Type 1 Error - Not really an effect.
Type 2 Error - Actually is an effect.
External reliability is assessed with the test-retest method. The experiment is done several times with the same or similar participants.
Internal reliability is assessed with the split-half method. Compare one half of the results to the other to see if scores are consistent.
*Take more than one measurement for each P.
*Conduct a pilot study.
*Collect and record data in a standardised way. Improves Inter-rater reliability.
Internal validity is improved with lots of control.
I.e. a lab study.
But these methods have their own problems.
Demand characteristics can be tackled with the Single-blind technique. Ps do not know which condition they are in.