I expect that you have asked many questions in your questionnaire, say 20 to have an idea about the new curriculum, and you have said that you have used the Likert scale, say a 5 point scale. In this case you can multiply 20Q X 5 = 100 scores (best opinion about the new curriculum). Then you are going to start the analysis, I suggest the following steps:

1. Calculate the frequencies of participants for each question (you can combine the 1,2 of Likert scale together and 4,5 together and leave the 3 as a separate entity. In this case you can know the strengths and weaknesses of each category (question) of the new curriculum. Proportion with and proportion against each category. 

2.You can compare the scores e.g. between males and females (better to use a non-parametric test like Mann Whitney test rather than the t test for two independent samples) or the Kruscal Walis test if you have three groups or more to compare.

3.You can also divide the score (100%), into those below or above the median (let us call it scores Cat). Then you can use the Chi square test (e.g. 2X2 table between gender and score Cat).

4.You can use the score Cat as a DV and use the logistic regression analysis.

5. You can use the multiple regression where the DV is the scores (of 100).

6. The best model (for me) is to compare the opinion scores of fathers about the old curriculum and the new curriculum assuming that they have children who have studied the old curriculum. Then you can use the the Man Whitney test to compare the old and new scores.

Hi! How many levels does your IV have? Do all your participants read about the same curriculum? If not, how many versions do they get?

If there are two versions and attitudes are measured as the DV on a likert, you'd use a t-test (probably independent sample assuming people only read about 1 of the curriculums).

If it is three or more versions, there would be a one-way ANOVA (again, between subjects assuming people only read about one curriculum and not all the versions).

You could also run a factorial ANOVA with one factor being Curriculum type and Type of Participant (e.g., Student/Parent) as another factor.

Let me know how many versions of the curriculum they read about. That will help a lot.

Thank you Michael, for taking the time to respond.

As far as I understand, my IV has only one level, or one version. The IV is a new ESL curriculum that the district has implemented for all non-native English speakers in all levels school - grades 1 - 12. All of those classified as ESL have been exposed to the same curriculum, but in different ways (the teachers have taught it, the students have received it, and the parents have been hearing about it from their children.

Now the district wants to know what perceptions are out there regarding this curriculum and its effectiveness. Other classmates are analyzing the hard data...test scores, etc...my task is to analyze the attitudes, beliefs and opinions of those involved in the curriculum. So I need to find a quantifiable way to collect and analyze the data.

Thanks again for your help.

Scot

looks as if a mixed methodology is what is needed - measure attitudes by a survey and determine which inferences are statistically significant by degree of agreement or disagreement.   

Scot, okay. So if everyone received the same curriculum, then your IV is the "who" factor which is three levels (teachers, parents, students) and your DV is the attitude. You'd probably want to run a One Way Between Subjects ANOVA with three levels and your DV would be your measures of attitudes. What system will you be using to analyze your data?

Scot,

How many ranks does your Likert scale have? Is every attitude measured by one question or are responses on individual questions combined to calculate an attitude (multiple questions asked to operationalize an attitude)? Your answer will determine the level of measurement of your DV and consequently the statistical tests that are appropriate to use. If your DV is of interval level (measured by a number of questions/statements combined), in addition to frequency (descriptive) analysis (how students, parents, teachers see the new curriculum), you can do One-Way ANOVA to explore whether differences exist in how different groups see the curriculum. If each attitude is measured by one question/statement and is of ordinal level of measurement, you will be able to run the descriptive analysis (frequencies). If you're interested to compare the groups, you will have to use Kruskal Wallis non-parametric test.

If you collect info about participants gender or length of practice (for teachers), you can then complement your analysis by looking at the differences within each group (gender of students, gender of parents,  gender and length of practice for teachers). And if you want to extend your study beyond understanding the general trends to exploring the nuances of attitudes, qualitative open-ended questions will be helpful (but then you'll have to perform a qualitative analysis of the data too). It depends what resources in terms of time and money you have.

Hope it helps

Thanks to all for your very helpful insights!!  I REALLY appreciate your time in helping ask some meaningful questions that have helped me think through and be specific about what I am trying to research.    I am amazed that you took time to engage the questions of a total stranger -- very much appreciated and I hope to be of help to others somewhere along the way!  Thanks again!!

