View Article

Human research is based upon an inexact science. Both 
						
				
				
						
								quantitative and qualitative research methods use tools of 
						
				
				
						
								inquiry that are no better than the other. It is the question 
						
				
				
						
								not the method that is the guiding force, in other words 
						
				
				
						
								research must be knowledge driven not method limited. In 
						
				
				
						
								my last article I wrote about Qualitative Research. The 
						
				
				
						
								purpose of this article is to introduce Quantitative Research. 
						
				
				
						
								We hear so much talk about evidence based practice but: 
						
				
				
						
								
										What is evidence? 
								
						
				
				
						
								
										
												Evidence is the data:
										
								
						
				
				
						
								
										•
										upon which a judgement or conclusion can be based 
								
						
				
				
						
								
										•
										by which proof or probability may be established 
								
						
				
				
						
								
										•
										which serve to indicate or suggest 
								
						
				
				
						
								
										In deciding about evidence upon which to base your practice 
								
								
										you have to consider very carefully the limitations of 
								
								
										qualitative research 
										
												(based predominantly on narrative and 
										
								
								
										
												interpretation) 
										
										and quantitative research 
										
												(based on numbers). 
										
										The 
								
								
										‘gold standard’ regarding ‘evidence’ is the randomized 
								
								
										controlled trial yet often results, while statistically significant, 
								
								
										are not clinically significant 
										
												(or will not ‘work’ in the real world). 
										
										So 
								
								
										you will need to consider carefully the assumptions upon 
								
								
										which any quantitative research is based and its strengths 
								
								
										and weaknesses in the same way you would for qualitative 
								
								
										research.
								
						
				
				
						
								
										
												What is Quantitative Research? 
										
								
						
						
								
										
												Quantitative research is: 
										
								
						
						
								
										
												• 
												a formal, objective, systematic process of generating 
										
								
						
						
								
										knowledge 
								
						
						
								
										
												• 
												conducted to describe events, examine relationships, 
										
								
						
						
								
										investigate causes, determine effectiveness 
								
						
						
								
										
												Some people argue that quantitative research is reductionist, 
										
										
												unable to view the whole, and that it loses the human 
										
										
												element. They compare it with qualitative research which is 
										
										
												thought to be holistic and human focused. This, however, is 
										
										
												an inaccurate assumption. 
										
								
						
						
								
										Humans are limited. Even if we thought we knew what the 
								
						
						
								
										whole truth was there are always alternative possibilities, 
								
						
						
								
										
												perspectives, beliefs and interpretations. We may argue there 
										
										
												is a cause and an effect, that there are relationships between 
										
										
												variables, generalisabilities etc., but we are confined by our 
										
										
												thinking, world views, perspectives, conflicts, inconsistencies 
										
										
												and uncertainties. We can use and abuse statistics. In other 
										
										
												words there is no one truth. For example, we are pretty sure 
										
										
												that smoking causes lung cancer – yet humans who have 
										
										
												never smoked have been known to get lung cancer and not 
										
										
												all heavy smokers get lung cancer. 
										
								
						
						
								
										Most text books are very clear on quantitative research 
								
						
						
								
										
												methods and you will find topics such as reliability, validity, 
										
										
												significance, costs, benefits and ethics. As consumers of 
										
										
												research we need to know the implications of these central 
										
										
												issues. 
										
								
						
						
								
										
												Quantitative Methods 
										
								
						
						
								
										Quantitative research involves some form of assigning 
								
						
						
								
										numbers to objects or observations so that methods of 
								
						
						
								
										measurement can be utilized and analysed by statistical 
								
						
						
								
										
												techniques. These techniques range from the very simple to 
										
										
												the very complex. In quantitative research we work with data 
										
										
												that are often called variables.Variables are qualities, 
										
								
						
						
								
										
												properties, or characteristics of persons, things or situations 
										
										
												that change or vary and that can be manipulated or 
										
										
												measured. Some variables are obvious such as height or 
										
										
												weight. Other variables, such as anxiety traits, coping 
										
								
						
						
								
										
												behaviours, are not so obvious in that they are not so clearly 
										
										
												convertible into numbers. 
										
