|
 |
||||||||
Data recording
If data from the evaluation of individual projects is used for program evaluation, it should be recorded in an appropriate form which has been pre-defined by the project evaluators. For the program evaluation, either raw data from the projects (original data sheet, see below) or descriptive statistics (see Page on Data analysis - Descriptive statistics) can be used.
Recording of questionnaire data
The explanations below refer to questionnaires with closed questions (i.e. with predefined response options). In order to analyse such data with a statistics program such as SPSS or Excel, you should first create a data sheet (cf. Example data sheet, pdf, 36kB). This is arranged in columns and rows:
| => | Each question in the questionnaire goes into a column (ideally in the same order as questionnaire); the first column should contain an identification number or something similar for each participant). |
| => | The rows contain the responses from the participants to the different questions. |
|
Participant |
Question 1 |
Question 2 |
Question 3 |
Question 4 |
etc. |
| 1 | |||||
| 2 | |||||
| 3 | |||||
| 4 |
Please make sure that the column headings always refer to the corresponding questionnaire item (i.e. to the corresponding question). We suggest that you assign a short name to each question and record it in a separate document.
Coding of responses
Each response option is coded in order to be entered into the data sheet and to be analysed. This means that a number is assigned to each response option (e.g. 1 for “disagree”, 2 “somewhat disagree”, 3 for “undecided”, 4 for “somewhat agree” and 5 for “agree”; or 1 for “male”, 2 for “female”, 1 for “yes”, 0 for “no”, etc.). The meaning of the codes should also be documented precisely!
IMPORTANT: Several questions about one topic (e.g. 4 questions about the usability aspect of “controllability”) constitute what is known as a scale. In the analysis, each question is not usually considered separately but a mean value is calculated for questions belonging to a scale. Here, it is important to invert the coding for the negatively formulated questions.
mean value
“The long-term average of occurrences; also called the expected value” (From: Statistical Glossary, Mean Value)
|
Example: |
Positively formulated item: “The software offers the option of interrupting thetask at any time and continuing later without losing data.” Coding: 1 for “disagree” – 5 “agree”. Negatively formulated item: “The software enforces an unnecessarily strict way of complying with the task steps.“ Coding: 5 for "disagree" – 1 for "agree". |
Missing Values
For an item which is not checked, the value of “0” should not be entered, as this would represent a negative bias for mean value calculations. Simply leave the cell blank.
[1] [2]
