In: Math
the evaluation of effect of a training program on earnings using a non experimental study is likely to suffer from selection bias because person with higher earnings will always lead to a tendency of lower training and lower earning will lead to higher training and thus the test will not be correct and will show false inferences
random assignment of participants is done to negate any bias occuring due to social ,lab or field factors .randomly allocating the participants also reduces the chance of error due to treatment and false practices in some cases
the three objections to lab experiments are :
1 lab produces unrealastic data which lacks relevance for understanding the real world
2 the sample sizes are very small
3 there will be bias when experimenter expectation affects behaviour
attrition bias is a systemmatic error caused by unequal loss of participants from a randomized control trial.it is a type of selection bias and needs to sorted out by interviews or following up participants otherwise will give unfavourable results