								
						
						
						
				
		
		
				
						Types of Variables
				
		
		
				
						
								There are a number of types of variables the main two being
						
				
				
						
								the:
						
				
		
		
				
						•
						
								
										
												
														Independent Variable
												
										
								
								(IV): a variable that is under the
						
				
		
		
				
						
								control of the researcher; one that is manipulated, and the
						
				
		
		
				
						• 
						
								
										
												Dependent Variable 
										
								
								(DV): a variable that responds to the
						
				
				
						
								independent variable.
						
				
		
		
				
						
								In an experiment the IV is controlled, however, IVs may also
						
				
				
						
								happen naturally where they can be observed and measured.
						
				
				
						
								An example of an IV in an experiment could be a ‘specific type
						
				
				
						
								of treatment’ and an example of an IV occurring naturally
						
				
				
						
								could be ‘anxiety.’
						
				
		
		
				
						
								The DV responds to the IV. It is not manipulated. Changes
						
				
		
		
				
						
								that occur in the DV are assumed to have occurred as a
						
				
		
		
				
						
								consequence of the change in the IV. The DV is observed and
						
				
				
						
								
										measured in order to understand, explain or predict the
						
				
				
						
								changes in the IV. For example, a new medication to prevent
						
				
				
						
								migraine may be experimented with. Participants receive
						
				
				
						
								either the new medication or a placebo (the IV) and the
						
				
				
						
								number of migraine attacks are measured (DV).
						
				
		
		
				
						
								Many qualities, properties or characteristics can be IVs or
						
				
		
		
				
						
								DVs, e.g. as you will note with ‘anxiety’. This is because
						
				
		
		
				
						
								independence or dependence are not inherent characteristics
						
				
				
						
								of the variable, they are a function of the role the variable
						
				
				
						
								plays. Note also that a change in the DV does not necessarily
						
				
				
						
								predict cause and effect; it may be relational in that the
						
				
				
						
								change occurs in relation to the change in the IV not because
						
				
				
						
								of it.
						
				
		
		
				
						
								A third type of variable to be cautious of is the 
								
										
												Extraneous
										
								
						
				
		
		
				
						
								
										
												Variable
										
								
								: a variable that is occurring that is not under the
						
				
		
		
				
						
								researcher’s control and sometimes the researcher is not
						
				
		
		
				
						
								aware of it. For example, a slight change in temperature may
						
				
				
						
								seriously affect an experiment.
						
				
		
		
				
						
								
										
												Levels of Measurement
										
								
						
				
		
		
				
						
								Levels of measurement determine the type of statistical tests
						
				
				
						
								allowed. If a researcher ignores the level of measurement
						
				
				
						
								and uses the wrong test the results are not reliable. There are
						
				
				
						
								four levels of measurement:
						
				
		
		
				
						
								1. 
								Nominal – the tests focus on frequencies
						
				
		
		
				
						
								2. 
								Ordinal – the tests focus on rankings
						
				
		
		
				
						
								3. 
								Interval – rankings have equal intervals between the
						
				
		
		
				
						
								numbers and an arbitrary zero point
						
				
		
		
				
						
								4. 
								Ratio – rankings have equal intervals and an absolute
						
				
		
		
				
						
								zero
						
				
		
		
				
						
								The level of measurement refers to the relationship among
						
				
		
		
				
						
								the values that are assigned to the attributes for a variable
						
				
		
		
				
						
								and are hierarchical. At the lower levels of measurement, e.g.
						
				
				
						
								nominal, ordinal, assumptions tend to be less restrictive and
						
				
				
						
								data analyses tend to be less sensitive. At each level up the
						
				
				
						
								hierarchy, the level includes all of the qualities of the one
						
				
				
						
								below it and adds something new. Higher levels of
						
				
				
						
								measurement e.g., interval or ratio are generally preferable.
						
				
				
						
								Let's assume that in the midwifery context the relevant
						
				
				
						
								attributes are "independent midwife", "core midwife", and
						
				
				
						
								"team midwife". For purposes of analyzing the results of this
						
				
				
						
								variable, we arbitrarily assign the values 1, 2 and 3 to the
						
				
				
						
								three attributes. The 
								
										
												level of measurement 
										
								
								describes the
						
				
				
						
								relationship among these three values. In this case, we are
						
				
				
						
								simply using the numbers as short placeholders for the
						
				
				
						
								lengthier text terms. We don't assume that higher values
						
				
				
						
								mean "more important" and lower numbers signify "less
						
				
				
						
								important". Here, we would describe the level of
						
				
				
						
								measurement as 
								
										"nominal".
								
