WorksheetsResearch Methodology – MCQs (PhD AI & DS)
Total questions: 50
Worksheet time: 25mins
Research methodology primarily refers to:
Tools used for data collection
Statistical techniques
Systematic approach to solving research problems
Writing research papers
The first step in the research process is:
Literature review
Problem identification
Hypothesis testing
Data analysis
Which of the following best defines a research gap?
Missing references
Unanswered or underexplored problem in literature
Incorrect results
Lack of funding
Exploratory research is mainly used when:
Hypothesis is clearly defined
Problem is well structured
Little information is available
Large dataset is available
Which research approach is most suitable for theory testing?
Qualitative
Quantitative
Exploratory
Case study
A systematic literature review differs from a narrative review because it:
Uses fewer papers
Is biased
Follows a structured protocol
Avoids databases
PRISMA framework is commonly used in:
Survey design
Systematic literature reviews
Hypothesis formulation
Statistical testing
Which database is most relevant for AI research?
PubMed
IEEE Xplore
JSTOR
SSRN
A conceptual framework primarily helps in:
Data preprocessing
Visualizing variable relationships
Model training
Statistical testing
The validity of a research design ensures:
Repeatability
Accuracy of measurements
Appropriateness of conclusions
Speed of experimentation
Which sampling method ensures equal probability?
Convenience
Snowball
Stratified
Random
In AI-based research, ground truth data refers to:
Raw data
Synthetic data
Verified correct labels
Noisy data
Which data type is most suitable for deep learning?
Small tabular data
High-dimensional data
Categorical-only data
Survey data
Bias introduced during data collection affects:
Model accuracy only
Research reproducibility
Generalizability
Computational efficiency
Which is an example of primary data?
Kaggle dataset
Survey responses collected by researcher
Published benchmarks
Government reports
A null hypothesis (H₀) states that:
There is a strong relationship
There is no effect or difference
The model performs well
The data is biased
Type I error occurs when:
H₀ is accepted incorrectly
H₀ is rejected incorrectly
Sample size is large
Model overfits
p-value indicates:
Model accuracy
Probability of H₀ being true
Which test is suitable for comparing two means?
ANOVA
Chi-square
t-test
Z-test
Overfitting in ML models is analogous to:
High bias
High variance
Under-sampling
Data leakage
Reproducibility in AI research requires:
High accuracy
Open-source code and datasets
Large datasets
Faster hardware
Cross-validation is used to:
Increase training speed
Prevent data leakage
Estimate model performance
Reduce dataset size
Which metric is best for imbalanced datasets?
Accuracy
Precision-Recall
MSE
R²
Ablation study helps in:
Dataset augmentation
Understanding model components
Feature scaling
Hyperparameter tuning
Benchmark datasets are important because they:
Reduce computation
Enable fair comparison
Eliminate bias
Guarantee performance
Plagiarism refers to:
Reusing own work
Using common knowledge
Using others’ work without credit
Poor writing
Self-plagiarism involves:
Copying from others
Reusing one’s own published work without citation
Fabricating data
Peer review manipulation
Data fabrication means:
Data augmentation
Synthetic data generation
Making up data
Data cleaning
Which organization provides AI ethics guidelines?
IEEE
WHO
UNESCO
All of the above
Informed consent is essential when:
Using open datasets
Collecting human subject data
Training neural networks
Using GPUs
Impact factor measures:
Paper quality
Journal reputation
Author productivity
Citation accuracy
A Scopus-indexed journal ensures:
Fast publication
Peer-reviewed quality
Open access
No plagiarism
IMRAD structure refers to:
Intro, Method, Results, Discussion
Idea, Model, Results, Analysis, Discussion
Introduction only
Abstract format
Which section explains experimental setup?
Abstract
Introduction
Methodology
Conclusion
DOI stands for:
Data Object Index
Digital Object Identifier
Document Organization Index
Digital Output Indicator
A research contribution should be:
Incremental or novel
Only theoretical
Only experimental
Always interdisciplinary
Transfer learning is useful when:
Large labeled data is available
Limited labeled data is available
Data is balanced
Models are simple
Meta-analysis involves:
Single experiment
Combining results of multiple studies
Data preprocessing
Model selection
Digital Twin research mainly focuses on:
Virtual simulation of real systems
Hardware fabrication
Data encryption
Image compression
Explainable AI (XAI) aims to:
Improve speed
Reduce data size
Improve model transparency
Increase accuracy only
Peer review primarily ensures:
Fast acceptance
Ethical compliance
Quality and validity
High citations
Double-blind review means:
Reviewer identity hidden
Author identity hidden
Both author and reviewer hidden
Open review
Citation count indicates:
Paper length
Research impact
Which of the following is a major factor affecting research reproducibility?
Research funding
Dataset size
Journal quality
Dataset size
Research reproducibility crisis is mainly due to:
Lack of hardware
Poor documentation
Proprietary datasets
All of the above
Open science promotes:
Restricted access
Transparency and collaboration
Faster rejection
Closed datasets
A research proposal must clearly define:
Budget only
Objectives and methodology
Publication targets
Hardware details
Gantt chart is used for:
Data analysis
Time planning
Literature review
Hypothesis testing
Key Performance Indicators (KPIs) in research measure:
Cost
Progress and outcomes
Publication speed
Hardware efficiency
Interdisciplinary research is important in AI because:
AI is standalone
Real-world problems are complex
Reduces computation
Avoids ethics
