Page 170 - Data Science class 11
P. 170

• Non-response bias: Chances are high that there will be individuals who are disinterested or unable to take the survey.
          This causes discrepancy between the respondents and non-respondents, resulting in the so-called nonresponse bias.
           • Response bias: Merely responding to the survey is not enough, respondents must also provide accurate and honest
          responses.  Less-than-truthful survey responses  can come as an outcome of both conscious  and subconscious
          cognitive factors.
           • Question order bias: The order of both questions and answers could cause survey respondents to provide biased
          answers. Sometimes, the initial questions of a survey could influence the answers respondents give to the subsequent
          questions later on in the survey.
           • Information  bias: Information bias is  caused due to misrepresentation of truthfulness that occurs during  the
          collection, handling, or analysis of data in a research study.

        Avoiding Sampling Bias
        Using careful research design and sampling procedures can help you avoid sampling bias.

        Here are three ways to avoid sampling bias:
           • Use simple random sampling. Probably the most effective method researchers use to prevent sampling bias is through
          simple random sampling where samples are selected strictly by chance.
           • Use Stratified Random Sampling.

           • Avoid Asking the Wrong Questions.


        4.5 LeveL Of COnfiDenCe: HOW Sure are yOu?
        When you ask someone, "How sure are you?" you try to gauge the level of confidence with which that person is
        putting forward an observation. With reference to this, let us now learn about confidence interval.

        4.5.1 Confidence interval

         In statistics, the term used to measure the accuracy of a result is called the confidence interval.
        When we estimate a population parameter, it is a good practice to give it a confidence interval. A confidence interval
        communicates how accurate our estimate is likely to be.
        Confidence, in statistics, is another way to describe probability. A confidence interval refers to the probability that a
        population parameter will fall between a set of values for a certain proportion of times. For example, if you construct
        a confidence interval with a 95% confidence level, you are confident that 95 out of 100 times the estimate will fall
        between the upper and lower values specified by the confidence interval.

                                 Lower Limit                                  Upper Limit
                                    2.92                                         5.62
                                         95% chance your population mean
                                           will fall between 2.92 and 5.62















                           2.5 %                                                        2.5 %
                          outliers                                                     outliers


                                                     Mean = 4.27
          168   Touchpad Data Science-XI
   165   166   167   168   169   170   171   172   173   174   175