Sampling method: Convenience sampling
1) The level of social competence (5-point likert scale)
2) Is there significant relationship between grandparenting and academic achievement?
3) Is there significant relationship between grandparenting and social competence?
4) Is there significant gender difference in social competence?
5) Is there significant age difference in social competence?
Academic achievement (grades A, B, C,D,E)
Relationship with grandparents (5-point likert-scale)
Social competence (5-point likert scale)
I think most of these hypotheses can be done with contingency tests such as chi-square and also followed by Pairwise Z-Tests.
You can also use Spearman correlation coefficient for Hypothesis 5.
Dear Lim Hoon
If the mean more accurately represents the center of the distribution of your data, and your sample size is large enough, use a parametric test. If the median more accurately represents the center of the distribution of your data, use a non parametric test even if you have a large sample size.
Best Regards
Dr. Fatemeh Khozaei
Dear Lim,
Some consider "Likert scale" outcome as not normally distributed and tend to analyze their data with non-parametric tests. Also, some try to conduct Kolmogorov–Smirnov test to assess data normality.
In your case, you could use correlation (Pearson or Spearman) for relationship investigation. Moreover, for assessing the difference, you could use ANOVA or Kruskal-Wallis.
Hello Lim,
If, by "5-point" Likert-type scale, you mean a single response, then four of your five variables are ordinal strength (unless academic achievement is an averaged value of many individual marks, in which case you may treat it as a pseudo-continuous variable). Age, if quantified to nearest year, may be considered also a (pseudo) continuous variable.
H1: Report median value; the Tukey 5-number summary would also be suitable (minimum value, lower hinge, median, upper hinge, maximum observed value).
H2: Spearman correlation.
H3: Spearman correlation.
H4: Mann-Whitney U test
H5: Not clear what you're after. If correlation/association, use Spearman. If you want to compare (up to 5 levels of) social competence levels for equality of mean age, you could use one-way anova (or the equivalent regression model).
Good luck with your work.
Apologies to alll but but where do you get p-values from a convenience sample. FGS STATISTICS DEPENDS ON PROBABILITY MODELS.. D. BOOTH, CURMUDGEON PROFESSOR OF STATISTICS
It is the nature of your data that tells you whether to go for a parametric test or a nonparametric one. You could use a parametric test if your data meets its assumptions; one of which is the normal distribution. If not, you could opt for a nonparametric test despite having a large sample. For a 5-point Likert scale, it depends on how you treat your scale(s): interval or ordinal; you could go through the publication by Chyung et al. (2017) and the article by Wu and Leung (2017) for relevant inputs. As regards the myth of not using parametric tests with a nonprobability sample, you might check out the article by Hunter and May (1993) for the demystification of five statistical testing myths, including the determination of statistical significance. You might also find the Libguide by Curtin University Library (2021) helpful. Here are the full citations.
Chyung, S. Y. Y., Roberts, K., Swanson, I., & Hankinson, A. (2017). Evidence-based survey design: The use of a midpoint on the Likert scale. Performance Improvement, 56(10), 15–23. https://doi.org/10.1002/pfi.21727
Curtin University Library. (2021). LibGuides: Introduction to statistics: Inferential statistics. Study and Research Guides—LibGuides at Curtin University. https://libguides.library.curtin.edu.au/uniskills/numeracy-skills/statistics/inferential
Hunter, M. A., & May, R. B. (1993). Some myths concerning parametric and nonparametric tests. Canadian Psychology/Psychologie Canadienne, 34(4), 384–389. https://doi.org/10.1037/h0078860
Wu, H., & Leung, S.-O. (2017). Can Likert scales be treated as interval scales? —A simulation study. Journal of Social Service Research, 43(4), 527–532. https://doi.org/10.1080/01488376.2017.1329775
Good luck,
Dear Lim,
You can apply Tukey Post-Hoc Test for equal variance for your H1. You can also apply non-parametric test Chi-Sqaure for other hypothesis.
If both dependent and independent variables are non-metric go for Chi-square. if dependent variable is metric and independent variable is non-metric go for t-test or ANOVA and if both variables are metric go for Correlation.
Good luck
Parametric tests are always more welcomed as compared to non parametric tests,,but some assumptions need to be fulfill before applying parametric tests.
In your case ,there are attributes to be checked therefore non parametric tests are more convenient.
Deleted research item The research item mentioned here has been deleted
I think it can be helpful.
Regards
Technical Report Proposed Guide for Questionnaire Analysis "Part 2" - Numeric...
Also this is the part 2 that is more detailed.
Regards
You should go for parametric test. For association testing you can use Chi-square test for your question number 2 and 3 and for significant correlation testing go for person correlation coefficient test. You should use ANOVA for question 4 and 5 to see if there's any significant mean difference.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Graphs