Page 183 - DMGT209_QUANTITATIVE_TECHNIQUES_II
P. 183

Quantitative Techniques-II



                      Notes         5.   Semantic Differential Scale is very similar to the ....................... Scale.

                                    6.   The Likert Scale consists of two parts – .................... and ..........................

                                    9.3 Criteria for the Good Test

                                    There are two criteria to decide whether the scale selected is good or not. They are shown in the
                                    diagram given below:
                                                           Figure 9.2: Criteria for the Good Test

























                                    Reliability

                                    For a research to be sound, reliability is an important criterion. Reliability means, ability of a
                                    scale to produce a consistent result if repeated measurements are taken. Reliability is the extent
                                    to which scales are free from random error and produce consistent results. Lesser the random
                                    error, the data will be more reliable. Random error is not a constant error but, it depends on the
                                    person using the measuring instrument or measurement situation. A random error leads to
                                    inconsistency in measurement; when the measurements are made on the same objects or persons.
                                    An example of random error would be the use of elastic scale to measure a person’s height. If 2
                                    successive measurements are made, the person who is measuring would stretch the elastic ruler
                                    to different degree on 2 successive occasions, therefore, the height measurement would never be
                                    the same, even though the person’s height has not changed.
                                    In addition to the random error, there is one more error namely systematic error. It is also called
                                    as non sampling error. This error includes all types of errors except random sampling error.
                                    Therefore, we can say
                                    Measurement result = true measurement + measurement error
                                    Measurement error = true measurement + systematic error + random error

                                    Test-retest Method

                                    There is an approach called test-retest to check the reliability. In this approach, respondents are
                                    given identical sets of scales at two different points of time under almost identical conditions.
                                    The time interval is between 2 to 5 weeks. The similarity between 2 measurements is determined
                                    by calculating the correlation coefficient. Higher the value of correlation coefficient, greater the
                                    reliability.



            178                              LOVELY PROFESSIONAL UNIVERSITY
   178   179   180   181   182   183   184   185   186   187   188