Feedback Scoring

Intro

A few days back I was pulled into a Data Center of Excellence (DCoE) meeting. The meeting got me thinking of a feedback scoring scenario and how I would go about handling it.

Say you have different stakeholders in a company (like a hospital). They ask patients to fill out questionnaire/survey on their customer experience. The question I asked myself was, collectively, can all the surveys tell us how we are doing as a hospital on a particular issue. The assumption is that the surveys have the same questions but the answer choices vary.

One example question can be as below:

How did you find your stay in the hospital?

• Very satisfying, Satisfying, Neutral, Dissatisfied, Very dissatisfying.
• Very satisfying, Satisfying, Average, Dissatisfying
• Satisfying, Average, Dissatisfying…. and so on

Step by step process

For all the surveys, breakdown the question and answer choices into a table (first two columns). If the tables exist in say an excel file, you could  continue to the next step (on excel) or load the files into R for the manipulation.

The last 2 columns are calculations that you would normally do in excel, R or any other platform/software you are comfortable with.

On a sequence of 1 : (number of answer choices) with increments of 0.001 - the increment size is dependent on the number of decimal places you are working with. Find the percentile the average score is on. This will be the overall percentage score on a scale of 0:100%.

# Using the example above (Code is in R)

score<- 3.690

(data.frame = (point = seq(1,5,0.001)) %>% # Create a sequence
mutate(Percentile = round(rank(point)/nrow(.) * 100, digits = 3)) %>%
#Create a column with the percentiles of all the values on the sequence
filter(point == score))$Percentile  
# Filter the row that matches our score and return the percentile

#The score from this is 67.26%

The other way would be a simple 3.690/5  which gives 73.8%

The two numbers are different because the second answer is based on a scale of 20:100%. That is, assuming everyone chooses very dissatisfied, the average score will be 29/29 = 1 and 1/5 gives 20%. The score will never go below 20%

Repeat this for the other surveys. Find the average for all. This will be the wholesome score regarding patient’s stay in the hospital.

Happy exploring!!

~NMN