MagicMock returns wrong return_value in python

When I only run the following test it works perfectly fine. But when I run all tests within its test class at the same time, it fails because a wrong return value is being given by data_backend.read_month_data. The wrong data is a dict, which is being specified as return_value in multiple previous tests.

def test_get_statistics(self):
    mock_month_data = {
        1: {
            "PyCharm": 4000,
            "IntelliJ": 6000
        }
    }

    data_backend = MagicMock()
    data_backend.read_month_data.return_value = mock_month_data

    data_repository = CodeTimeDataRepository(data_backend)
    result = data_repository.get_statistics(datetime.date(2020, 1, 1))

    expected_result = {
        "date": "Jan 01 2020",
        "total_time": 10000,
        "activities": [
            {
                "name": "IntelliJ",
                "time": 6000,
                "progress": 0.6
            },
            {
                "name": "PyCharm",
                "time": 4000,
                "progress": 0.4
            }
        ]
    }

    self.maxDiff = None
    self.assertDictEqual(expected_result, result)

The mock object is set up in the previous tests using the following code. self.get_default_month_data() returns a dict looking similar to mock_month_data, but with different data.

    data_backend = MagicMock()
    data_backend.read_month_data.return_value = self.get_default_month_data()