Assignment Remit

3000 words

60%

Module Learning Outcomes:

This assignment is designed to assess the following module learning outcomes. Your submission will be marked using the Grading Criteria given in the section below.

LO 1. Apply analytical tools and techniques to evaluate marketing data effectively.

LO 2. Critically evaluate analytical methods, assessing their suitability and limitations for solving marketing challenges.

LO 3. Analyse data and interpret results, connecting findings to marketing scenarios and consumer behaviour.

LO 4. Understand how marketing data is captured and processed, and use it to address real-world marketing problems.

LO 5. Integrate behavioural science and marketing principles to generate insights and develop data-driven strategies.

LO 6. Communicate findings effectively, presenting actionable recommendations through professional reports and visualizations.

Assignment:

Background and Context

You are part of an analytics team at a consulting firm that specialises in consumer insights and employer branding research. The firm has been approached by a group of corporate clients who are increasingly interested in understanding how employees express satisfaction and dissatisfaction online, particularly through platforms such as Glassdoor.

You have been given access to a raw Glassdoor review dataset containing text reviews, ratings, job titles, and metadata such as company, date, and location. Your task is to apply textual analytics and data transformation techniques to turn these unstructured reviews into structured, analysable insights.

The ultimate goal is to help the client understand online reviewing behaviour.

Your Task

You are expected to design and execute a textual analysis project that demonstrates your ability to manage, analyse, and interpret unstructured data to answer a business-relevant research question.

You are expected to:

Formulate a Research Question

Identify an interesting and specific question related to online reviewing behaviour (e.g., “What topics are most discussed in positive vs negative reviews?” or “Can we predict review sentiment based on linguistic features?”).

Explain why this question is important and relevant to the organisation or broader research context.

Understand and Prepare the Data

Describe the dataset, including number of reviews, structure, and variables.

Show how you cleaned and preprocessed the text data, including tokenisation, lemmatisation, stopword removal, and filtering.

Explain how you converted text into structured data, such as through word frequencies, TF–IDF representations, sentiment scores, or topic distributions.

Provide descriptive statistics and visualisations to give an overview of the data and initial insights.

Apply Advanced Analytical Methods

Use at least one advanced text analytics method introduced in the module. Options include (but are not limited to):

Topic Modelling (e.g., LDA or STM) to uncover latent themes in reviews.

Text Classification or Regression using text-derived features (e.g., to predict sentiment, rating, or job satisfaction).

Justify your chosen approach and explain how it helps address your research question.

Where relevant, compare multiple methods or visualise topic/sentiment relationships.

Interpret and Communicate Insights

Present key results clearly using tables, charts, and visualisations (e.g., word clouds, topic maps, sentiment distributions).

Discuss what your findings reveal about employee reviewing behaviour, perception drivers, or communication patterns.

Translate your analytical findings into business-relevant implications for the client.

Reflect Critically

Acknowledge limitations (e.g., data quality, model assumptions, sample bias).

Suggest possible improvements or future analytical directions.

Report Requirements

Structure:

Executive Summary

Introduction and Research Question

Data Preparation and Methodology

Results and Interpretation

Discussion and Managerial Implications

Conclusion

References (Harvard Style)

Appendices (including annotated R code and extended tables)

R Code:

Include all relevant R code in an appendix, structured into clear subsections (data cleaning, feature extraction, modelling, visualisation).

Code should be well-commented, reproducible, and match the analysis presented.

Notes and Technical Guidance

Word Count:

The report must not exceed 3,000 words, excluding the title page, executive summary, references, tables, figures, R code, and appendices. There is no permissible 10% excess on this limit.

Dataset Size and Computational Considerations:

The Glassdoor review dataset is large and may require significant computational time and memory to process, depending on your personal computer’s performance.

It is therefore strongly recommended that you begin by exploring and testing your workflow on a smaller subset of the data (e.g., random samples or filtered segments) before scaling up to the full dataset.

