Not_Analyzed Field Splits On Spaces

I have a field marked as type string, not_analyzed. When visualising data from this field each word is separated.
It is as if the field were mapped as analyzed.
I have another index which has the same field not_analyzed and visualisation does not separate the words in the string.
Perhaps a bug?

May need more details here. Could you post the mapping in ES and a screenshot of how you are creating the visualization?

Just a simple mapping:
(extract)
},
"source_name" : {
"index" : "not_analyzed",
"type" : "string"
}

With the visualisation it is just a basic pie or bar chart.
So, with a bar chart using the timestamp for X-Axis and a Split Bar Sub Aggregation using terms and them 'source_name' field for example.
If the value of the source_name field is "my own application" then the bar will be split into my, own and application.
The same with a pie chart. It's as if the field is mapped as analyzed.

I've not seen that happen. And you say you have another index where this works as expected? Any errors on the Elasticsearch side?

I'm seeing this issue as well. In my case it also extends to the queries being run directly on elasticsearch. In my example it's a field being generated from 3 copy_to statements

        "field_1": {
            "type": "string",
            "fields": {
              "raw": {
                "ignore_above": 256,
                "index": "not_analyzed",
                "type": "string"
              }},
            "copy_to": "field_123"
        }, 
        "field_2": {
            "type": "string",
            "fields": {
              "raw": {
                "ignore_above": 256,
                "index": "not_analyzed",
                "type": "string"
              }},
            "copy_to": "field_123"
        }, 
        "field_3": {
            "type": "string",
            "fields": {
              "raw": {
                "ignore_above": 256,
                "index": "not_analyzed",
                "type": "string"
              }},
            "copy_to": "field_123"
        },
        "field_123" :{
            "type": "string",
            "index": "not_analyzed"
        }

I'm seeing this issue. Have you found a solution to this?