Patient Experience survey of GP and local nhs services 2011/12 Volume 1: Technical Report

Annex E: Significance Testing

a. Comparison of results at National, NHS Board and CHP over the two survey periods (2009/10 and 2011/12)

For each question (where comparison is possible), the test statistic can be calculated as follows:

For a result in 2011/12 to be not statistically significantly different from 2009/10, then

To illustrate this, let the:

• percentage positive score in 2009/10 and 2011/12 for Scotland, an NHS Board or CHP be denoted by and respectively.
• number of valid responses in 2009/10 and 2011/12 be denoted by and respectively.
• design effect in years 2009/10 and 2011/12 be denoted as and respectively.

Then the standard error can be calculated as:

b. Comparison of results for NHS Board and CHP with Scotland

For each question (where comparison is applicable), the test statistic can be calculated as follows:

For the NHS Board or CHP to be not statistically significantly different from Scotland results, then

To illustrate this, let the:

• percentage positive score for NHS Board or CHP be denoted by .
• percentage positive score for Scotland be denoted by .
• number of valid responses for be denoted by NHS Board or CHP (geography) by .
• number of valid responses for Scotland be denoted by .
• design effect for NHS Board or CHP and Scotland be denoted by and respectively.

Then the standard error can be calculated as