If the road is not yet clear to you,

we will show you the way how to get through.

A data analysis plan is a necessary part of the research design that describes which analyses will be carried out to answer the research question.

Planning the way the data will be analysed has several advantages:

• It makes clear if the research question can be answered;
• It makes clear what type of analysis must be performed and what knowledge is required to perform them with the available software programs;
• It prevents the scientist from collecting wrong data;
• It prevents the scientist from collecting unnecessary data;
• It helps to select the right cases or subjects;
• It ensures that sufficient respondents are sought; Knowing what to do.

Although all these aspects are interrelated, we will describe them briefly separately below.

Answering the research question

The basis for (scientific) research is to answer a question. The answer cannot be given immediately, new data have to be collected and analysed. For this reason it is very important to have an answerable research question.

Type of analysis

It makes quite a difference if the gathered data are texts or numbers. If it consists of text the data should be analysed with qualitative techniques, and if the data or numbers quantitative techniques have to be used. If the data are digits be sure the right technique can be applied. For instance, comparing two groups with a dependent variable on interval level can be analysed with a t-test. However, if the dependent variable is on ordinal level a Mann-Whitney test should be used. The scientist should know these types of analysis and must have the software to perform such an analysis.

Preventing collecting wrong data

Supplementary to both aspects above, the right data should be collected to answer the research question. The data analysis plan is a (final) check to see if the research question can be answered with these data.

Preventing collecting unnecessary data

Scientists like to collect a lot of data and see afterwards if it makes any sense. Collecting more data however means more time has to be spent on collecting, correcting and analysing the data. Most of the time these activities are only raising the costs. Therefore only collect the data needed.

Collecting data from the right cases

If a study is about a difference between groups, data from all groups should be gathered. In a study in which organisation size matters, not only small companies should be included. Though there are less big companies and probably it is more difficult to persuade them to participate in the research, they should also be included too to get an (external) valid result. However, to make a statistical well done analysis, big companies should be included in about the same amount as the small companies. This leads to opposite requirements, but should be taken care of before the data are gathered.

Collecting enough data

For statistical analyses usually less cases are needed for data measured at an interval or ratio scale than for data measured at a nominal or ordinal scale. Make use of our sample calculator to find the appropriate sample response size.

Knowing what to do

When there is a plan what to do with the data, a prediction of the time to spend on analysing it can be given. It prevents the scientist from doing unnecessary analyses. Some analyses are very attractive, but time consuming and not needed for answering the research question. It might give some insight in the data, but are seldomly reported. So having a plan is more straightforward.

The time spent on analyses can be taken into account when an offer has to be made; a more precise price can be offered. Afterwards the time spent and the costs can be calculated so a future offer can be more accurate.

Related topics to data analysis:

Research question

• Research design
• Design / Conceptual model
• Respondents
• Instruments
• Analysing qualitative data
• Analysing quantitative data

Do professional research. Read our manuals:

• How to formulate a good research question
• How to create research designs and conceptual models
• Nine ways to draw a sample
• How to calculate the sample size
• Representativity
• How to choose the correct statistical test

Return to the index of this dictionary