You are on page 1of 4

GENERAL COMMENTS

1. In accordance to the Data Privacy Act implemented in the Philippines, each group should include
a Data Privacy Consent Section in their survey (preferably at the front page). This consent form
should state the purpose of the survey and should explicitly declare that the information taken
from the respondent will be used sole for the purpose of the research study that will be
conducted. There are many examples online, but it generally looks something like the given
example below.

2. All survey instruments must undergo a “reliability” test. This test statistic is commonly known as
the KR20 Statistic (a.k.a Cronbach’s Alpha). Here is the link to a tutorial in YouTube and further
discussion on interpretation of Cronbach’s Alpha:
https://www.youtube.com/watch?v=ANxZRi_achQ
Generally, we want to target a Cronbach’s Alpha of 0.7 – 0.9. Values between 0.7 – 0.9 indicate
that survey instrument you have created is internally consistent. To test for the Cronbach’s Alpha
of your instrument you have to do the following:
a) Prepare a survey instrument that will gather relevant data needed to prove or disprove a
hypothesis.
b) With at least 20 respondents first, let them answer as honestly (and unbiasedly) as possible
the survey instrument you have prepared.
c) From the initial 20 respondents, calculate the Cronbach’s Alpha according to the tutorial in
the link given above.
d) If your Cronbach’s Alpha is between 0.7 – 0.9, your survey instrument is good to go. But do
take note however of the comments of your initial respondents. There may be some questions
that are vague and there may be also questions that need revision.
e) If your Cronbach’s Alpha is below 0.7 or higher than 0.9. Consider revising your questions by
either adding questions or omitting vague ones.

3. The demographic profile section of your instruments still needs to be polished. Information that
you will be getting from the demographics of your respondents are crucial for statistical analysis.
It is essential that you can collect the correct demographic information. Some common
demographics that are usually asked are the following:

- Name (usually optional unless it is needed for your analysis)


- Sex (M or F)
- Age
- Permanent Address
- Present Address (for those boarding etc.)
- Boarder/Non-Boarder
- K-12 School Graduated From
- K-12 Track
- K-12 Strand
- High School Graduated From (for non-K12 Graduates)
- Family Income (usually this is confidential unless this information is crucial for
analysis; be sure to draft a good consent form if you are going to ask for this
information; a range rather than the exact family income is usually preferred)
- GWA (same comment as family income; a range rather than the exact GWA is usually
preferred)
- Grades in Specific Subjects (same comment as family income; a range rather than the
exact Grade is usually preferred)
- Course (or Degree Taken)
- Shifter/Non-Shifter
- Regular/Irregular
- Year Level
- Subjects taken in K-12 Curriculum
- etc.

You may refer to this link: https://blog.hubspot.com/service/survey-demographic-questions for a


good way of asking demographic questions. An example is given below.
4. Your survey instruments must also include an instruction section. The instructions should be clear
so that the respondents will conveniently answer your survey.

5. Generally, no one would want to answer a survey so therefore make your survey as aesthetically
good as possible. Make your survey pleasing to the respondents so that they will be encouraged
to answer to the best of their ability. No matter how good a survey is if the respondents “chamba”
their answer, your results will not be good.
6. It is suggested that the respondents shade their answers to your survey questions. You may
provide an answer sheet for the respondents’ convenience. Shading of answers allow more
“focus” and will minimize respondents doing “chamba” on their answers.

7. The number of respondents is also vital for a reliable analysis. Though usually sampling from a set
population is preferred, the reliability and validity of the sampling method (for example random
sampling) used is always scrutinized. Unless you are very sure of the validity of the sampling
method you will be using, sampling the whole population (for example: all 2nd year students rather
than just 20 students per section) is unbiased and promotes reliable statistical analyses.

8. The objective of survey is to capture the subjectivity of the respondents’ perception and/or
opinion about a subject matter and in turn quantify this subjectivity. Hence, it is usually preferred
to ask questions following the format of Likert-Scale question. Just make sure that the quality of
questions and the questions have made are in-line with the hypothesis you are proving and/or
testing. Below are examples of Likert-Scales used for survey questions.

9. Which Likert-Scale to use? A Likert-Scale of 1 – 5 is usually preferred. Though a Likert-Scale of 1 –


7 produces more accurate statistical results, these types of instruments are usually prone to
“chamba” of respondents. It is already proven in literature that the best Likert-Scale to use are 1
– 5; hence there is no need to create larger Likert-Scales. The quality of the response is what
matters the most when statistically analyzing the results.

You might also like