You have answered your question: "different tests used in subtly different circumstances." Since you haven't provided information about your circumstances, I assume you are not expecting an answer (you will get some but by people who have not recognized what you already have said!). Presumably for you to have noticed this you have already come across Lord's paradox and issues depending on the distributions (of the gain scores, the residuals, etc.).

if you describe the context of your research and 'circumstance' that you are in, we might be able to help you. otherwise, as you said, there is an ocean of analytical techniques to be drowned in :)

There is an important difference between single-item LIkert-scored variables and multi-item scales that are composed of Likert-scored items. The former is ordinal and requires non-parametric tests, while it is possible that the latter can be considered interval-level and used with t-Tests, correlations, etc.

Sorry I didn't give enough details above. There are two questions, one asking subjects to rate their agreement with the statement, "I understand my diagnosis," and the other, "My diagnosis explains the symptoms I experience". People were asked these once before and once after treatment. There are 11 subjects in total. The answers for each seem to cluster around a mean, but I'm not sure I can say what the distribution is with so few participants.

As you can likely tell, I'm a little statistically impaired, so please let me know what other information is needed to make this a meaningful question.

Okay, I have gone through the question and the details provided to guide you choose a statistical tool for analysis. True, the comments already given are helpful, although, the one given by David Morgan is close to given you a solution. However, to get an appropriate tool for your analysis you will be required to provide the following information:

1. state the aim of the study

2. state the research questions

3. Your design is Quasi- experimental, because your work is an intervention study with pre and post test. So you should expect some comparison test too.

4. Are you using a standardized instrument or a self constructed instrument? If you are using a standardized instrument, then scoring and analysis won't be an issue.

5. Provide the instrument with its items.

6. No doubt, sample size affects the distribution and probability of statistical significance. It is better with large samples above 30 subjects.

I wish this would be helpful.

Thanks

Hey David, as per what i have undersood from the above discussions you have asked same questions form same resondents before the treatment and after the treatment and have meausred the changes in the responses on 10 point likert sacle ranging from strongly agree to strongly disagree. If i am on right track u must have a response like

1 = strongly Disagree

2

3

4

5

6

7

8

9

10= Strongly agree.

if so kindly reply

The paired samples Wilcoxon test (also known as Wilcoxon signed-rank test) is a non-parametric alternative to the paired t-test used to compare paired data. You could use this for each of your two variables.

Ankita Mandoliya , yes, that is correct.

David L Morgan , thank you. It looks like that is exactly what I need.

Thanks for your help :)

David L Morgan has guided you with right kind of test to apply specially in cases where paired data is there. Otherwise u can always look for similar studies conducted in the past for reference.

all the best for the study David D G Palmer

I am expert in evaluation of paired ordered categorical data- There are good powerful statistical methods taking into account of the rank-invariant properties of the data - i order to obtain valid results. Please do not be tempted to use parametric statistical methods that assume quantitative normally distributed data. see added file

True, from the nature of your data (ordinal data) as confirmed by "Ankita" you have to go with a non-parametric statistical tool for your analysis. As such, the recommendations from "David L Morgan" and supported by "Svensson (2001)" is a perfect guide for you.

Goodluck David Palmer.