Scales are the indexes that measure those types of variables that are not directly observed but are instead inferred from the other variables that are directly measured.
One of these types of scales, called the Likert scale, is the most popular type of scale. This scale is generally used while carrying out a survey. In a survey item on a Likert scale, response categories may be responses such as ‘strongly agree,’ ‘agree,’ ‘don’t know,’ or ‘disagree.’
Another scale is called the Guttman scale. In this scale, the items are made in a single dimensional series. This is so that the response given by the respondent for a certain item predicts the responses for all of the previous items in that series. For example, according to the assumption of this kind of scale, if the respondent responds to a particular question in a positive manner, then it is predicted that the respondent will also respond to a less complex or a less sensitive question in a positive manner.
There is another type of scale, called a Mokken scale. Since Guttman scales are quite rigid and determinable, Mokken scales are available to do what Guttman scales cannot. The major difference between these two types of scales is that the Guttman scales are deterministic, while the Mokken scales are probabilistic. In the previous example, according to the assumption of Mokken scales, if the respondent responds to a particular question in a positive manner, then there will be a significantly greater probability that the respondent will respond to a less complex question in a positive way as well.
In another type of scale, called a proximity scale, if a respondent responds in a positive manner on a certain item, it is not necessarily assumed that the respondent will respond positively to a less difficult question. Such a scale implies that if the respondent answers a difficult question, then that respondent might not wish to answer the less difficult question (some reasons for this include the respondent not being in a good mood, or some other psychological reason).
Standard error is the standard deviation of the sampling distribution of a statistic. Standard error can also be abbreviated as S.E. Standard error plays a very crucial role in the large sample theory. Standard error also forms the basis for the testing of a hypothesis. The statistical inference involved in the construction of the confidence interval is mainly based on standard error. The magnitude of the standard error gives an index of the precision of the estimate of the parameter. The reciprocal of the standard error is generally taken as the measure of the reliability or the precision of the statistic. In other words, the standard error is inversely proportional to the sample size. This means that the greater the standard error, the smaller the size of the sample. Thus, in order to double the precision, the standard error should be reduced to one half, and for this, the sample size should be increased to four times the original size. The standard error of a sample is generally designated by the Greek letter sigma (σ). The standard error can also be defined as the square root of the variance present in the sample.
Contact Statistics Solutions today for a free consultation on scales and standard errors.
