Modules for the TNM 800

  1. Motivation
  2. The Protocol : Elements: Requirements
  3. The Questions/hypotheses
  4. The Title
  5. Background
  6. Study design
  7. Methodology
  8. Measurement
  9. Sampling
  10. Descriptive statistics
  11. Inferential Statistics & Statistical test
  12. Sample Size
  13. Research Ethics
  14. Research Budget
  15. Timelines/Planning
  16. Resources/infrastructure

Day 1 : Motivation & Protocol

  1. What are the advantages of doing research? What could you hope to personally gain from the experience?
  1. What are the things to consider when picking a research topic? (Or in other words how to find a research topic that matters and that is likely to lead to success).
  1. What is your research question/topic and why did you choose it?
  1. List the essential elements needed in a protocol.
  1. Discuss the advantages of having a “good protocol”. What distinguishes a good protocol from a bad one?
  1. Discuss the elements/requirements of a good research question?
  1. Can you phrase your research question in the form of a hypotheses?
  1. Discuss the requirements of a good title.
  1. Discuss the research question and titles of your group. Decide on the one that is the “best” to present.

Day 2 : Background

  1. What are the key elements of a good literature review? What is the purpose of the literature review?
  1. What is the information needed in the motivation/relevance section of your protocol? Why is this part of your protocol critically important?
  1. Discuss in your groups who has completed their literature review? Can that person defend his/her literature review as meeting the criteria of a “good “ literature review?
  1. Can anyone give a good paragraph on the motivation for relevance of their study? All group-members should try.
  1. Give a definition of:

a)experimental vs non experimental research

b)descriptive vs analytical research

c)a cross sectional study

d)a case control study

e)a cohort study

f)a randomized controlled trial (RCT)

g)a qualitative study

  1. The best study design depends on the research question being asked. “Explain”
  1. What are the similarities between a RCT and a cohort study?
  1. Most students tend to design purely descriptive studies. They should try to re-phrase the question and change the study design to analytic: Do you agree/disagree? Discuss
  1. What are the key elements usually included under “methodology”?
  1. In your group discuss the various research questions and decide which study design would be best to use, discuss the methodology – what does the group think, is it detailed enough?
  1. For those doing intervention studies more details is needed : See website

For those doing diagnostic studies more details is needed: See website

For those doing epidemiology studies more detail is needed: See website

For those doing qualitative research more detail is needed : See website

Day 3: Measurement

  1. What does: “Bias” mean and why is it a threat to any study?
  1. What are the 3 different forms of Bias and give an example of each.
  1. Name steps that can be taken to minimize bias in the various study designs.
  1. What is meant by measurement error? How can it be minimized?
  1. Describe the purpose of a pilot study and discuss how you would implement it in practice.
  1. Discuss what steps you have taken in your study to minimize bias and measurement error?

Sampling

  1. Why do we study a sample and not the whole population? What is the price we pay for this ( the downside of only having a sample) ?
  1. What are the criteria for a good sample?
  1. Name and describe the various types of sampling.
  1. What is meant by a sampling frame ?
  1. How do you plan to get a sample for your study? What do the others think, is it good enough or not?
  1. What is the difference between a parameter and a statistic?
  1. What do you understand the following to mean:

a)sampling variation

b)sampling distribution of the mean

  1. What would we get if we had the mean of sample means?

15.We don’t have the mean of the sample means we only have our sample means. Depending on our sample size our mean may of may not be closed to the population mean:

a)how does one get a standard error of the mean (SEM)

b)how is it used to help us decided whether we are close to the population

c)if this is what we use the (SEM) for, for what do we use the Standard Deviation (SD)

16.Name and define 3 measures of central tendency.

  1. Name and define 3 measures of spread.
  1. Give a definition of a P – value.
  1. Give a definition of a 95% confidence interval.
  1. Why is a 95% confidence interval preferred above just a P value?
  1. What is meant by hypotheses testing? List the steps
  1. What is meant by type I and II error?
  1. How are they interrelated?
  1. What are the determinants of sample size calculation for comparing two groups and

a)a continuous measurement and

b)a categorical measurement

c)Show a calculation of each

  1. Name the different types of measurements.
  1. Why is it important to know what “type” of data/measurements you are collecting?
  1. When deciding on which statistical test to use when comparing two groups with continuous data what are the factors to consider? Give examples of which tests are used when.
  1. Which test is used to compare groups of categorical data (e.g. % HT in group I vs group 2)
  1. In your study: a) discuss the measurements you will be making: what data types do they represent. b) which variables will go into Table 1: descriptives of study population (sample). Which descriptive statistics will you use. Show a “dummy” Table1. b) Which hypotheses will you be testing? c) Which statistical test will you use. Give “dummy” results. d) What sample size have you calculated for your main outcome?

