You are doing a conceptual mistake here, but don’t worry you are in good company, according to this article:
The example with the coin is useful to explain what is the mistake. As opposite the the coin, the elementary particles can be in a quantum superposition: https://en.wikipedia.org/wiki/Quantum_superposition
Observation is a misleading word, let’s use the word interaction. Quantum particles can be in a superposition state until there is an interaction with those particles.
For example, let’s assume that we have qubit state of 1 and 0. When you interact with the qubit to “read” if it is in state 1 or 0, you’ll get a 1 with a probability p, and 0 with 1-p. But, before the interaction the qubit is in a superposition of 1 and 0. This allows quantum computing: it is possible to influence the probability in which the state of a set of qubits will decay when performing a reading (A.K.A. interacting with the qubits). So, basically, it is possible to increase the probability that a set of quantum qubits will provide the correct solution to an algorithm when interacted for a reading.
A particle with a defined hidden state (the coin example) behaves differently from a particle in a quantum superposition.
In the double slit experiment event 1 single photon will produce an interference pattern, as if it was able to pass through both slits. If you try to measure (thus interact) with the photon, the interference pattern will disappear.
