---

---

---

---

---

---

Chapter wise analysis of Shodhganga SRHU Dehradun PhD Thesis critique,

▶ AIM AND OBJECTIVES: Chapter 3

• Missing consistency (Aim of study: “Evaluate”, research statement: “Evaluate effectiveness”; simultaneous use of Yojana and Scheme; use of “Tertiary Hospital”, “Tertiary Care Hospital” and “Tertiary Care Multi-specialty Hospital” to refer to the same hospital; use of Ayushman Card to indicate coverage under AAUY, using both “Health Scheme” and “Health Insurance scheme” to refer to one and the same concept).
• “Quality service” does not mean “Quality of healthcare services provided by the hospital for patients covered under AAUYHIS”
• Objective 3 is to “compare the expenditures” while Hypothesis 2 refers to “financial viability”
• Hypothesis 1 generalizes the combined objectives 1 & 2 to be applicable to “all segments of the Community in Uttarakhand”

■ MATERIAL AND METHODS: Chapter 4

• The phrase “hospitalized Ayushman discharge patients” remains ambiguous. It does not clearly refer to “patients who were previously hospitalized under the AAUYHIS and later discharged.”
• The statement “Ayushman counter contact information was gathered” does not clarify that “the patient’s contact information was collected from the AAUYHIS Office or Desk.”
• The questionnaire does not include levels for perceptions.
• The inclusion criteria incorrectly imply that AAUY is a scheme of the said hospital, which it is not.
• The exclusion criteria will bias both Objective 1 and Objective 2. As a result, it will also bias Hypothesis 1.

Sample size

• If the research population includes only “patients admitted to the said hospital,” then the findings cannot generalize to Hypothesis 1, which concerns “all segments of the community in Uttarakhand.”
• The correct spelling is “Cochran,” not “Cochrane”: https://en.wikipedia.org/wiki/William_Gemmell_Cochran
• The researcher did not adjust the Cochran formula for the convenient sampling technique.
• The calculated sample size suits only binary outcomes like “satisfaction versus non-satisfaction” under the assumption of normal distribution.
• The pilot study did not inform any adjustment to the sample size.
• The methodology uses “Simple random sampling,” “Convenient sampling,” and “Purposive sampling” without clarifying their respective roles or differences.

⭓ OBSERVATIONS AND FINDINGS: Chapter 5

• The study does not include pictorial representation.
• The data shows a skewed age distribution.
• The researcher grouped businessmen, self-employed professionals, and unemployed individuals into one occupation category.
• The information on AAUY exceeds 100% because the categories overlap and are not mutually exclusive.
• The data on reasons for choosing the hospital also exceeds 100% due to overlapping, non-exclusive categories.

Questionnaire

• The questionnaire contains flaws. Question A-AD_5 introduces the word “Safe,” which adds a construct that may not relate to satisfaction. This weakens the question’s reliability. Its “deleted alpha” is lower than the total alpha, and the correlation is only 0.351.
• Question B-AD_6 reduces reliability. It has a higher “deleted alpha” than the total alpha, and its correlation is only 0.259. The researcher should remove this question.
• The questionnaire asks about individual process steps but fails to assess satisfaction with those steps directly. A better questionnaire would have asked about satisfaction with each step.

Analysis

• The analysis confuses perception and satisfaction.
• The report does not mention the Cronbach alpha for perception level measurement.
• The researcher did not explain the perception score calculation.
• The report also omits the satisfaction score calculation.
• The author did not mention the method used to test statistical significance, making the p-value unreliable.
• The report misapplies the Chi-square test and fails to describe the methodology. It also lacks a clear explanation in plain text.
• The analysis of Objective 3 appears arbitrary and lacks depth.

---

---

Framing and Objectives

From a research framing perspective—covering objectives, questions, hypotheses, and alignment—the thesis reveals several key points: The thesis frames some research questions unclearly and fails to align them fully with the analysis methods. As a result, only some questions are statistically testable. The thesis maps some hypotheses to suitable statistical tests and makes them testable. The researcher links some variables to conceptual constructs within the academic field. The study ties some outcome metrics directly to the research goals and objectives. The analysis fails to connect explicitly with academic standards, which weakens domain-specific depth. The researcher does not fully document the study’s limitations and omits potential sources of bias, data issues, and methodological constraints.

---

Data Integrity & Provenance

From the perspective of data integrity and provenance, the thesis presents several key issues: The researcher does not document or trace data provenance. They do not confirm raw data integrity or verify it against the original source. The researcher does not document the data preprocessing and cleaning pipeline. They also fail to make the steps reproducible. The researcher does not analyze missing data. He handles missing data without transparency and provides no justification for the imputation methods used. The researcher does not define or justify outlier detection methods and fails to apply them consistently across datasets. The researcher does not document the data transformation process (e.g., normalization, encoding). They also fail to maintain data lineage and transformation logs. The researcher cites all external datasets and clearly defines their provenance.

