Hi there, so I have this survey sample that I need to see if I can get some valuable insights from it. I have not worked with this type of data before, so I am looking for some help with this.
The negative values are replacements for N/As.
Hope you can help, thank you!
Sample_2.zip (331.3 KB)
Not knowing anything about your business operations, I can’t say what would or wouldn’t be valuable insights. However, here’s what I think is a pretty good roadmap for getting there:
The number and type of analyses you can run from a big survey data set like this are nearly infinite. Thus, to focus your efforts I would suggest developing a formal Analysis Plan:
Attached is an example analysis plan I developed as part of our church strategic planning process to focus our assessment of a huge survey data set we collected. Obviously, a very different context, but developed along the same principles as above.
I hope this is helpful.
Thank you, Brian!
I’m more interested in the practical how-to in Power BI with this dataset.
As you’ll see in the pbix file, I’ve created a visual that provides some insight but I am not sure whether the techniques I’ve used are appropriate or not. In this visualize, the organization wants to see if there’s a difference between male and female responses with respect to whether the answer provided by the company representative was helpful or not.
So, before I continue creating numerous DAX Measures for each question, I need to get a sense that I am approaching the visualization of the data in an appropriate manner, or not.
Also, for a larger number of survey questions, the number of DAX Measures required would grow substantially, so how to deal with that eventuality as well?
Wow – sorry. I totally misinterpreted what you were looking for in my first response. Let me get back in the batting cage and take another swing:
Chi-square is the right test for addressing this question. Two questions for you:
How committed are you to doing this in Power BI? It’s doable in power BI, but executing the analysis and visualization(s) will be a LOT more work in Power BI than in R. (I think I recall from some of our previous conversations that you have experience with R).
How statistically/analytically sophisticated is your audience? There are some good visualization options for chi-square results that are easy to do in R (I really like balloon plots for this purpose), but not possible in Power BI w/o R or Python scripting . However, if your audience is not versed in basic statistical analysis, I would stick with simple bar charts showing actual versus expected observations of non-helpful responses by gender and company (plus one analysis for all companies aggregated together), and then maybe put a card visual above each set of bars indicating whether the result was statistically significant at the chosen p value (probably .05).
You are absolutely right that even a moderately scaled analysis of this dataset will generate a boatload of measures (compared to a single R script for all the analysis AND visualizations). If you’re committed to Power BI for this, I’d recommend creating separate measure tables for each group of analyses (e.g., gender, age, location, etc.).
Hopefully, this is more helpful than my first response…
Hi @powerbideveloper, a response on this post has been tagged as “Solution”. If you have a follow question or concern related to this topic, please remove the Solution tag first by clicking the three dots beside Reply and then untick the checkbox. Thanks!