# Survey analysis help: what tests to run?

#### Noraft

##### New Member
I'm doing research investigating why South Korean students are reluctant to speak English in ESL classes. I've developed a survey with nine statements describing different types of problems students face (developed from a pilot study), with a 3 point Likert scale (often, occasionally, never). So, for example, one item is: I am afraid of making a mistake (often, occasionally, never). Students are also asked to identify the top three issues, in rank order. Age, grade level, and gender is also collected.

This survey is being administered to elementary, middle school, high school, and university students, almost 1000 surveys in all. I'm hoping to eventually survey around 2000 students, as those numbers will be safely generalizable to the entire student population of South Korea.

My research questions are:

What are the major reasons students are reluctant to speak English?
Do these reasons vary by age? Gender? School level? Or a combination of these?

Now, the problem is that my last graduate level stats class was 9 years ago. I still remember the basic concepts and measures of central tendency, but then I start to get fuzzy.

If I remember right, in order to determine if variance by any of the demographic features is statistically significant, I should run dependent T-tests with everyone of a single variable compared to everyone of another. For example, on each test item, all boys compared to the whole sample and all girls compared to the whole sample. I can also do combinations, right? All middle school boys to all middle schoolers, and all middle school girls to all middle schoolers.

What other kinds of tests am I going to want to run in order to get interesting descriptive statistics?

#### bourne

##### New Member
hi, because you have two research questions, it is better you run more tests rather than to just run descriptive stat.
1. for knowing what factors are important for reluctance in speaking English, you can run Factor Analysis.
2. for knowing the responses from various level (primary, secondary, graduate, etc.) or from gender, etc. you can run crosstab, or ANOVA, etc. in SPSS
3. for t-test, what you have written is correct.

#### Noraft

##### New Member
hi, because you have two research questions, it is better you run more tests rather than to just run descriptive stat.
1. for knowing what factors are important for reluctance in speaking English, you can run Factor Analysis.
2. for knowing the responses from various level (primary, secondary, graduate, etc.) or from gender, etc. you can run crosstab, or ANOVA, etc. in SPSS
3. for t-test, what you have written is correct.
Okay, I'm assuming that since the responses are ordinal (Often=2, Occasionally=1, Never=0), I'm going to want to do something nonparametric: a Kruskal-Wallis one way ANOVA, right?