Characteristics
The 'ideal' number of observations in a time series
100
Measure the trend -> intervention -> measure the effect
Simple interrupted time series
Limited to the data that was already collected
External validity
______ effect does not decay over time
Researchers make observations before and after a treatment, for both treatment and control group
Nonequivalent no treatment control group
The composition of the experimental group changes abruptly at the time of intervention
Selection
________ effect does not persist over time
Refer to board
Removing treatment at a known time
Outside forces that occurred prior to the time period of the treatment impact the dependent variable
History
In this type of effect, there is a noticeable change after the intervention
Can be used when time series with no treatment group is feasible
The intervention diffused slowly, with the exact diffusion rate and pattern being unknown
Statistical conclusion validity threat
The 3 dimensions of an interrupted time series
Form, permanence, immediacy
2 dependent variables, one of which treatment should affect, and the other one that treatment should not affect
Nonequivalent dependent variables
If a new police chief takes over, he might restructure the way crimes are classified leading to a shift in how things might be interpreted
Implementation bias