Hypothesis testing an Example

Steps 1.define H0 and HA

  1. collect relevant data from a sample of subjects
  2. calculate the values of the test statistic specific to Ho
  3. compare the values of the test statistic to values from a known probability distribution
  4. interpret the P- value and results.

Our Example.

Step1

H0 systolic BP in diabetics = systolic BP in non Diabetics

HA systolic BP diabetic ≠ systolic in non Diabetics.

Step2

We collect data from a sample of the population including diabetics and non diabetics (how would we have done this sampling?)

We find SBP in Diabetics = 132mm Hg (SD=12)

And SBP in non Diabetics = 128mm Hg (SD=13)

So we find a difference of 4 (not 0 as the Ho states) Remember the concept of sampling distribution that underpins all statistical tests. Another sample could have given a difference of 2 and another -1 and another 5. If we could have had all these mean differences and obtained their mean we would have had the mean difference in the population.

However we only have our sample mean difference. So the question arises how likely is it to find a difference of 4 in a sample when the real difference of 4 in a population is 0. If it is very possible (likely) we stick to 0 being the true population mean (fail to reject Ho). If a sample value of 4 is very unlikely (rare) given a population value of 0 then we reject 0 as being the true population difference (we reject Ho).

Step3

Calculate the test statistic specific to H0 (e.g. t statistic)

Diabetics π=50 Mean SBP = 132 SD = 12

Diabetics π= 50 Mean SBP = 128 SD = 13

Non Diabetic π=50 mean SBP = 128 SD = 13

Pooled SD = √(49 X 12² + 49 X 13²

(50 + 50 -2 ) = 12.5

Test statistics t = 132 – 128 = 4=1.6

Step 4

Compare with Appendix at back of any statistics book (t-table)

T statistic =1.6 degrees of freedom =98 look up in t table

P lies between 0.1 and 0.2

. ttesti 50 132 12 50 128 13

Two-sample t test with equal variances

------

| Obs Mean Std. Err. Std. Dev. [95% Conf. Interval]

------+------

x | 50 132 1.697056 12 128.5896 135.4104

y | 50 128 1.838478 13 124.3054 131.6946

------+------

combined | 100 130 1.260792 12.60792 127.4983 132.5017

------+------

diff | 4 2.501999 -.965136 8.965136

------

diff = mean(x) - mean(y) t = 1.5987

Ho: diff = 0 degrees of freedom = 98

Ha: diff < 0 Ha: diff != 0 Ha: diff > 0

Pr(T < t) = 0.9434 Pr(|T| > |t|) = 0.1131 Pr(T > t) = 0.0566

Step 5

P>0.05: Chance of finding a sample of mean difference = 4 when the population mean difference is 0, is more than 5%, i.e. quite possible (by convention) and we fail to reject Ho. So we cant say there is a difference in SBP between diabetics and non-diabetics.

Day 4

  1. Name and discuss the key ethical principles governing clinical research.
  1. Name the key resources you need to consult to ensure that you comply with the ethical rules to do research on human subjects.
  1. Mention the steps/requirements you need to fulfil before submitting your protocol to the ethics committee.
  1. Discuss the common problems that the ethics committee find that prevents student protocols from being accepted.
  1. Design an informed consent form for your study – make sure it meets all the requirements. Discuss it with the group for their review and approval.
  1. What is the difference between traditional authorship and contributorship?
  1. When does someone qualify as co-author?
  1. Draw up a list of contributor (authors) specify each one’s contribution. Leave a space to sign and date.
  1. Which actions/associations can be seen as “conflict of interest” when publishing and should be declared to the authors?
  1. Make sure you read the University’s policy on plagiarism.
  1. What are all the elements that should be included in your budget?
  1. Draw up a budget sheet for your study and discuss it with the group for discussion. What is important about your budget when your protocol is reviewed either by the ethics committee or a grant review board.
  1. What is a Gant Chart? Draw a Gant Chart for your study.
  1. What resources and infrastructure will you need? What do you already have and what will you need to get? What are your contingency plans (what if the ECG machine breaks?)