More likely you will be doing a cross sectional SAMPLE survey that intends to provide a description of your sample, identify relationships, or determine differences. If you do PROBABILITY sampling, here are your possible tests= Z (or t) test of one mean or proportion, Z (or t) test of difference between two independent samples, t -test of, relationships, and chi-square test.

Basically, your objective is to describe the opinions, satisfactions, suggestions and impact of new curriculum. so in this situation, you need a questionnaires through which you can measure all these parameters. so first of all develop or adopt questionnaires from literature and then collect the data. after collection of data, first of all give descriptive statistics e.g mean and percentages and then if you want to see the gender effect and want to compare any groups then you can apply t test or chi square test.

hope it will help you. i you need further clarification then let me know....

Scot, to my understanding,  analysis of perceptions/opinions of different groups towards single variable (in your case the new curriculum) can be best done through One way ANOVA.

I expect that you have asked many questions in your questionnaire, say 20 to have an idea about the new curriculum, and you have said that you have used the Likert scale, say a 5 point scale. In this case you can multiply 20Q X 5 = 100 scores (best opinion about the new curriculum). Then you are going to start the analysis, I suggest the following steps:

1. Calculate the frequencies of participants for each question (you can combine the 1,2 of Likert scale together and 4,5 together and leave the 3 as a separate entity. In this case you can know the strengths and weaknesses of each category (question) of the new curriculum. Proportion with and proportion against each category. 

2.You can compare the scores e.g. between males and females (better to use a non-parametric test like Mann Whitney test rather than the t test for two independent samples) or the Kruscal Walis test if you have three groups or more to compare.

3.You can also divide the score (100%), into those below or above the median (let us call it scores Cat). Then you can use the Chi square test (e.g. 2X2 table between gender and score Cat).

4.You can use the score Cat as a DV and use the logistic regression analysis.

5. You can use the multiple regression where the DV is the scores (of 100).

6. The best model (for me) is to compare the opinion scores of fathers about the old curriculum and the new curriculum assuming that they have children who have studied the old curriculum. Then you can use the the Man Whitney test to compare the old and new scores.

I agree with Namir. This study is of qualittative nature, therefore Non-paramteric Statistics is advisable. U Mann-Whitney test to compare two groups and Kruskal-Wallis test to compare three or more groups of diferent categories (gender, ethnic groups, neigbourhood). SPSS is very practical and useful and it is very easy to use it. Videos can be watched on YouTube. I use thes test very often in my designs.

Scot,

I am an educational researcher.  I think that the One Way ANOVA is a good idea...if you are looking for differences in those groups.  In your original question you stated that you were looking at their opinions, not differences in opinions.  Is there any type of summary statement/question?  You know...a question that ties it all together? 

It appears that you ae dealikng with ordinal data for wich the appropriate satistic is interquartile deviation.

Friends,

Your collective responses have been a gold mine!! I sincerely thank you all for your insights!

Were I am getting a little snagged on my project (or in my thinking) is that it seems that most quantitative analyses compare two or more groups...one group receives the treatment, the other group does not, and you run the test to see if there is any statistically significant difference between the two.

I am planning on conducting a cross-sectional survey to examine the attitudes, beliefs, and opinions of the students, parents, and teachers, regarding their perceived effectiveness of the new curriculum.

I am not too interested (maybe I should be??) in comparing the groups – what do the men think about the curriculum, what do the women think about the curriculum, the teachers, students, etc. I was thinking of just using the survey to gather their collective responses and present this as my data. For example, “75 percent of the respondents think the curriculum has met the intended objectives.” or “43% of those surveyed believe that additional time in discussion groups would be beneficial to learning.”  Is this a valid kind of study?

But now that I think about it, maybe it indeed would be a stronger study if I were to compare groups? Or maybe I actually have to compare groups in order for it to qualify as a valid quantitative study?

For our class we are using SPSS and slowly learning the ropes of this program.

Thanks again to each one for your time and insights. Very, very helpful great food for thought!

Scot

Scott, you have already received a wealth of useful suggestions and there is not much I can add. But turning attention a bit away from data analysis to more design issues, I think you may wish to think carefully about whether you are using a 5-point or a 4-point rating scale. Some people think that in a 5-point scale, respondents may use the mid-point as a way of avoiding taking a firm position. On the other hand, some respondents feel cheated if they are not allowed the option. So think about that.

Also, in designing the questions, you would want to ensure that all aspects of your research question(s) are adequately represented across all 20 questions (or whatever the number of questions you are using). The Alpha Cronbach or other appropriate test for reliability would be appropriate.

