Sample Populations - Speech or Presentation Example

Can one person’s views be representative of the collective view of several employees, as was in this case? Could the single employee surveyed truly represent the larger population? The answer to both is most certainly no. Different employees perform different roles in the organisation and as such are exposed to different external stimuli that shape their opinions and experiences with the organisation. In statistics there are two components of validity – external validity and internal validity – that are used to see whether a particular study is valid or not, where a valid research is that which uncovers principles and facts that explain or predict (Marion, 2004).

In this instance, the sample of a single employee fails the external validity criteria. According to Trochim and Donnelly (2006), external validity refers to the extent to which the conclusions in this study would hold for other employees in other places and at other times. This survey violates all three threats to external validity which are people, places and times. The single employee selected for this survey could have been selected for being an extraordinary candidate (person) or he/she could have been selected because he works under marketing planning (place) or the results are only true because the surveyed employee works in the early morning shift (time).

Lastly, a sample is a part of a whole which means that for one to generalize the results from the sample have to be extrapolated which leads to two kinds of errors: sampling errors and non-sampling errors (Freedman, n.d.). Sampling error results when we get a few too many units of one kind, and not enough of another. In this instance where we have only one employee as the sample, sampling error is very high. On the other hand, non-sampling error – often referred to as bias – is also high because of selection bias and response bias.

Selection bias is a big issue here because there is no indication that the employee was chosen randomly. Though not guaranteed we also believe that in this instance the response bias would be high because the interviewer could easily have led the respondent to influence the results. Recommended actions to correct inadequacies in the sample To correct the inadequacies of the sample above the first activity would be to select a sample fairly to represent the population which in this case is the company.

According to Freedman (n.d.) the best methods for choosing a sample involve use of probability methods. Probability samples minimize bias which is a serious problem in applied work. To draw a probability sample start by identifying the population of interest then create the list of units to be sampled (sampling frame). Considering that a company often has a structure that defines employees either by function (e.g. accounts, human resource, manufacturing etc.) or job group we would suggest use of stratified random sampling.

Here, we would first divide the company’s employees (population) by say function or department or shift (into homogeneous groups) then take a simple random sample in each group. According to Trochim and Donnelly (2006) the key benefits of stratified random sampling are: (1) it will ensure that not only the entire population is represented but that key subgroups are