While you are encouraged to work with the full dataset if your system allows, for fairness and practicality:

You are only required to analyse a minimum of 10,000 reviews for your submission.

You may choose any random or stratified subset as long as your sampling strategy is clearly documented and justified.

Remember that text preprocessing and model training can be computationally intensive—allow sufficient time for data cleaning, model fitting, and visualisation steps.

Reproducibility:

Your report should demonstrate that the analysis could be replicated. Clearly state any sampling, filtering, or randomisation methods used.

Practical Tip:

Begin early to allow time for testing and model optimisation.

Grading Criteria / Marking Rubric

Your submission will be graded according to the following criteria:

Research Question & Justification [15%]

Data Preparation & Transformation [20%]

Analytical Method & Modelling [25%]

Interpretation, Visualisation & Insight [25%]

Structure, Clarity & Professional Presentation [15%]

See the marking rubric at the end of the remit for more information on how your work will be marked and graded.

Ethical Use of Generative AI (GenAI)

You are permitted to use GenAI to support your submission for this assessment. You may use it for the following activities:

Researching and refining your ideas

Information retrieval or background research

Drafting an outline to organise or summarise your thoughts

Refining research questions

Checking spelling and grammar

Applying GenAI tools should be done with human oversight and control. You should carefully review and use the results carefully as AI can generate authoritative-sounding output that can be incorrect, incomplete, uncritical, or biased.

You may not submit any work generated by an AI tool as your own. Where you include any material generated by an AI tool, it should be properly declared just like any other reference material. Alongside your assignment you should also provide a commentary in the Cover Sheet detailing how GenAI has been used to develop your final submission. If you have not used GenAI tools, you should clearly state so.

Plagiarism, including that which results from using GenAI, is a form of academic misconduct that will be dealt with under the University’s Code of Practice on Academic Integrity. https://intranet.birmingham.ac.uk/as/registry/policy/conduct/plagiarism/index.aspx

University guidance on ethical use of GenAI can be found here:

https://intranet.birmingham.ac.uk/as/libraryservices/asc/student-guidance-gai.aspx

Further Guidance:

Feedback to Students:

Both Summative and Formative feedback is given to encourage students to reflect on their learning that feed forward into following assessment tasks. The preparation for all assessment tasks will be supported by formative feedback within the tutorials/seminars. Written feedback is provided as appropriate. Please be aware to use a web browser and not the Canvas App as you may not be able to view all comments.

Referencing:

Please use the Harvard System throughout your assignment. You should consult https://intranet.birmingham.ac.uk/as/libraryservices/library/referencing/icite/harvard/index.aspx for specific guidance on using the Harvard method of referencing.

Plagiarism:

It is your responsibility to ensure that you understand correct referencing practices. You are expected to use appropriate references and keep carefully detailed notes of all your information sources, including any material downloaded from the Internet. It is your responsibility to ensure that you are not vulnerable to any alleged breaches of the assessment regulations. More information is available at University’s Code of Practice on Academic Integrity

https://intranet.birmingham.ac.uk/as/registry/policy/conduct/plagiarism/index.aspx.

Wellbeing, Extensions and Extenuating Circumstances:

The processes for extensions and extenuating circumstances (ECs) are to support students who have experienced unforeseen issues that have impacted their ability to engage with their studies and/or complete assessments. Students should notify Wellbeing of any extenuating circumstances as soon as possible via the online form, following the guidance provided.

https://intranet.birmingham.ac.uk/social-sciences/college-services/wellbeing/index.aspx

Marking Rubric:

Note that the information below is guidance and feedback only and not a quantitative measure to calculate the grade.

The final grade represents the overall quality of the work taking these criteria into account but is the academic judgement of the marker(s).