I think that you can get more details for discussion wnen you compare global results  by sections or groups. But now you have the scope and you can decide.

Thank you very much.  Indeed, your insights have been very helpful.  I really appreciate the strong responses and suggestions.  SM

I agree with previous comments that subgroup comparisons are not only appropriate, but much more informative -- I can certainly see how parents, students, and teachers could possibly have very different opinions about the new curriculum. 

There's a couple of things that I haven't seen in your discussion about your project at all though -- first of all, it sounds like you don't have a control group (a group that didn't experience the new curriculum), correct?  Can you do a pre-post comparison (e.g., probably not for students and parents, but at least for teachers, are there some who had to teach the old curriculum?  What are their opinions of the old curriculum?.  Because right now, all you have are descriptive, summary statitsics -- curriculum type (old vs. new) is not really an independent variable because you have no data about the old curriculum.

The other thing to think about are the specific constructs you'll be asking people about.  You mention attitudes, opinions, and beliefs.  About what, exactly?  the quality? its effectiveness at meeting stated objectives?  Its relevance? the need for a new curriculum in the first place?  In any case, you will want multiple items to tap each construct, and you'll want to do a little bit of psychometric analysis to make sure the items that you think all are tapping the same construct really are (look up Chronbach's alpha in the SPSS help file).  Assuming the psychometrics check out, then you average all the items that belong to a single construct together, and that's what you run your subgroup ANOVAs on (They're Likert scales, so your DV is continuous and your subgroups are categorial, which is what ANOVA is designed for). You'd normally do a lot more for a full blown validation study, but since this is for a class, this is at least a minimum. 

In my view, prior to commencing SPSS, your raw data needs to be subjected to rigorous item-testing to test for validity and reliability and to ensure that your questionnaire instrument can accurately measure the conceived variables and constructs. I'm, thinking Rasch (computed by Winsteps) as it has the computational capacity to execute an item-level analysis and identify misfit items, recommended for modification.

Also worth mentioning at this point is factor analysis procedures that many studies use to test their items, based on the assumption that these test items are normally distributed continuous variables. As a matter of fact, the frequent use of exploratory factor analysis fails to account for the measurement, distributional and structural properties of the un-weighted data, based on Likert-type ordinal scales. This can run the risk of generating biased and misleading estimates that can yield discrepant findings. You may read more about this in: Rowe, K. J. (2002). The measurement of latent and composite variables from multiple items or indicators : Applications in performance indicator systems.  

Rasch is like the prelude to SPSS. If you choose to do it in this sequence.... Rasch to SPSS.....It paves the way for SPSS to report on descriptive and inferential statistics based on your converted "ratio-type" data. In my opinion, ultimately, Rasch and SPSS will both really compliment your findings :)

SPSS software package could be used to analyze survey data. However, the questionnaire designed for gathering data is needed to fine tune with an expert in the field of research. 

In the same vein of K.K.K. I could suggest you to read "Scale Development" wrote by Robert DeVellis an edited by SAGE as an excelent and very friendly tool to understand how to build a measurement questionnarie. Also Springer has a very good one Valliant, Dever & Kreuter, 2013. Practical Tools for Designing and Weighting Survey Samples

You can carry out descriptive statistics using SPSS software package to analyze survey data. YOu can use ANOVA test to test the significance or can carry out regression analysis to assess impact on IV on DV. However, the questionnaire designed for gathering data is needed to fine tune with an expert with the objective of the study. 

What I think you are doing is you have one IV (old or new curriculum) and many DVs (perhaps 20 questions). Some of the previous responses assume that you are going to sum the individual DVs (scored 1 through 5) into one grand scale. I do not recommend this, unless someone else has already used that scale and determined properties of validity and reliability (especially internal reliability). So, I am recommending twenty separate Mann-Whitney calculations. If you have severe ceiling or floor effect on items, consider two group Kolmogorov-Smirnov.

read about exploratory data analysis and cross tabulation. you will get all the information. if you wish i will attach a file to read

The statistical test you can use in a survey is heavily dependent on your research objectives and hypotheses. You are free to use both quantitative and qualitative statistics depending on the nature of your data.

Use of SPSS based on weighted average and mean, standard deviation and T test.

Thanks T.L. Brink .

Nice Contribution Sylvia Hysong.