In: Statistics and Probability
More than $70 billion is spent each year in the drive-thru lanes of America’s fast-food restaurants. Having quick, accurate, and friendly service at a drive-thru window translates directly into revenue for the restaurant. According to Jack Greenberg, former CEO of McDonald’s, sales increase 1% for every six seconds saved at the drive-thru. So industry executives, stockholders, and analysts closely follow the ratings of fast-food drive-thru lanes that appear annually in QSR, a publication that reports on the quick-service restaurant industry.
The 2012 QSR magazine
drive-thru study involved visits to a random sample of restaurants
in the 20 largest fast-food chains in all 50 states. During each
visit, the researcher ordered a modified main item (for example, a
hamburger with no pickles), a side item, and a drink. If any item
was not received as ordered, or if the restaurant failed to give
the correct change or supply a straw and a napkin, then the order
was considered “inaccurate.” Service time, which is the time from
when the car stopped at the speaker to when the entire order was
received, was measured each visit. Researchers also recorded
whether or not
each restaurant had an order-confirmation board in its
drive-thru.
Here are some results from the 2012 QSR study:
Hypothesis to be tested
VS
given p1=0.881 p2=0.902 n1=1327 n2=726
The test statistic where q=1-p
= -1.46
we reject null if at 5% level of significance
also the p-value calculated is 0.14 which is greater than 0.05(alpha) thus strongly accepting the null hypothesis
1.46 < 1.96, thus we do not reject the null and accept the null and conclude that there is no significant difference between the population proportions
we can see that the confidence interval is quite narrow and indicates stability with the results. The estimate is stable and indicates the range of difference which is very small (below 0.05) indicating little or no difference. The confidence interval gives us the range of possible values of the difference between the two population proportions. 95% of the time the difference in population estimate will lie in the interval -0.049 to 0.00649 thus proving the null hypothesis to be true that there is no significant difference between the two population proportions.