I'm a bit confused about autocorrelation.
I understood that I should not expect my residuals to be autocorrelated.
But what about the time series data itself, should I expect to find autocorrelation in them?
Tank you in advance.
It depends on the measure used in the times series. If you were tracking GDP, which is a measure of the performance of the economy, you would expect to see autocorrelation over shorter time periods. Whether you are in a boom or a depression/recession, consecutive measures of GDP would typically be dependent upon each other. However, a counter example where you would typically not expect autocorrelation would be a dimension measured from a manufacturing process. Most of these (though not all) would exhibit random variation around a mean value over time, unless disturbed by an external noise variable. Consecutive measurements are independent of each other.
In addition, you have likely already read up on this - but the concept of "lag" comes into play. For example for environmental data, every winter temperatures drop. In addition, you can see autocorrelation in multilevel models where values are clustered together by a group variable. For example imagine geographical locations. In the US some large cities can tend to be democratic in political views and as you move away from the city toward more rural areas the political leanings toward the democratic party may attenuate.