1 Reply Latest reply on Nov 22, 2016 10:01 AM by diego.medrano

    Confidence Intervals

    Ridha Joudah

      Hello Everone,

       

      I got the following question in a homework. Would you please help me to answer it. The Excel data  file and an example for the required vis are attached with this discussion.

       

      Question 2

       

      Create confidence intervals for the proportion of students in soccer teams at each city as shown in the example view on the next page (the view shown is for your reference and not based on data provided to you) . Does the calculation of these intervals make the President of DCOF rethink his assessment of participation rates in soccer for various cities as compared to the simple bar chart in Question 1? Based on the intervals, which cities or groups of cities can be clubbed together regardless of the level of confidence (90%,95%,99%) used and what is the relative ranking of the groups as regards participation? Does your answer stay the same if you study the combined sample v/s per grade? Comment at will and in depth. Note: When creating confidence intervals for “proportions” we can use the following formula: p� ±          2 �p�(1−p�)      where, p� = estimated proportion having the attribute (in our case being in the soccer team) n = sample size and          2 = table value from the normal distribution at the α level of significance. Caution: The normal distribution approximation works well only if: (sample size * true proportion on a team) > 5. Since we don’t know the true proportion of students, an estimate for performing this check is: (sample size * sample proportion on a team) > 5 (which is effectively : number of students on team > 5 – but regardless, you should have this check embedded in your viz so that it can be clearly verified) Hints for Question 2: Please create the following calculated fields. The formulae below do not adhere to syntax but are for your reference. Please adhere to the field names in bold. % in Soccer Team: = In Soccer Team/(In Soccer Team + Not in Soccer Team) Sample size (or n): In Soccer Team + Not in Soccer Team Standard Error (                                     �p�(1−p�)      ) : Sqrt(% in Soccer Team * (1- % in Soccer Team)/n) Z Upper (values from the Normal distribution): Create a parameter as shown in confidence interval file discussed in class and posted online. Margin of Error: Standard Error*Z Upper Lower Limit: % in Soccer Team – Margin of Error Upper Limit: % in Soccer Team + Margin of Error np: n * % in Soccer Team (this can be used for the check hinted at in

        • 1. Re: Confidence Intervals

          Hey Ridha,

           

          Are you still looking for help with this question? If so, it'd be best to share what you've already tried or where you're running into issues. In the Community we like to make a point of teaching users how to use the product but not necessarily doing someone's work for them.

           

          Thanks,

           

          -Diego