						
				
				
						
								Knowing the level of measurement helps you decide how to
						
				
				
						
								interpret the data from that variable. When you know that a
						
				
				
						
								measure is nominal (like the one just described), then you
						
				
				
						
								know that the numerical values are just short codes for the
						
				
				
						
								longer names. Second, knowing the level of measurement
						
				
				
						
								helps you decide what statistical analysis is appropriate on the
						
				
				
						
								values that were assigned. If a measure is nominal, then you
						
				
				
						
								know that you would never average the data values or do a test
						
				
				
						
								on the data.
						
				
		
		
				
						
								In 
								
										ordinal 
								
								measurement the attributes can be rank
								‐
								ordered.
						
				
				
						
								Here, distances between attributes do not have any meaning.
						
				
				
						
								For example, on a survey you might code pain as no pain = 1; a
						
				
				
						
								little pain = 2; moderate pain = 3; severe pain = 4. In this
						
				
				
						
								measure, higher numbers mean 
								
										more 
								
								pain. But is distance
						
				
				
						
								from 0 to 1 same as 3 to 4? Of course not. The interval b
				
				
						
								etween values is not interpretable in an ordinal measure.
						
				
				
						
								In 
								
										interval 
								
								measurement the distance between attributes
						
				
				
						
								
										does 
								
								have meaning. For example, when we measure
						
				
				
						
								temperature, the distance from 10 
								‐
								20 is the same as distance
						
				
				
						
								from 20 
								‐
								30. The interval between values is interpretable.
						
				
		
		
				
						
								Because of this, it makes sense to compute an average of an
						
				
				
						
								interval variable, where it doesn't make sense to do so for o
				
				
						
								rdinal scales.
						
				
		
		
				
						
								Finally, in 
								
										ratio 
								
								measurement there is always an absolute zero
						
				
				
						
								that is meaningful. For example, in interval measurement
						
				
				
						
								ratios don't make any sense 
								‐
								30 degrees is not twice as hot as
						
				
				
						
								15 degrees (although the attribute value is twice as large).But i
				
				
						
								n ratio measurements you can construct a meaningful
						
				
				
						
								fraction (or ratio) with a ratio variable. Most "count" variables
						
				
				
						
								are ratio, for example, the number of women delivered in past
						
				
				
						
								six months. Why? Because you can have zero women
						
				
				
						
								delivered and because it is meaningful to say that "...in the
						
				
				
						
								past six months twice as many women delivered in our
						
				
				
						
								practice than in the previous six months."
						
				
		
		
				
						
								
										Types of Statistics
								
						
				
		
		
				
						
								Statistics derived from these levels of measurement can be
								
								
						
				
		
		
				
						
								
										DESCRIPTIVE or INFERENTIAL
								
						
				
				
						
								
										
												DESCRIPTIVE
										
								
								statistics help organize and give meaning to
						
				
				
						
								data. Statistics from all levels of measurement can be
						
				
				
						
								summarized and displayed visually as tables, charts, pie
						
				
				
						
								charts, or graphs.
						
				
		
		
				
						
								
										INFERENTIAL 
								
								statistics allow inferences to be made from a
						
				
				
						
								sample to the larger population from which the sample is
						
				
		
		
				
						
								drawn and allows hypotheses to be tested. Only the two
						
				
		
		
				
						
								higher order levels of measurement, interval and ratio, can be u
				
				
						
								sed to make inferences from the data. Because data is
						
				
				
						
								collected from a sample and not the whole population
						
				
		
		
				
						
								inference is always based on incomplete information and it is
						
				
				
						
								possible to make errors in accepting or rejecting a
						
				
		
		
				
						
								hypothesis.
						
				
		
		
				
						
								
										Hypotheses
								
						
				
		
		
				
						
								Hypotheses are statements about populations 
								
										(not samples). 
								
								A
						
				
				
						
								well formulated hypothesis clearly identifies the relationship
						
				
				
						
								between variables. The researcher’s aim is to accept or reject  
				
				
						
								the null hypothesis. The decision to accept or reject is based 
				
				
						
								upon how probable it is the findings are a true result or are
						
				
				
						
								due to chance and this probability is subject to error.
						
