Resit Regression report FAQs by Mark Harwood

Participants

py5201uel — 2019-06-18 18:50:10 UTC

1= Male, 2 = Female
Mix of undergraduate and MSc students

Ages: one should state the mean age, and some spread information (either the range and/or the standard deviation, SD).

Ethnicity coding:
1 = White

2 = Black

3 = Asion

4 = Other

5 = British

(I didn't design the study- so don't know why British is an ethnicity, but fewest people chose that anyway)

Appendices

py5201uel — 2019-06-18 18:53:42 UTC

Unnecessary. Can include questionnaires, but probably best to keep it simple and omit.

RESULTS: I am having trouble on how to input data into SPSS

py5201uel — 2019-06-20 23:41:02 UTC

In SPSS, you need to go to the File menu \ Import data\ CSV data ... and select the file from the Resit Quant folder (resits_2019_data.csv)

Then, by having the Word file from the resit folder ('Results help for Mindfulness Lab.doc'), and carefully working through each line there, you should get the Results that you need.

Methods: How do I put in examples from the questionnaires carried out?

py5201uel — 2019-06-21 19:22:01 UTC

Week 3 on Moodle at top has the Questionnaire file, which has 3/4 of the question

Questionnaires

py5201uel — 2019-06-25 08:54:36 UTC

Do I pretend that the questionnaire was given out to the participants, as if I had collected the data myself?

You can use the passive tense e.g. "Questionnaires were given out to participants", which would be true and not explicitly reveal that you did not collect the data yourself. Or you can just be transparent and say that the data were collected and given to you. Just keep is short and simple, not convoluted.

Update vs New?

py5201uel — 2019-06-25 08:58:28 UTC

Am I expected to completely re-write the piece from scratch or can I update the existing one using the feedback and guidance given?

Yes, you do need to resit the Quant regression lab with the new data provided in the folder. There is some overlap in the variables/ topic, which is deliberate to make things more familiar. However, you do have to be careful not to copy/paste your own work as this will count as self-plagiarism. The best thing is to use the same framework and literature as before, but rewrite the words.

Correlation vs Regression

py5201uel — 2019-06-25 09:00:30 UTC

Is this about correlation as well? Or is it just, which independent variable predicts mindfulness?

You are correct about the design (i.e. DV = mindfulness; predictors = TA, AA, AC), and that AC positively predicts Mindfulness. However, it is not the only predictor that is a significant independent predictor of Mindfulness.

Correlation always precedes a regression analysis. Its purpose is to test whether all the predictors are included in the regression. If a predictor variable is not significantly correlated to the outcome (DV) variable, then it is left out of the regression.

Therefore you need to include a correlation analysis before the regression analysis in the Results.

In the report, it is important to make it clear that you understand that there is a difference between correlation and regression. Correlation is just an association (a link/ relationship) between a pair of variables, whereas regression determines if one can usefully predict one from the other (and how much one variable can predict another). In the Discussion, you should concentrate on the regression results, but you should also comment on the correlations. For example, one or more variable may be significantly correlated to Mindfulness without independently predicting Mindfulness.

Video Results mismatch

py5201uel — 2019-06-25 09:21:28 UTC

I was looking at the video you made.

I realised the results I got were completely different compared to yours? I don't know where I went wrong? Because I followed all the steps carefully

I tried to stress in the video that the numbers were deliberately different because at the start of the video I deleted 80% of the data! The point was to show people the correct steps, but not to stop people actually doing the steps themselves and just copying down the numbers from the video screen.

Asterisks * in Results

py5201uel — 2019-06-25 09:23:52 UTC

i don't understand the asterisks?

one * means p < 0.05, which means that result (e.g the correlation) is significantly different from 0 (i.e. it is significantly correlated, since r = 0 means no correlation). Two **, means p < 0.01, which means that it is even more significantly different from 0 (i.e. more significantly correlated). The direction of the correlation is just given by the sign; the significance is given by how low the p values is (lower = more significant); and finally, the "strength of the correlation" is given by how far the correlation coefficient is away from 0 (e.g. -0.325 is further away from 0 than -0.188).

The strength and significance of a correlation are two distinct things, although a strong correlation tends to be significant. (The sample size has a big effect on whether something is significant or not.)

Crucially, the correlation is just done to see which predictor variables are "allowed" to be included in the regression. Only significant ones are included.

We most care about the significance of the regression itself-- e.g. which predictors are independently predictive of Mindfulness.

What Tables needed in Results?

py5201uel — 2019-06-25 13:08:40 UTC

The Results needs a minimum of 3 tables.

1) Descriptives stats (means/SDs/min/maxs as a minimum for the 4 key variables),

2) Correlation table (again the 4 variables as in the video and Word doc),

3) at least one Regression table. The Word doc suggests a compact way of condensing the R-squared, ANOVA and Coefficients tables that SPSS spits out for the Regression into one convenient table; you may as well follow that formatting.

How do I interpret Beta values?

py5201uel — 2019-06-25 13:11:04 UTC

The beta values: the furthest one from 0 is the strongest, the closest the weakest. That's about all you need to comment on with respect to the betas.

Ethics

py5201uel — 2019-06-25 13:13:37 UTC

There should always be some comment on ethics, often in the Procedure section.

Informed consent was obtained from each participant before data collection began, and the study was approved by the UEL research ethics board.

Qual/Quant

2019-06-27 08:58:24 UTC

If i passed the exam but messed up in both essays if i only do one essay can i still pass the module