Now it’s time to analyze your data – or crunch the numbers. This last part of a needs survey or assessment can be as easy or hard as you want to make it. With a small sample and less at stake, perhaps eye-balling the numbers will provide the information needed. If you are working with a larger sample or the decisions made from your results may be more impactful, then the analysis should be more scientific. Either way if your needs survey was well-constructed, analyzing the data shouldn’t be too difficult. Here are the basic steps you need to follow to for this last process:
- Clean the data
- Group the data
- Run analytics on the data
- Prepare your needs survey report
Cleaning your data
When we clean data, we are looking through it to make sure that responses are consistent. Some people will also code responses during this activity for easier analysis. But that discussion comes a little later – first we need to clean the data.
Cleaning the data simple means looking through the responses and making sure that the answers are in a standardized format. Examples of this include:
- If the question was asking for a true/false response that all responses are either a t or an f – no maybes or sometimes or only when I’ve had my morning coffee.
- If the question was asking for a rating between 1 and 4, there are no 5’s or decimal number in the responses.
If you’re working with a paper and pencil survey, the responses will almost always be less standardized than digitally collected data. The cleaning of written surveys will also probably happen while you are entering data into a spreadsheet or other software for data analysis. Whatever you decide to clean/standardize your data, be consistent in applying the cleaner.
While software is not required to analyze data, it makes the process much easier and less likely to have errors. If you’re worked with an online survey service, the data will probably be fairly clean from the start. You will also have eliminated the data entry step of analysis.
Some people will also code their data at this point for easier analysis. Coding means to assign a number or letter value to similar responses. Let’s say you have a short answer question in your survey about problems people in your company have using Excel. Here are five of the responses you received to this question:
- I don’t understand pivot tables
- PivotTables are easy for me but I need help with macros
- I would like to learn about pivottables
- I don’t know how to make good graphs
- I don’t use Excel in my job for anything
These responses are all over the topic of using Excel PivotTables and so it would be hard to do any analysis of this question without first coding the responses. Your coding system may appear like this:
- P1 – doesn’t use PivotTables
- P2 – uses pivot tables but doesn’t need help
- P3 – would like training on PivotTables
- P4 – other response
Applying this coding to the responses you received would probably give you something like this:
- Response 1: P3
- Response 2: P2
- Response 3: P3
- Response 4: P4
- Response 5: P4
Coding the responses in this manner will make analyzing much easier because responses are now standardized and so software (or people) can group responses better for analysis.
Three notes before we move on to the next step:
- First, when coding data never delete the original responses – add another column or put the clean data in another worksheet or workbook. You may need to refer back to the original data at some time.
- Second, decide how you are going to handle non-responses – blanks. Are you going to ignore them and say the number of responses for each question – 38 out 42 said yes on this question #78 while 44 out of 50 yes on question #79. Or is your question one where no response needs to be counted as a response?
- Third, the better your questions are written the easier this part of running a needs survey becomes. Refer to part 2 of this series for more information OK. So, I have to do a Needs Survey. How do I write the questions?
Group the data
With your data cleaned and coded, now it is time to group data for analysis (often step two- Group the data and step three- Run analytics are run at the same time). Grouping the data means that you will start to think about how you want to see the results displayed. Are you interested in what type of training administrative assistants need? Are you interested in training already taken be staff members with more than three years of service? Are you interested in what the supervisors see as needs as compared to what the employees see as needs?
Having both data and the ability to run it through the software is meaningless if you don’t know the right questions to ask. These questions come from the original design and intent of the needs survey we talked about back in part 1 – Why Should I Have to Run a Needs Survey? The discussions held at that time should give you the initial groupings of data.
When I teach Excel, I talk about the story that leads to the creation of the spreadsheet/workbook. The story is a story problem like you had in middle school math class. If this happened and that happened, how many are left. You were to take the words and reduce them to math. Think of your original discussions for this needs assessment as your story problem. If we have this many dollars and those are our needs, how can we maximize the result – take the words and reduce them to math. So what this boils down to is for you to use those initial stories or needs or questions and decide how to group your data to answer them.
Run analytics on the data
This is where software really makes your life easier. There are many options available from your standard Excel software to IBM’s SPSS. What you use is more dependent on your experience with different software, the size of your data and your budget. For example, SPSS can run very complicated calculations and very large data sets. However, it is rather expensive and can have a fairly steep learning curve if you’ve never used it before. On the other hand, Excel (especially if you load the data Analysis ToolPak) can run many of those same tests but usually works better on a smaller data set. Somewhere between those two is PowerBI. Though not intended for this specific use, PowerBI can run most or all the same functions as Excel can and can create dynamic visuals to report results. These are just three of many software options available to you for running analytics on data sets.
Whatever tool you use for the analysis, keep several things in mind:
- Always double check your math/formulas/logic to make sure you are getting the output you think you are. It usually helps to walk away occasionally during this part of data analysis and come back with fresh eyes to review your thought processes.
- Have someone else look at your math/formulas/logic to make sure it says what you think it says.
- Ask the same question in a different way to see if you get the same answer. Instead of asking how many people are ____, try asking how many people aren’t ____. See if the math still adds up. If not figure out why.
Prepare your needs survey report – The Last Step!!!!
You are about to enter the world of three kinds of lies, “lies, damned lies, and statistics.” An observation attributed to Mark Twain and British Prime Minister Benjamin Disraeli among others. To enter this world, you are going to take your data, look for trends and then tell someone else what you believe those trends are saying using number to back up you opinion. Until you enter this world and present your findings, everything up to now has been an academic exercise – lots of fun but no real value. You must return to your boss, stakeholders, constituents with the results of your work. This report may be as simple as you sending an email with a couple of research questions, their answers and a few recommendations or as involved as a multipage printed report and a presentation to communicate results to an annual stockholders meeting. You will probably be somewhere in the middle. To get the best results from your presentation, here are my final suggestions of this journey:
- Say it with pictures. Most people respond better to images than to text. Whenever possible use simple, clean charts or graphs to present the hard data from your survey. People also like color better than black and white images and charts.
- Make it easy to understand. You have wrestled with the deeper questions for your survey for quite a while but your audience may be seeing this for the first time. Hep them out by boiling it down to broad strokes or trends that are easier for them to follow.
- Facts at back or available on request. Include that hard data or make it available for those how want to see the number that lead to your broad strokes. There will always be someone how really is interested in how you got to your conclusions.
Needs surveys are an invaluable tool in developing good training. The process is long but very manageable if you take it step by step. GOOD LUCK!!!