In the Data Grid, when checking a few indicators, without checking their parent indicators, the order of the indicators shown as top-level nodes of the indicator tree is unexpected.
The indicators in SERAM form a directed graph without loops, and not a tree. The tree representation is built by adding all children of a node recursively, which can lead to the situation where one distinct indicator is repeated multiple times in the tree representation.
For instance, if there is a calculation (with "show details" enabled) using a value and an emission factor, and another similar calculation which is using another value but the same emission factor, this emission factor will appear twice in the displayed tree.
When selecting indicators, checking one of these indicators that appear multiple times will check and uncheck all visual occurences at the same time, since they effectively represent the same indicator. In consequence, there is no deterministic selection order.
In the loaded Data Grid, the top level of indicators are ordered by their sort key. The sort key can be configured in the indicators' metadata, and when left empty it will use the indicator code for sorting (in contrast to names the code is language-independent and therefore deterministic).
By defining a meaningful order and setting this in the indicator sort key, you can influence the order to match yor needs.