STAGE 2 · PARTICIPANT PRACTICE PACK
Screen a Shared Practice Set
Work through a small, consistent dataset before applying the same process to your own review records.
Published review examples: Article Evidence Lab
Use the five review PDFs and the Li et al. (2024) IoT security case to practise locating screening counts, quality assessment, data extraction and synthesis. The extended case includes a supplementary ZIP with the article and research materials. Record the page, section, table or figure supporting each answer.
Try each task before opening its self-check. If the article does not report a detail, write “not stated” rather than guessing.
Exercise scenario and protocol
Practise a review question about feature selection or feature extraction methods used in machine-learning intrusion detection for Internet of Things (IoT) environments.
Include when all apply
- Peer-reviewed journal or conference primary study.
- Published from 2019 to 2025 inclusive.
- Studies IoT devices, networks or IoT traffic.
- Uses machine learning for intrusion/attack detection.
- Evaluates feature selection, ranking or extraction and reports relevant results.
Exclude when any apply
- Outside the date range, or not a primary empirical paper.
- Wrong context, such as cloud-only, Android-only or non-IoT industrial systems.
- Not intrusion/attack detection, or no machine learning.
- No evaluated feature selection/extraction method.
- IoT results cannot be separated or extracted.
Apply these as the exercise protocol. In a real review, criteria must be justified by the review question and documented in the protocol.
Step-by-step activity
Inspect and preserve the source set
Download the CSV. Note the number of imported records and keep an untouched copy before changing anything.
Find duplicates before screening
Compare title, authors, year and abstract. Record duplicate pairs; retain one copy of each unique record. Do not confuse a duplicate database record with a related but distinct study.
Calibrate independently
Each reviewer screens the same small batch without discussing decisions first. Mark Include, Exclude or Uncertain and write the criterion that supports each decision.
Compare and clarify
Compare decisions, especially disagreements and uncertain records. Explain which rule caused the difference. Clarify the rule before completing the rest of the set.
Complete title and abstract screening
Record one primary reason for every exclusion. Use Uncertain when the abstract does not provide enough information; do not infer missing details.
Release and assess full-text notes
After abstract decisions are recorded, the facilitator may provide the full-text simulation notes. Reassess the candidate reports and record the final decision and reason.
Reconcile the flow counts
Enter the counts in the Stage 2 guided workspace. Explain any difference and verify that each record is accounted for.
Debrief and transfer
Discuss ambiguous criteria, then repeat the workflow with your own Stage 1 exports. Keep the actual review decisions in the selected review system and maintain a backup.
Files for the exercise
| File | When to use it |
|---|---|
| Practice records CSV | 23 imported records, including 3 duplicate copies; abstracts are synthetic. |
| Screening decision log | Record independent decisions, reasons, reviewer and conflict outcome. |
| Full-text simulation notes | Release after title/abstract screening to practise full-text decisions. |
Do not open the facilitator key until reviewers have completed their independent decisions.