lecture 4

Question 1

research process steps

Answer 1

Development of research question
Design of study
Collection of data Description/analysis of data Interpretation of results

Question 2

Study design refers to

Answer 2

the methods used to select the study participants, control any experimental conditions, and collect the data.

Question 3

study design

Answer 3

Interpretation of results depends on the study design.

The study design should be tailored to the research question.

Methods of statistical analysis and information produced will depend on the study design.

Good research requires good study design.
- Often given too little attention.

Question 4

Broadly identify two main types of study design

Answer 4

descriptive studies

analytic studies

Question 5

descriptive studies

Answer 5

studies which describe things such as surveys

nerve seek to intervene or split groups, only collecting data, no intervention

Question 6

analytic studies

Answer 6

studies which test hypotheses
have a particular claim/value we want to test against
experimenting on humans or entities

two types = experimental studies and observational studies

Question 7

experimental studies

Answer 7

type of analytic study

e.g. randomised controlled trial (e.g. testing a drug)

Question 8

observational studies

Answer 8

type of analytic study

e.g. cohort studies, case-control studies

Question 9

aim of descriptive studies

Answer 9

Aim: To describe, for example,
the characteristics of people with a disease (person; place; time);
lifestyle patterns in a population;
attitudes to health care.

Descriptive studies are often simply referred to as surveys.

Generally use a sample from the population of interest.

do not go into it with any preconceived hypothesis (can develop hypotheses through the results of these studies)

Question 10

population

Answer 10

Complete set of entities or elements or units or people that we wish to describe or make inference about.

should be well defined such as all patient diagnosed with colorectal cancer in NZ in 2015 rather than the population of NZ (because is this right not? past? future?)

Question 11

census

Answer 11

whole population investigated

Question 12

sample

Answer 12

a subset of population

Question 13

what does a sample need to be to provide a good description of a population? (exam)

Answer 13

To provide a good description of the population, samples need to be ‘representative’ of the population they were drawn from; this is why random sampling (also known as random selection) of study units is important.

sample must be representative of the population and randomisation is necessary to achieve a representative sample

Question 14

What if random sampling is impossible

Answer 14

Need to consider carefully how likely it is that our sampling procedure
will produce a representative sample
- Sometimes we can compare the characteristics of the sample to known
facts about the population; e.g. are the age distributions the same?

Question 15

Sampling frame

Answer 15

List of items in a population from which a sample is drawn.

A good one will encompass the population of interest

Rarely coincides with the entire population of interest:

• Telephone numbers (in the past this was good, the present not so good as less and less people have landlines therefore not as good to get a random sample from this)
• Electoral roll

Often doesn’t exist:

• All people with depression
• All potential users of a new drug

Even without a list we can ensure an ‘unbiased’ sample if every individual has the same chance of being drawn.

Question 16

Method

Answer 16

Select a subgroup (sample) of people and measure (this is generally the process)

Question 17

Random sampling

Answer 17

Choose the sample in such a way that every individual in the population has a known chance of being selected.
In a simple random sample, everyone has an equal chance of being chosen.
This method is the best way of obtaining a sample which is representative of the population.

randomisation is good because it aim to get a representative sample

Question 18

simple random sample

Answer 18

In a simple random sample, everyone has an equal chance of being chosen. - probability of choosing people is known and equal
This method is the best way of obtaining a sample which is representative of the population.

Question 19

equation of sample mean =

Answer 19

population (true) mean + error

error can be systematic error or random error
error can be positive or negative

Question 20

error

Answer 20

is how far from the truth the estimate is

Question 21

sample mean

Answer 21

is the estimate of the true corresponding parameter value

you hope that it is close to the true population mean

Question 22

random error also known as

Answer 22

random variability or uncertainty or chance

this is good as it can be dealt with

Question 23

Random error (chance)

Answer 23

Due to natural variability.
Increasing the sample size will reduce the random fluctuations in the sample mean.
Statistical methods allow us to quantify the influence of random error on our estimate.

Question 24

Systematic error is descriptive study =

Answer 24

bias

Question 25

bias as a systematic error in descriptive study

Answer 25

bad because it arises from flaws in the statistical methodology such as poorly collected data or purposefully influencing the data Due to aspects of the design or conduct of the study which systematically distort the results. Occurs if a sample is not representative of the population (Selection bias). Occurs if the information collected from the sample members is incorrect (Information bias). Cannot be reduced by increasing the sample size.

Question 26

Information bias

Answer 26

type of bias where the way the information is collected influences the results such as the measuring tools being calibrated wrong, or different tools at different locations that extract the information differently

Question 27

assuming data is unbiased can lead to

Answer 27

misleading conclusions

Question 28

probability sampling

Answer 28

Probability sampling is important because it helps to justify the statistical models which will be introduced in this course. The key characteristic is that we know the probability of being selected for everyone in the sample frame Simple random sample is the simplest form of probability sampling. For a finite population of size N draw a sample of size n such that each possible sample has the same probability of being selected.

Question 29

simple random sampling and probability sampling

Answer 29

Simple random sample is the simplest form of probability sampling. For a finite population of size N draw a sample of size n such that each possible sample has the same probability of being selected.

Question 30

Stratified sampling

Answer 30

Much of statistical design theory is about controlling variation. More variation in the data means less precise inference (signal vs noise). Stratified sampling is useful when the population comprises several groups of similar individuals. - A stratum is a population sub-division of similar units. Take a simple random sample from within each stratum. stratified sampling is important when looking at different subgroups and testing the same treatment for example probability is proportional to size - everyone has the same change of being selected sample with equal numbers from each strata - those in smaller strata are more likely to be selected More precise estimate than for the same sample size from a simple random sample Can take different sized samples from different strata (a device for reducing overall variability) Useful if you are interested in the results for each stratum and some of the strata are small. Example: colon cancer treatment, samples of colon cancer patients, stratified by ethnicity.

Question 31

Example: colon cancer treatment - simple random sample or stratified sample

Answer 31

some common, some rare groups therefore if you take a simple random sample the rare ones may be missed therefore doing stratified sampling ensures that these groups are represented and analysed

Question 32

cluster sampling

Answer 32

The population may be composed of similar and naturally occurring groups. Cluster samples take a simple random sample of groups: - For a single stage cluster sample include all the units in the selected groups. Cluster sampling is a probability sampling method in which you divide a population into clusters, such as districts or schools, and then randomly select some of these clusters as your sample

Question 33

Method of cluster random sample

Answer 33

population divided into clusters - done when individually sampling is difficult. Simple random sample of clusters. Everyone in each sampled cluster is included in the study

Question 34

Method of two stage cluster random sample

Answer 34

population divided into clusters. Simple random sample of clusters. Simple random sample of one person from each cluster. Probability of someone being in the study depends on the number in their cluster (still selecting people with a known probability). e.g. schools, homes for example schools are the clusters then get the kids of the age you want from there For a two stage cluster sample, take a simple random sample of units within a group. Example: a simple random sample of schools. A single-stage cluster sample would include all students at the selected schools. A two-stage cluster sample would take a simple random sample of schools, then a simple random sample of students at each school.

Question 35

Uses of cluster sampling

Answer 35

Useful when the clusters are easy to sample – a frame might exist for them. Usually cheaper and usually less precise than a simple random sample or a stratified sample of the same size. Can compensate by taking a larger sample. Can combine methods - stratified sampling of clusters etc.

Question 36

The formulae for computing estimates from simple random sampling, stratified sampling, cluster sampling etc are

Answer 36

all different

Question 37

study design influences

Answer 37

the tools of analysis