Issues with small sample sizes

[Tord74]

Hi there, 

I've built a range of visuals, based on anonymous staff survey data but need to manage, staff anonymity but at the same time allow for as much granular analysis as possible.  I have demographic filters (Gender, Job role, experience etc), but the issue i have currently is that some total samples will be much bigger than others i.e. 300 vs 50. At the moment I've fixed you can only filter on one dimension, which is limiting for the analysis. 

Ideally I'd like to ensure there are at least 5 respondents on at any time, but grey out the filter option if the numbers don't allow. 
Any ideas appreciated. Thanks 

[v-kaiyue-msft]

Hi @Tord74 ,

May I ask if the guarantee of at least 5 respondents refers to the fact that there are at least 5 entries left in the filtered data?

"At the moment I've fixed you can only filter on one dimension, which is limiting for the analysis." may I ask what is meant here? What have you fixed?

Can you share sample data and sample output in table format? Or a sample pbix after removing sensitive data. We can better understand the problem and help we can better understand the problem and help you.

Best Regards,

Clara Gong

If this post helps, then please consider Accept it as the solution to help the other members find it more quickly.

[Tord74]

Hi Clara 
Thanks for the response. Yes, I'm looking to ensure there is always 5 entries left in the filtered data. 
We have developed a feedback and scoring system from a staff feedback survey.   
On the top right hand side, we have 'filters' that allow you to segment by user demographic and then a range of answers to questions are shown in various formats. 

The concern is with smaller user bases, that the user could filter to such a granular level they can identify individual respondents.  So for now, we've 'grouped' some of the demographic segmentation to a more top line level (which is not ideal) i.e experience instead of saying (Under 2 years, 3 to 5 years, 5 to 10, 10+ etc) we've said Under 10, Over 10.  

Ideally i'd like to make the segmentation more dynamic, so for the larger sample sizes you can segment in more detail, but if you get to a point where you're filtering reduces the sample to under 5 you get a message saying sample size too small.   Any tips gratefully appreciated.. thanks 

See screen shot below. 

[v-kaiyue-msft]

Hi @Tord74 ,

You can try using the count and if functions to achieve this, similar to the following.

Can you share sample data and sample output in table format? Or a sample pbix after removing sensitive data. We can better understand the problem and help you.

Best Regards,

Clara Gong

If this post helps, then please consider Accept it as the solution to help the other members find it more quickly.