Survey results in a Dashboard - advice needed

[michaelsh]

Hello friends,

I am building a dashboard for school parents' survey.

It has many multiple-choice category questions as well as scale (1-7) questions.

I'd like some advice mainly on 2 things:

1. What is the best way to present the results?

I don't want to only use bar/pie charts, but also to bring some insights - such as comparison or trends...

What are best practicies?

I would appreciate any samples.

2. Is there an efficient way to automatically show new questions if they are added?

(this is secondary, most of the survet stays unchanged)

Thank you!

[BA_Pete · Accepted solution]

Hi @michaelsh ,

In terms of presentation, this is incredibly hard to answer. This is based on your own style as a designer, the receptiveness to different styles of your end-users, the gravitas of the information you are reporting, as well as the story you are ultimately trying to tell with the output.

However, I would say that you shouldn't be too quick to write-off bar/column charts. While they may seem 'basic', most laymen are perfectly comfortable reading them and understand what they are being told. Bar charts (not column charts) are particularly good at displaying comparisons between values as the horizontal orientation is more conducive to the binocular vision of the human eyes to discern. This can't always be said for some of them more 'flashy' visuals. You are, however, correct to write-off pie charts. They are rubbish and make it very hard for the human eye to distinguish between similar values.

In terms of insights, I think anything that compares your score values to parent age, number of children, and occupation type should be quite interesting. This should give some insight into whether you should be offering more support to older/younger parents in full/part/self-employment with more/less kids. These are the kind of insights that could easily be utilised to apply predictive/proactive actions.

Additionally, if these surveys are an ongoing occurrence, you should definitely do some 'over time' analysis i.e. showing the changes in response values as time goes by. This type of visualisation is essential in order to track whether changes/actions previously taken are having the desired effect.

Regarding the dynamic absorption of new survey questions: with your data in Power query I would select the date column, go to the Transform tab, then hit Unpivot Columns > Unpivot Other Columns. This will 'normalise' your data structure, creating two new columns [Attribute] and [Value]. You can change the [Attribute] column name to "Question" or similar. This should make it easier to write more generic measures where you only need to reference the [Question] field, rather than a specific-question-named field.

Hope this makes sense.

Pete

[v-shex-msft · Accepted solution]

Hi @michaelsh,

AFAIK, current power bi not suitable for these operations, it does not contain write-back features so that it not able to collect or record the survey results from your customers.

In my opinion, I'd like to suggest you use other products(e.g. ms form, power app) to collect these results and use power bi to get data from them to show the resulting graph in charts.

Microsoft Forms - Collect better data and make better decisions. 

How to Automatically Import Microsoft Forms Responses to Power BI 

Connect MS Forms with Power Apps 
Regards,

Xiaoxin Sheng

[BA_Pete · Accepted solution]

Hi @michaelsh ,

In terms of presentation, this is incredibly hard to answer. This is based on your own style as a designer, the receptiveness to different styles of your end-users, the gravitas of the information you are reporting, as well as the story you are ultimately trying to tell with the output.

However, I would say that you shouldn't be too quick to write-off bar/column charts. While they may seem 'basic', most laymen are perfectly comfortable reading them and understand what they are being told. Bar charts (not column charts) are particularly good at displaying comparisons between values as the horizontal orientation is more conducive to the binocular vision of the human eyes to discern. This can't always be said for some of them more 'flashy' visuals. You are, however, correct to write-off pie charts. They are rubbish and make it very hard for the human eye to distinguish between similar values.

In terms of insights, I think anything that compares your score values to parent age, number of children, and occupation type should be quite interesting. This should give some insight into whether you should be offering more support to older/younger parents in full/part/self-employment with more/less kids. These are the kind of insights that could easily be utilised to apply predictive/proactive actions.

Additionally, if these surveys are an ongoing occurrence, you should definitely do some 'over time' analysis i.e. showing the changes in response values as time goes by. This type of visualisation is essential in order to track whether changes/actions previously taken are having the desired effect.

Regarding the dynamic absorption of new survey questions: with your data in Power query I would select the date column, go to the Transform tab, then hit Unpivot Columns > Unpivot Other Columns. This will 'normalise' your data structure, creating two new columns [Attribute] and [Value]. You can change the [Attribute] column name to "Question" or similar. This should make it easier to write more generic measures where you only need to reference the [Question] field, rather than a specific-question-named field.

Hope this makes sense.

Pete