Offshore wind energy - draft updated sectoral marine plan 2025: consultation analysis report

3 Analysis Method and Reporting

3.1.1 Responses to the consultation were submitted via Citizen Space^[19] or directly to Scottish Government’s Offshore Wind Directorate (OWD). Responses received by hard copy or email were uploaded to Citizen Space by OWD in line with Scottish Government consultation procedures.

3.1.2 Following the completion of the consultation period, responses were compiled by OWD into a spreadsheet and shared with the consultants, Arup, who were responsible for the analysis of the responses and compiling this report. Personal information supplied by respondents was not shared.

3.1.3 Responses were organised by type (individual or organisation). Responses from organisations were further assigned to a respondent ‘sub-type’ (organisation type) as well as sector (see Table 1 and Table 2). This enabled analysis of the types of organisations that had responded and whether there were differences or commonalities across responses from different types of organisations.

3.1.4 Two forms of analysis were utilised for responses based on the format of the question being analysed. For multiple choice questions, basic summary statistics were produced showing the breakdown of responses by respondent type, sub-type and sector. For free text questions, responses were analysed using topic modelling. Topic modelling is a type of natural language processing which uses artificial intelligence methods such as machine learning to identify and categorise topics within a collection of text.

3.1.5 The purpose of topic modelling is to discover thematic structures and patterns within sets of text. It helps researchers and analysts to understand large volumes of information by grouping similar themes or subjects together. Topic modelling is now regularly used in various fields which deal with large volumes of free text data including academic research and community engagement^[20].

3.1.6 Topic modelling was identified as a suitable technique to analyse free text questions in expectation of receiving large volumes of text data. To facilitate this, a natural language processing computer model developed by Arup and tailored the specifications of the draft updated Plan consultation was utilised (herein after referred to as ‘the model’).

3.1.7 The steps in the topic modelling analysis included:

• Pre-processing: Cleaning the free text responses to remove unnecessary characters, repeated words and grammatical inconsistencies, which helped the model perform better.
• Topic identification: Common and unique topics in the responses were identified by running the model on each question. The Arup team then analysed these results to improve the model for each question, ensuring the model looked at the larger topic context and not just specific words when assigning topics.
• Model validation: An iterative process of running the model and reviewing the results was undertaken with subject matter experts, ensuring the model correctly identified key topics and assigned responses to the right areas for each question.
• Final output: The validated and improved model was applied to the responses for each question to determine the key topics and how the topics were distributed between different types of respondents.

3.1.8 Outputs from the model were then validated against individual responses for quality assurance, using a sample of 10% of responses for each question. Further analysis was undertaken which identified differences and commonalities in responses across respondent sub-types and sectors.

3.1.9 It was also recognised that there was the potential for the model to not identify unique responses as common topics. In order to ensure that these responses were captured in the overall analysis and reflected in this report, unique responses were identified in the topic identification and model validation steps and individually analysed by subject matter experts. Specific analysis of unique responses was provided for each question and is detailed in Sections 5 and 6 of this report.

3.1.10 Sections 5 and 6 of this report document the results of the analysis of the formal consultation and present the main topics and views expressed in the responses.