Probability and Non Probability Sampling
Part 1
Probability and Non Probability Sampling
Sampling is a statistical method of collection of data which can be used to deduce various conclusions. It is usually random and does not need initial planning over what will be collected and the materials to be sampled. Researchers have credited it as a reliable method of data collection that gives almost accurate results. However, the method has been known to only sample a small part of the actual data being dealt with. For instance, in sampling a people to get some solid information that can be used for further reference, most people may be unaware of the data collection by the time the findings are publicized. Similarly, it is necessary to consider a diverse group that may give various suggestions. Due to this fact, studies have proved that researchers rarely survey the entire population for two reasons. First, it may be too costly for them to fund the survey of a large group of people over a particular specified period of time. Secondly, the population and the nature of people changes with time, thus, sampling of a population too large may lead to spending of more time over which the perceptions of people might change, thus, giving differing findings. The methods of sampling may either be probability or non probability sampling.
Probability Sampling
(Mclntosh, 2008) defines probability sampling as the sampling method that requires usage of some form of random selection. He further argues that in order to have a random selection method, a procedure or a directive process must have been set up that all the selected units of the population intended have equal probabilities of being chosen. This is beneficial towards getting positive results since it eliminates chances of bias or tendency to rely on one sample more than the other. As the case above illustrates population, the collection of data from a certain population should ensue that the people selected to give the information had equal chances of being picked s as those who did not. The following of this principle will further ensure that the information is in no way manipulated or interfered with. Doherty, (1994) goes further to illustrate the steps that should be taken to achieve this objective by stating that, “In probability-based sampling, the first step is to decide on the population of interest, that is, the population we want the results about. This could be, for example, all persons aged 18 years or over who are resident in private households in New Zealand.” This helps to reduce confusion that may lead to a result oriented research which may compromise on accuracy and quality to achieve the results.
Probability sampling can be done using various techniques depending on the ones that may deem ft for the situation at hand. Similarly, the techniques depend on the subject matter being used and for what information it is used for. According to Mclntosh, (2008), “Sample random sampling is akin to pulling a number out of a hat. However, in a large population, it can be time-consuming to write down 3000 names on slips of paper to draw from a hat.” It is one of the techniques applied where the advantages and disadvantages such as the consumption of too much time are considered to compare with others. Cluster sampling is also a technique that researchers have used to gather information regarding matters affecting the environment and people. This stepwise process is useful for those who know little about the population they’re studying. First, the researcher would divide the population into clusters (usually geographic boundaries). Then, the researcher randomly samples the clusters (Mclntosh, 2008). The final process when using this technique is measuring the units in this cluster. It is worth noting that all the methods/techniques applied in the probability sampling are keen on advanced planning and procedure. Also, the measurement of the findings is put into consideration as an accuracy measure/step.
Non Probability Sampling
Non probability sampling is also a method applied by researchers as an option when collecting data through sampling. Unlike the probability sampling, the method obviously does not use random selection methods in its data collection. This is the most significant difference between the two methods. The methods used in this option specifically are very different from that of probability sampling. It also has different techniques that describe the collection of data. First, the method involves the use of convenience sampling. This entails the use of who’s available when the sampling is taking place. Unlike the probability sampling which would require equal probability of choice of either, this uses the ones available, thus has higher chances of bias and manipulation of data given. The other trait of non probability sampling is the purposive sampling. Here, the group selected is based on the purpose of the research.
The method has also been analyzed as one that exercises expert sampling. The data is not based upon who can give the information but rather, the experts for the intended opinion of study. In this method, a target is always set which the person responsible for collecting data is expected to keep going on until the target is reached. This aspect is commonly referred to as the quota date. The quota is normally issued by the relevant authorities and in cases where the research is independent; it is based on the desires of the researcher According to Doherty, (1994), “Quota sampling may be appropriate when there is no suitable list of population we are surveying.” The other significant and positive aspect of the sampling method is the diversity sampling. The information should be collected from a wide range of people who not only have the same features in common but those of different opinions and exposure. This way the results will be more reliable and realistic.
