yes, whether the detected difference is truly there and whether

yes, whether the detected difference is truly there and whether the detected difference is due to the effect of the independent variable they are manipulating are two very different questions. If a significant difference is detected and you suspect it's a measurement error due to variability, it's type I error. If a significant difference is detected and you wonder whether the difference is truly due to the effect of independent variable or a statistical artifact of a spurious relationship which could be cause by a slew of factors, some of which are addressed by veggiedoggie, that's a totally different issue. You should be examining your experimental design and the internal validity. If your experimental subjects are human beings, this could very well be due to observer-expectancy effect.