Data collection

 

Types of measurement

 

Interval level measurement

"Real" measurements, such as height and weight, the size of the difference between measurements can be assessed.

 

Ordinal level measurement

Where is possible to say that one is bigger than an other, so data items may be ranked relative to each other, e.g.  1. no infestation  2. slight infestation  3. moderate infestation  4. heavy infestation  5. severe infestation

Ordinal means ability to order but it is not possible to say 4 is twice as infested as 2.

 

Nominal or catogoral level measurement

A Piece of data can only be attributed to a particular category. Categories can not be ranked or ordered, e.g. gender distribution, HIV status, type of IV fluid

 

Data collection may be direct or indirect

 

Indirect

Indirect is data which has been collected by other researchers or institutions and may come from;

1. Official statistical material, Office of Population and Census Surveys, Local councils, Health Authorities.

2. Scientific publications

3. Data files from past research

 

Direct

Measurement at interval level, e.g., numbers of things, pulse rate, units of alcohol consumed

Observation, overt or covert, participant on non-participant, need to account for Hawthorne effect.

Instruments such as Questionnaires

Use of open and closed questions

Likert –type responses

Interviews, structured or unstructured

Group interviews

Vignette

 

Possible errors in qualitative data collection

Trustworthiness - Honesty of researcher and the participant

Confirmability - Consistency and repeatability of decision making processes

Transferability - The degree to which findings may be applied to another individual or group

Credibility- The confidence that the researcher can have in the truth of their findings

 

Possible errors in quantitative data collection

Introduction of human subjectivity in group allocation

More than one assessor

Poorly constructed and ambiguous questionnaires

Related to principles of reliability and validity

Sampling errors

Sampling

 

Sampling involves the selection of a group of people, events or behaviours to enter into a study.

 

The individual components in a study are referred to as the elements, when elements are people they are called subjects.

 

The sample must be mathematically selected to accurately represent the whole population under study.

 

 

Sampling or eligibility criteria

These are developed from the research question.

 

Inclusion and exclusion criteria.

 

If the sampling criteria are good the sample should yield results which may be generalised to the whole target population.

 

 

Representativeness

The accessible population must be representative of the target population.

 

 

Sampling error

This is the difference between the sample statistic and a population parameter.

 

Sampling error reduces the power of a study to identify differences between groups.

 

Sampling error may introduce random or systematic variations.

 

 

Randomization

Every individual in a target population should have the same chance of being selected into a sample.

 

All subjects should be randomly selected and the randomly allocated to an experimental or control group.

 

In order to select from a population randomly every individual in a population must be identified, this is called the sampling frame. Individuals are then selected from the sampling frame.

 

A sampling plan should be used to reduce bias and promote Representativeness.

 

In many research studies a convenience sample is used.

 

Sample size must be adequate, more will be needed to detect subtle effects.

 

 

 

 

 

Identify the type of data being given in this table;

 

 

Name                            Eye            Social           Temperature    Income

                                       Colour       Class                   `C     

 

Kim Hanson                  Red               4                        36.8            £97 000

 

John Campbell             Blue               3                       17               £3  250

 

Jean Longrigg               Green            1                       40`C           £103  002

 

Cath Boyes                   Orange         2                       36.9            $54 000

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Likert –type scales

 

 

How often do you fall asleep in lessons?

 

 

Always       Often          Occasionally           Never

 

 

 

 

 

Attending research lessons causes;

 

 

No stress                                        Extreme stress

 

 

 

 

John Campbell is good looking;

 

 

Strongly agree     Agree         Disagree     Strongly

    disagree

 

 

 

 

 

 

 

 

Sampling

 

How would you obtain a random sample of voters to predict the result of a general election in the UK