Research Question & Justification (15%)

Data Preparation & Transformation (20%)

Analytical Method & Modelling (25%)

Interpretation, Visualisation & Insight (25%)

Structure, Clarity & Professional Presentation (15%)

Distinction (70% or more)

Research question is highly original, clearly articulated, and theoretically or managerially significant. Demonstrates deep understanding of online reviewing behaviour and provides a compelling rationale for the investigation. Analytical objectives are precise, achievable, and well aligned with the data and chosen methods.

Demonstrates excellent understanding of textual data structure and applies comprehensive, transparent preprocessing (tokenisation, lemmatisation, stopword removal, stemming, feature creation). Shows mastery in converting unstructured text into structured data (e.g., TF–IDF, sentiment scores, topics). Explains all steps logically with clear justification and reproducible workflow.

Applies advanced and well-chosen analytical techniques (e.g., topic modelling, sentiment analysis, text classification, or regression) with strong justification. Implementation is accurate, methodologically sound, and demonstrates critical awareness of assumptions and limitations. May combine methods creatively.

Produces highly insightful, data-driven interpretations supported by clear and compelling visualisations (topic maps, sentiment distributions, key-term networks, etc.). Explains findings convincingly in business and behavioural terms, showing deep understanding of reviewer sentiment, themes, or patterns. Provides strong managerial implications that link directly to evidence.

Exceptionally well-structured and polished report. Logical flow, professional tone, and concise writing. Figures and tables are labelled and integrated effectively. Code is clearly organised and annotated. Referencing is complete and accurate in Harvard style. Presentation is client-ready.

Merit (60–69%)

Research question is relevant and clearly explained, though may lack novelty or conceptual depth. Rationale is logical and shows awareness of business context. Objectives align with data and chosen analysis.

Good preprocessing and transformation steps applied, with minor omissions or limited explanation of certain stages. Demonstrates sound understanding of text structuring and feature extraction, though workflow may lack complete transparency.

Appropriate analytical method(s) selected and correctly implemented. Analysis is mostly accurate and justified. Some minor gaps in discussion of assumptions or rationale. Demonstrates technical competence and good interpretation of results.

Provides clear interpretation of findings with reasonable depth. Visualisations are relevant and generally well-presented, though some may lack clarity or integration. Links results to business meaning, though insights could be developed further.

Report is clear, coherent, and professional. Minor issues with structure, language, or flow. Code included and mostly understandable. Figures and tables formatted adequately. Referencing accurate with small inconsistencies.

Pass (50–59%)

Research question is basic, weakly justified, or too broad. Motivation or connection to business context is limited. Objectives are somewhat unclear or poorly linked to analytical approach.

Some data cleaning and preprocessing performed, but steps are partially incomplete or poorly documented. Limited demonstration of understanding of how unstructured text becomes structured.

Basic or partially correct analysis performed, with limited justification. Advanced techniques may be used superficially or with small errors. Evidence of method application but weak interpretation of model output.

Descriptive interpretation of results, but lacks analytical depth or clear connection to business implications. Visualisations are present but underdeveloped, unclear, or poorly formatted.

Adequate structure but lacks polish and professional clarity. Some inconsistencies in section flow or presentation. Code may be incomplete or poorly labelled. Referencing inconsistent or partially incorrect.

Marginal Fail (40–49%)

Research question is unclear, poorly defined, or lacks justification. Minimal attempt to explain importance or relevance.

Limited or unclear preprocessing. Key steps (e.g., tokenisation, stopword removal, or feature creation) missing or incorrect. No clear explanation of how text data were transformed.

Inappropriate or incorrectly applied methods. Weak understanding of technique, with little justification or interpretation.

Weak or missing interpretation. Visualisations confusing, irrelevant, or missing. No clear connection between results and business context.

Poorly structured and difficult to follow. Code largely missing or unorganised. Numerous writing and referencing issues.

Fail (0–39%)

No meaningful research question or rationale provided.

No evidence of valid text preprocessing or structuring.

No valid analytical work performed.

No interpretation, insights, or visuals.

Disorganised and incomplete submission. Little or no referencing.