Part 2
Scales of Measurement
Data comes in various sizes and shapes and it is important to know about these so that the proper analysis can be used on the data. There are usually scales of measurement that must be considered in order to transform them into material that can easily be understood by a person who never took part in the data collection process. According to Heffner Media Group, (2003), “Statistical information, including numbers and sets of numbers, has specific qualities that are of interest to researchers.” Qualities such as magnitude, equal intervals, and absolute zero are used in the determination of the scale of measurement to be used, thus, evaluation of the statistical procedure deemed best. It is also worth noting that there relationship between the level of measurement and the appropriateness of various statistical procedures. For example, it would be silly to compute the mean of nominal measurements. However, the appropriateness of statistical analyses involving means for ordinal level data has been controversial. One position is that data must be measured on an interval or a ratio scale for the computation of means and other statistics to be valid. Therefore, if data are measured on an ordinal scale, the median but not the mean can serve as a measure of central tendency (Heffner Media Group, 2003). The four recognized scales of measurements include nominal data, ordinal data, ratio data and interval data. These have been research proven to give accurate results.
Nominal Data
This scale of measurement gives values that indicate qualitatively different kinds. It also puts three major aspects in to consideration. These include classification of data, arbitrary levels and that there are no orders. Its qualities are however not clearly defined by the statistical descriptions. Examples of the key issues that would be considered include names and lists of words (Heffner Media Group, 2003). For instance in the case of a particular political party in a country like Kenya, the names as are obviously recognized would be the fist to be analyzed since their evaluation would be the basis. This could be the Orange Democratic Movement (ODM), Party of National Unity (PNU) and NARC. An example of soil would include the identification of the three types which are sand clay and loam soil. These are the qualitative variables in the assessment or using the nominal scale of measurement to make sense of results/findings.
Ordinal Data
The recognizable scale quality of this measurement scale is the magnitude. This is the analysis of the size of the data that the researchers need to be made easily understandable. Here, the examples that could be used in demonstration are the likert scale and anything rank ordered. It is also a quantitative variable unlike the previously discussed nominal data which the researchers have proved to be qualitative. It also gives the actual degree of measurement of the data being analyzed. The category can be such as social class of a people or the sunlight magnitude. This can be described in simpler terms of magnitude where the reviewer can spot the difference. Social classes as an example can be either first, second or third class. This quantifies it for people to easily understand the disparity in the classes. The illustration of the sunlight magnitude can be full sun, full shade or partial shade. This elaborates the degree as to which the sunlight is experienced and felt. From the discussion, it is easier established that the components included in this scale are ordered, whereas the difference between the values is not an important factor. For instance, likert scales, rank on scale of 1…5 your degree of satisfaction (Heffner Media Group, 2003).
Interval Data
The scale quality in this category as a scale of measurement is the factoring of magnitude and equal intervals. Unlike the ordinal data, here the differences are recognized. Examples that are applicable in this area are such as temperature. It also gives the degree of distance, difference between values that can be specified. For instance, temperature is measure in centigrade. The ideal measurement here is usually 37 degrees, which can increase or decrease slightly depending on the environment and the health of the person. Where the temperature increases to 42 degrees, it is noticeable that the difference in 5 degrees. This difference is the interval data, thus measurable. It can also be used as a measurement variable which is mostly discrete and continuous. Researchers seeking to understand the application of this scale in measurement are yet to come to an agreement over whether it is more of continuous or discrete due to the fact that it can have constant scale but no natural zero.
Ratio Data
The ratio data as opposed to the interval has both the constant scale and the natural zero. Under most circumstances, it is the last stage of measurement where it analyzes and quantifies the findings to make it easier to understand. The scale qualities in this case include magnitude, equal intervals and the absolute zero. When keenly evaluated, we realize that it possesses the qualities of all the three aspects of the other measurements. Examples that may be used in this case are height, weight, age and percentage. This are the most commonly used statistical measure in every day life, thus proving the popularity of this scale over most other. Length is also another specification of this category.
References:
Doherty, M. (1994) Probability versus Non-Probability Sampling in Sample Surveys, The New Zealand Statistics Review, pp 21-28.
Heffner Media Group. (2003). Descriptive Statistics. Research Methods. Retrieved from http://allpsych.com/researchmethods/measurementscales.html
Mclntosh, J. (2008). Probability Sampling Techniques. Retrieved from http://scientificresearchmethods.suite101.com/article.cfm/probability_sampling_techniques