				
		
		
				
						
								
										
												What is a null hypothesis? 
										
								
								– that there is no relationship or
						
				
		
		
				
						
								difference between the variables being studied. For example,
						
				
		
		
				
						
								there is no difference in caesarean section rate between
						
				
		
		
				
						
								women whose LMC is an obstetrician and women whose
						
				
		
		
				
						
								LMC is a midwife. The symbol for the null hypothesis is Ho:
						
				
		
		
				
						
								The symbol for the alternative hypothesis is H
								1: 
								and states
						
				
		
		
				
						
								that there is a difference , for example, the rate of caesarean
						
				
		
		
				
						
								sections in women whose LMC is an obstetrician is
						
				
		
		
				
						
								significantly higher than the rate for women whose LMC is a
						
				
		
		
				
						
								midwife.
						
				
		
		
				
						
								
										
												Errors
										
								
						
				
		
		
				
						
								There are two main errors that can be made:
						
				
		
		
				
						
								
										TYPE 1 – 
								
								rejecting the null hypothesis when it is TRUE
								
								
						
				
		
		
				
						
								
										TYPE 11 – 
								
								accepting the null hypothesis when it is NOT TRUE
						
				
		
		
				
						
								These errors can be reduced by setting a level of significance
						
				
		
		
				
						
								before the study begins e.g. 0.05, 0.00001. The sign used for a
						
				
		
		
				
						
								type 1 level of significance is α= alpha, and for type 11 is β =
						
				
		
		
				
						
								beta..
						
				
		
		
				
						
								Question: 
						
						
								if a study were conducted 100 times at what level
						
				
		
		
				
						
								would the researcher feel confident about if they were to
						
				
		
		
				
						
								reject the null hypothesis?
						
				
		
		
				
						
								Answer: 
						
						
								if it were 5 times out of 100 the level of significance
						
				
		
		
				
						
								set would be 0.05. The researcher accepts the possibility that
						
				
		
		
				
						
								she would be wrong 5 times out of 100. Most social scientists
						
				
		
		
				
						
								accept this level. However, would a drug company trialling a
						
				
		
		
				
						
								new drug accept being wrong 5 times out of 100 i.e. 5 people
						
				
		
		
				
						
								might die. However, the company may be willing to accept
						
				
		
		
				
						
								one error in 10,000 or 10,000,000.
						
				
		
		
				
						
								
										
												Process for formulating a hypothesis
										
								
						
				
		
		
				
						• 
						
								An idea emerges
						
				
		
		
				
						• 
						
								Brain storm the idea
						
				
		
		
				
						• 
						
								Review the literature using key words from your ideas and
						
				
		
		
				
						
								brain storming session
						
				
		
		
				
						• 
						
								Identify the variables from the literature and your own ideas
						
				
		
		
				
						• 
						
								Formulate a research problem or question
						
				
		
		
				
						• 
						
								DEVELOP THE HYPOTHESIS
						
				
		
		
				
						
								
										
												Statistical Tests
										
								
						
				
		
		
				
						
								Statistical tests fall into two categories:
						
				
		
		
				
						
								PARAMETRIC or NON
								‐
								PARAMETRIC
						
				
		
		
				
						• 
						
								
										Parametric Tests 
								
								specify certain conditions concerning
						
				
		
		
				
						
								the parameters of the population from which the
						
				
		
		
				
						
								research sample is drawn. They are used to observe
						
				
		
		
				
						
								differences between natural groups e.g. age groups,
						
				
		
		
				
						
								experimental/control groups.
						
				
		
		
				
						• 
						
								
										Non
								
								
										‐
								
								
										parametric Tests 
								
								are also used to compare
						
				
		
		
				
						
								groups but only when data is collected at the nominal
								
								
						
				
		
		
				
						
								or ordinal level i.e. whether the frequencies in each
						
				
		
		
				
						
								group differ from what might have been expected and
						
				
		
		
				
						
								not by some extraneous factor.
						
				
		
		
				
						
								
										Types of Quantitative Research
								
						
				
		
		
				
						• 
						
								Experimental
						
				
		
		
				
						
								Random sampling
						
				
		
		
				
						
								Researcher control the experiment using control
						
				
		
		
				
						
								groups
						
				
		
		
				
						
								Researcher controls the manipulation of the
						
				
		
		
				
						
								independent variable
						
				
		
		
				
						• 
						
								Quasi Experimental
						
				
		
		
				
						• 
						
								Non
								‐
								experimental
						
				
		
		
				
						• 
						
								Correlational
						
				
		
		
				
						
								As you can see above Experimental research designs have
						
				
		
		
				
						
								three components.
						
