DAX not matching for single selected records and matching with select all records

[Anonymous]

Hi , I have created DAX and the data is about employees 

I created a DAX which is working fine 

If i select the single record in slicer option it is showing wrong data 
If i select all it is showing correct data 
The shown data is taken in Card Visualisation

Thank you in advance

[v-alq-msft]

Hi, @Anonymous 

I wonder what your dax is like, which is important to solve the problem.

Best Regards

Allan

[Greg_Deckler]

Please see this post regarding How to Get Your Question Answered Quickly: https://community.powerbi.com/t5/Community-Blog/How-to-Get-Your-Question-Answered-Quickly/ba-p/38490

[Anonymous]

Hi @Greg_Deckler 

Please find the below Measeures i have used and let mw know why am i getting wrong values

ENDINGS = [STARTINGS]+COUNT(OnBoard[Date of joining (MM-DD-YYYY)])-(COUNT(Separation[Exit Date]))

Startings = [Last Period check]+[Last Period Employee]

Last Period Check =

var _min = minx(ALLSELECTED('OnBoard'),('OnBoard'[Date of joining (MM-DD-YYYY)]))

Return

CALCULATE(COUNT('OnBoard'[Date of joining (MM-DD-YYYY)]),filter(ALL('OnBoard'),'OnBoard'[Date of joining (MM-DD-YYYY)]< _min))

Last Period Employee =

var _min_date = minx(all('Calendar'),'Calendar'[Date])

var _Expression=if(ISFILTERED('CALENDAR'[Month Year]),maxx('CALENDAR',ENDOFMONTH(DATEADD('CALENDAR'[Date],-1,MONTH))),maxx('CALENDAR',DATEADD('CALENDAR'[Date],-1,YEAR)))

Return

CALCULATE(COUNTx(FILTER(Separation,Separation[Date of joining]<=_Expression && Separation[Date of joining]>=_min_date && (ISBLANK(Separation[Exit Date]) || Separation[Exit Date]>_Expression)),(Separation[Employee ID])),CROSSFILTER(Separation[Date of joining],'CALENDAR'[Date],None))

 Here in the total I have a a total count from Measure Endings and that dax is working fine. and for each individual division it is showing wrong values .Please help on this

Division	ENDINGS
	187
	187
a	191
b	189
c	188
d	239
e	209
f	198
g	191
h	188
Total	284

[Greg_Deckler]

@Anonymous , Those are reasonably complicated measures. It is going to be difficult to solve this without sample source data posted in text and then a definition of what wrong and right is.