The impact of the physical workplace on behaviors and attitudes at work is a much-studied topic. Major research streams over the last decades investigated either satisfaction with offices in relation to physical comfort, or how layout decisions influenced interaction and collaboration in the workplace with a focus on open-plan offices. Rather little is known on the effect a workplace layout (such as its openness) has on perceptions of staff regarding teamwork, focused work and perceived productivity. We aim to close this gap by taking a differential approach which appreciates detailed variations within open-plan offices. Not every corner of an office is the same, so the question arises whether satisfaction with workspace differs depending on where someone is sitting. Bringing results of a staff survey in the UK headquarters of a global technology company together with a detailed analysis of spatial qualities at desks based on isovist and visual field analysis, we find that staff are less likely to rate their workplace environment favorably when they have higher numbers of desks within their own field of vision; and when they are facing away from the room with a relatively larger area behind their back compared to the area surrounding them. Aspects of teamwork that are negatively affected include sharing information with others, as well as team identity and cohesion. Focused work (concentration) and working productively are impacted even more so with the largest effect sizes throughout. These findings highlight the relevance of investigating detailed spatial qualities of micro-locations in workplace layouts. Our results also raise important questions regarding the current popular practice in workplace design of providing large open-plan offices for technology companies.

Fieldwork was undertaken in spring 2018 in the UK headquarter of a global technology company. In the study we collected floor plans of the office (including the detailed desk numbers and seating position of staff) as well as responses to a workplace satisfaction survey. Participants were asked to rate how much the working environment supported or inhibited their working life, for instance regarding sharing of information with colleagues, team identity and cohesion, concentration and productive work. 167 responses were obtained.

Based on space syntax as a method and the concept of an isovist, in the analysis we calculated visual field information from the seats of participants in the office, e.g. how many desks someone is directly surrounded by (covering a full 360 degree view from someone's desk), how many desks are directly visible in someone's forward facing vision, etc. We also classified desks by type according to their micro-location, e.g. next to a window, corridor, wall or mid row. These spatial characteristics of a seat were matched with survey responses in order to answer our research question whether office workers perceive teamwork, focused work and productivity differently according to the spatial features of their desk. We investigate this relationship between desk features and job satisfaction using multiple ordinal regression models.

Due to restrictions on data sharing by our Ethics Committee, we cannot share survey responses but the repository contains floor plans (in dxf format) as well as the R script we used to undertake the analysis. This allows researchers to replicate our method with their own data sets.

To replicate our analysis, the following steps are required:

1. Prepare a dxf file of a floor plan (using the floor plan we have provided as an example)
2. Extract spatial information from the dxf file in R using the R package dxfspatial. See: https://github.com/pklampros/dxfspatial
3. Calculate isovist properties (for example area of 360 degree isovists from all seats) using the R package rdepthmap. See: https://github.com/pklampros/rdepthmap
4. Collect data from a workplace survey including seating positions of respondents (see file survey_questions.pdf)
5. Bring spatial metrics and survey responses together in a single csv file (following the structure in the file compact-data.csv, which contains a single made up sample answer)
6. Use the R script provided as R Markdown (generators_rev1.Rmd) to run the analysis.