Posts tagged quantitative

Introduction

The Coronacrisis continues to create a massive impact on global physical and mental health. Our research team in Industrial Design, based at the Delft Institute of Positive Design, aims to make it easier to understand wellbeing needs during this time. Therefore, over the past few months, our team has helped TU Delft gather anonymous wellbeing information from thousands of socially isolated students and employees. With more insight into self-reported health and wellbeing, we want to empower the administration to take smart actions to help struggling students and employees.

This post will focus on our findings from the TU Delft Staff Wellbeing Survey. Using a participatory design process, we iteratively designed 24 survey items to assess the needs of TU Delft Staff members. In June 2020, this survey was sent to over 6,500 TU Delft staff members. Of those who opened the survey, over 85% completed it. This gave us responses from over 2,700 staff members, which include support staff, scientific staff, PhDs and others. This brief report aims to share some of the key highlights of our quantitative findings; a separate report will cover our key highlights from the qualitative free response data.

What’s the effect on workload?

How has the Coronacrisis affected workload? 30% said that their workload has increased a little, 18% said their work increased by a lot, 40% said it had no change and only 12% said their workload had decreased.

How are people doing?

A central measure in our survey is the measure of overall “Life Satisfaction”. This item is used in a variety of surveys around the world as a measure of subjective wellbeing. Below, we show the scaled response questions that asked people to choose a number from 0-10, along with the average response. The graphs show the normal distribution of responses across all of the scales. This shows that most staff at Delft are generally satisfied: the average score was 6.8 out of 10 and only 17% of staff members gave a score of 5 or lower. 

Beyond life satisfaction, we also investigated satisfaction with working at TU Delft, physical health, working from home, etc. Staff generally rate working at TU Delft very highly: there was an average score of 7.8 and only 5% gave a score of 5 or lower. Physical Health is also strong, with an average score of 7.4. However, 10% of staff rated their physical health as 5 or lower, which shows a population of people who are struggling with their health. Working from home, however, stands out for its wide spread. Some people love it and some people hate it. This had an average score of 6.2 out of 10 and over 30% gave working from home a score of 5 or lower. 

How does language and function affect wellbeing?

One way to break this down further is to look at the effect of job function and language. Across the groups, average life satisfaction varies from 5.9 (for non-dutch PhD students) to 7.0. Most notably, PhD students are struggling much more by the working from home situation.

Which other factors most affect wellbeing?

We asked a number of questions in the survey about people’s individual situation. Which of these additional factors best predicts their subjective wellbeing? We created a regression model using all of the other items in the survey to predict “life satisfaction” (our measure of subjective wellbeing). 

Our model showed that the most predictive factor was the person’s rating of working from home, followed by their physical health and then their rating of working at TU Delft. Other factors that were highly predictive of life satisfaction were items related to stress, loneliness, fatigue, balance, engagement, optimism, personal autonomy and a person’s home situation (e.g., are they living alone or with partners, children, etc). One thing that wasn’t predictive in this model was the particular faculty where people work. 

For the statisticians out there, p<.0001, RMSE=1.228, RSquared=0.51. In the figure above, only the most predictive factors are shown. LogWorth can be intuitively understood as how much a particular item contributed to the prediction in the context of the other items. So, even though a factor might not be predictive in this particular model doesn’t mean it isn’t important or insignificant. 

Conclusion

We are all affected by the ongoing Coronacrisis. Our statistical analyses show how different circumstances are affecting different groups of staff at TU Delft. These results can help the administration gain a broad and diverse perspective.

However, our goal is not just to create perspective — it is to catalyze useful actions that help improve wellbeing in our community. How do we plan to transform our survey data into this action? We have recently brought together diverse staff to participate in “wellbeing design workshops.” These workshops are based on each participant’s qualitative analysis of the thousands of written responses in the survey. By having a broad group of people read the anonymous responses, we can ensure that we are truly listening to the voices in the community.  Statistics can reveal much — but so can reading about personal stories, which bring color, detail and ideas of how to help. In a separate post, we will share the organization of these workshops and our findings. 

Stay tuned — and know that we continue to recruit staff members who would like to volunteer to help support the analysis and design. Do you have ideas and suggestions for how Delft can respond? What would you like to be asked in the next survey? Let us know.

You can email me personally at j.d.lomas@tudelft.nl. Put “Wellbeing Survey” in the subject line. Thank you!