renaming project, create a known set of harmless text, image and document files. Record each original type, create renamed copies, then compare an extension-based classification with the output of a documented content-based method. Keep the files and settings fixed. Report the number correctly identified out of the known total, errors by file type, and any files the method could not classify.

The conclusion should describe this test set. It should not claim that a tool is universally reliable, that the exercise reproduces a criminal investigation, or that one small result proves evidential validity.

Common proposal problems

Questions before committing to a topic

Do I need real crime records?

Not for the controlled exercises above. Known, permitted training material can make the design more reproducible and avoid the access and privacy problems of real case files.

Can a questionnaire be a forensic accuracy study?

A questionnaire can examine reported knowledge, attitudes or experiences. Those responses alone do not measure a forensic method's accuracy against known evidence.

How narrow should the title be?

Name the measurable outcome, comparison and setting when those are known. Leave room for an inconclusive result; do not write the desired conclusion into the title.

Sources