Many students get stuck at the same point. The questionnaire has been distributed, the responses are already collected, but once SPSS is open, the workflow suddenly feels confusing. The output has many tables. The numbers also look similar. In fact, they answer different questions.
If you are working on a thesis and need validity and reliability testing, the process is not as complicated as it first appears. The usual problem is simpler than that: the data is still messy, and the output is read as if every statistic means the same thing. This tutorial stays practical. The goal is to give you a working sequence you can actually follow.
Before opening SPSS, make sure the data and questionnaire items are already tidy
Start with the basics. Each questionnaire item should have a clear variable name such as X1, X2, X3, and so on. The answer scale also needs to stay consistent. If one part uses a 1 to 5 Likert scale, do not mix another part with yes or no answers unless you have prepared the coding properly.
Before running the analysis, check these points:
- there is no unexplained missing response,
- the score direction is consistent, especially if there are negative items,
- the number of respondents and the number of items match the instrument draft you actually used.
This step looks basic. Still, this is where many testing results begin to go wrong.
How to test item validity in SPSS
For many thesis projects using Likert-scale questionnaires, the common route is Analyze → Scale → Reliability Analysis. Move all items into the Items box, choose the Alpha model, then open Statistics and check Scale if item deleted. After that, run the analysis.
The output section that is often used to read item validity is Corrected Item-Total Correlation. This value shows how well each item is connected to the total score of the variable. In practical guides, there are usually two ways to read it. First, compare the item correlation with the product-moment r table based on the number of respondents. Second, use 0.30 as an early practical check.
If one item shows a weak value, do not rush to label the whole instrument as bad. Read the wording again. The item may be ambiguous, too long, or scored in the opposite direction.
How to check reliability and read Cronbach alpha
The menu is the same. The focus now moves to Cronbach's Alpha. This value is used to see whether the items inside one variable are internally consistent as a single instrument.
A practical benchmark often used in many campuses is alpha above 0.60 as an early acceptable level. If it is already above 0.70, the instrument is usually considered safer. Even so, do not stop there. Read the Item-Total Statistics table, especially Cronbach's Alpha if Item Deleted. There are cases where the total alpha looks decent, yet one item is actually weakening the scale.
That is why item validity and instrument reliability are not interchangeable. A weak item can still exist even when total alpha looks fine. A decent alpha also does not automatically rescue an item that does not match the variable being measured.
One mistake appears again and again in thesis guidance sessions. Students see the whole output as one block and conclude that every item is safe. It does not work that way. A cleaner reading order is this:
- check each item through Corrected Item-Total Correlation,
- remove or revise the clearly weak items,
- run the reliability test again for the cleaned version of the instrument.
If the result improves after an item is removed, that does not mean your data is ruined. It usually means you are cleaning the instrument so the research result makes more sense. If you still feel unsure when reading the SPSS output, Bimbingan Informal can help you review the data, interpret the tables, and translate the result into thesis-ready explanation for the methods and results chapters.