---

Exploratory and Descriptive Statistics

From the perspective of exploratory and descriptive statistics, the thesis reveals several key points: The researcher does not perform exploratory data analysis (EDA) with visualizations to assess variable distributions. The researcher computes descriptive statistics (mean, median, variance, etc.) for some relevant variables and considers their scale and distribution. He does not compute skewness or kurtosis for continuous variables and does not discuss outliers. The researcher does not conduct correlation or association analysis, or they interpret direction and magnitude incorrectly. They either fails to document missing values clearly or omits visualizations such as heat-maps to highlight data gaps. The researcher either does not apply dimensionality reduction methods where needed or fails to justify the number of retained components.

---

Inferential Statistics

From the perspective of inferential statistical analysis, the thesis reveals several key issues: The researcher does not verify assumptions of statistical tests such as normality, homoscedasticity, and independence. They do not perform tests like Shapiro-Wilk, Levene’s test, or Durbin-Watson. The researcher does not use non-parametric tests (e.g., Mann-Whitney U, Kruskal-Wallis) when data violates assumptions for parametric tests. The researcher does not apply corrections for multiple comparisons using methods like Bonferroni or FDR. They also fail to document such procedures. The researcher does not report effect sizes (e.g., Cohen’s d, R²) or include 95% confidence intervals with significant results.

He does not consider Bayesian methods where appropriate or fails to define priors clearly and include convergence diagnostics. The researcher does not justify model selection using metrics like AIC/BIC or omits cross-validation results. They do not analyze residuals for bias patterns or fails to test for heteroscedasticity using the Breusch-Pagan test. The researcher conducts sensitivity analyses only partially and does not fully evaluate the robustness of findings against different assumptions or model specifications.

---

Reproducibility & Transparency

From the perspective of reproducibility and transparency, the thesis reveals several key issues: The researcher fails to list or document all dependencies (e.g., Python/R packages, specific versions) in a requirements file or Docker container. They do not automate the full analysis pipeline using tools like Makefile, Snakemake, or similar systems. Some stochastic processes use fixed random number generators and initial seeds, but the setup does not ensure reproducibility. Although containerized environments (e.g., Docker) are available, they do not guarantee that the analysis runs across all systems. The researcher annotates scripts with comments but does not clarify non-standard or complex code segments. The README file lacks either detailed instructions or omits essential information like runtime, required packages, and input/output descriptions.

---

Ethical and Responsible Conduct as presented on Shodhganga by SRHU Dehradun in a PhD Thesis

Perspective of ethical and responsible conduct; following are some of the key points that were observed from this thesis. Firstly, Ethics approval is documented, including the relevant protocol numbers and dates. Informed consent forms cover all aspects of data collection, storage, and reuse. Secondly, Data anonymization is irreversible, with a clear process outlined for handling sensitive data. Thirdly, Informed consent is aligned with analysis use. Fourthly, Either bias mitigation strategies (e.g., fairness metrics, algorithmic transparency) not implemented or not discussed. Fifthly, Data fairness was not discussed in the report. Lastly, Ethical limitations are not acknowledged, and the potential societal or individual impacts of the findings are not considered.

---

Reporting & Documentation as presented on Shodhganga by SRHU Dehradun in a PhD Thesis

Point of view of reporting & documentation; following are some of the key points that were observed from this thesis. Firstly, None of the figures and tables reference their corresponding code block, ensuring transparency. Secondly, None of the captions include sample sizes, test types, and units, with full transparency on methods. Thirdly, Plain-language summaries do accompany some of the statistical results, making the findings inaccessible to non-experts. Fourthly, Raw output (e.g., model coefficients, test statistics) is not included in supplementary materials. Lastly, The final report does not track iterative changes, and each version is not well-documented, version control is not maintained.

---

Data Analysis Health Score (DAHS) for Shodhganga SRHU Dehradun PhD Thesis critique,

The overall data analysis health score for this PhD thesis evaluates to be -36%, as shown in the graph above. Furthermore relative scores for individual analytical processes are also indicated in the above graph, and are as follows. The thesis scores -10% on Framing and Objectives, -83% on Data Integrity & Provenance, -62% on Exploratory and Descriptive Statistics, -62% on Inferential Statistics, 0% on Predictive and Prescriptive Modeling, -33% on Reproducibility & Transparency, 64% on Ethical and Responsible Conduct and -70% on Reporting & Documentation.

---

---