I would be grateful if can guide from the theoretical or statistical point of view
1. What is the best in stating the research objectives, hypotheses or discussing the findings to use the term "Association", "Relationship" or "Effect" in the cross-sectional studies?
2. Should we consistently use only one term?
I would be grateful to here from many perspective..
Warm regards,
It depend on the statistical method used. For association term, it suit for chi-square test to determine the association between variables. For relationship and effect term is applicable for SEM to determine its relationship btween indepndent and dependnt construct.
Best regards
I think the differentiation must rather be between the terms“association” and “effect”, and the two are kinds of relationships.
The term "association" is closely related to the term "correlation while the term ”effect” is more related to the “causality”. The difference between the two terms is difficult and depends on the nature of the variables and the aim and the direction of the analysis.
I tend to be wary of non-specific, and what I call non-descript statements especially when quantitative methods and/or statistics are involved.
For example the words relationship/related which are used widely without being specific.This tends to allow the less experienced researcher to become involved in conjecture and could possibly weaken the development of a firm research strategy. In quantitative research being specific can apply to the wordings used in hypothesis tests, for example, these can be incorporated into objectives such as the examples that follow:
Testing Association or Independence - nominal data - chi-square
Testing mean differences - scale data - t-tests and/or ANOVA
Testing correlation - scale data - Pearson correlation
Testing Rank correlation - ordinal data - Spearman correlation
Testing for regression - usually scale data - simple/multiple regression
Some also choose slightly more ambiguous terms in objectives such as testing perceptions, attitudes, beliefs, intentions, service quality etc. (in other words testing suggested instruments). This is a popular alternative - as long as this is later developed more specifically, later, in a research strategy to develop hypotheses as above, I think this is acceptable.
Thanks for every thing, I am very grateful your responding..
Hayes and Preacher (2014) stated that, no statistical model can prove causality in the mediation analysis.
Imai, Keele and Tingley (2010) said, the causal inference can be if the researcher can argue that variables are modeled in causal sequences with no omitted variables.
So, What is the appropriate to use consistently in the final report (objectives, hypotheses and discussions)?
Do you agree if the researcher stated the research questions, objectives and hypotheses in a causal sequences but in the discussions of the findings and the conclusions to use the term association is more appropriate for the cross-sectional studies?
Please further input would be very helpful..
Warm regards,
Note what Wan says above - I tend to agree - the word association is a word commonly used in the hypothesis of a chi-square test which is a comparatively weak statistical test undertaken on nominal data. Personally I wouldn't use when reporting quantitative findings (unless with respect to a chi-square test).
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics