Random Error: It is caused by any of the factors that have a random effect on the measurement of the variables across the sample. To explain better by an example, the performance if a person can pick up degrade by the effect of their mood. In a particular kind of testing on children the mood of n the children will have an effect on the measure of performance and artificially inflate or deflate them from the others. An important characteristic of random error is that its consistency does not remain same across the sample. It does the task of pushing scores up or down in a random fashion. It means that if the all the random errors in a distribution could be seen, the sum of them would be equal to zero. The number of negative errors is same as the number of positive errors. Though Random errors add variability to the data, what is important is that it does not affect the average performance of the group. It is because of this that sometimes random error is also called noise.
Systematic Error: It is the kind of error that is caused by any factors that have a systematic affect measuremennt of the variables that are there across the sample. It can be explained by an example that says that if there is loud disturbance or noise outside the classroom the scores of all the students would be affected. As opposed to random error, it has a more systematic characteristic of either being consistently positive or negative. It is because of this trait, it is sometimes considered to be bias in measurement.
Reducing the Error: Whether the error is systematic or random, there are certain means to reduce it. One thing that can be done at a preliminary stage is to pilot test the instrument and get feedback from the respondents so as to understand how easy and hard they found it and whether the testing environment would affect their performance. The second way to reduce the error is to train the people who are involved in data collection. The third important thing to keep in mind is to cross check the data thoroughly, particularly when it is being entered into the computer. Some statistical measures can also be used for adjustment of the measurement errors.