				
		
		
				
						
								Lastly there are 
								
										two major principles 
								
								in the Quantitative
						
				
		
		
				
						
								Paradigm
						
				
		
		
				
						
								Reliability and Validity
								
								
						
				
		
		
				
						
								As consumers of research midwives will be assessing the
						
				
		
		
				
						
								reliability and validity of the measurement instruments in the
						
				
		
		
				
						
								studies they are reviewing to be certain they are appropriate
						
				
		
		
				
						
								and sound, in other words, that you can trust the evidence.
						
				
		
		
				
						
								Measurement tools are RELIABLE if they are consistent,
						
				
		
		
				
						
								accurate, precise and reproducible.
						
				
		
		
				
						
								Measurement tools are VALID if they measure what they
						
				
		
		
				
						
								are supposed to measure.
						
				
		
		
				
						
								For example, if your weighing scales are not calibrated so
						
				
		
		
				
						
								that they weigh consistently, accurately, precisely and the
						
				
		
		
				
						
								same weight can be reproduced (i.e. if it has not changed)
						
				
		
		
				
						
								then the scales are not reliable. When you read research
						
				
		
		
				
						
								articles critically think about whether the researchers have
						
				
		
		
				
						
								demonstrated the reliability of their instrument. Many
								
								
						
				
		
		
				
						
								questionnaires are not reliable.
						
				
		
		
				
						
								If your CTG is not placed accurately to clearly pick up the fetal
						
				
		
		
				
						
								heart rate and uterine contractions then the output graph is
						
				
		
		
				
						
								not valid (or reliable).
						
				
		
		
				
						
								Invalid or unreliable measures produce invalid and unreliable
						
				
		
		
				
						
								estimates of the relationships between variables and affect
						
				
		
		
				
						
								the internal validity of the study e.g. uterine contractions and
						
				
		
		
				
						
								fetal heart. The use of invalid or unreliable measures may lead
						
				
		
		
				
						
								to inaccurate generalizations and affect the external validity
						
				
		
		
				
						
								of the study. A famous case is the use of IQ tests that were
						
				
		
		
				
						
								based on Western concepts that disadvantaged minority
						
				
		
		
				
						
								groups and led to the conclusion that they were less
						
				
		
		
				
						
								intelligent.
						
				
		
		
				
						
								Sources of error can arise from the researcher or the
						
				
		
		
				
						
								participants, or from outside influences on the researcher,
						
				
		
		
				
						
								participants (e.g. bias, test fatigue) tests or measuring
						
				
		
		
				
						
								instruments (e.g. poorly designed or uncalibrated), and
						
				
		
		
				
						
								inconsistencies in the measuring or testing processes (using
						
				
		
		
				
						
								different assessors or instruments). Such sources of error
						
				
		
		
				
						
								have to be controlled for.
						
				
		
		
				
						
								
										So, returning to: What is Evidence?
								
						
				
		
		
				
						
								Evidence is the data:
						
				
		
		
				
						• 
						
								upon which a judgement or conclusion can be based. The
						
				
		
		
				
						
								data must be 
								
										reliable and valid.
								
						
				
		
		
				
						• 
						
								by which proof or probability may be established based
						
				
		
		
				
						
								upon 
								
										reliable and valid 
								
								data.
						
				
		
		
				
						• 
						
								that is 
								
										reliable and valid 
								
								and serve to indicate or suggest.
						
				
		
		
				
						
								Don’t be fooled into accepting ‘evidence based research’ that
						
				
		
		
				
						
								has not met the criteria for reliability or validity. Don’t just
						
				
		
		
				
						
								accept the rhetoric!
						
				
		
		
				
						
								
										Gill White
								
						
				
		
		
				
						
								
										Professor (Education Adviser)
								
						
				
		
		
				
						
								
										OSNI <?xml:namespace prefix = st1 ns = "urn:schemas-microsoft-com:office:smarttags" /?><st1:country-region><st1:place>Oman</st1:place></st1:country-region>