Methods of Data Analysis, Presentation and Interpretation Victoria 0. Oladoyin, was urn wir rma Outline Recap, ‘Data analysis ‘Data presentation + Data interpretation RECAP What is Biostatistics? + Itis.an applied science that is concerned withthe application Cf statistical methods to health and health related data + It deals withthe scientific methods forthe collection, ‘organization, summarization, analysis, presentation and Interpretation of health and health related data. How do we define data? + Data Is a collection of related observations or information + Its the raw information from which statistics are created + tis considered as raw data if thas not been treated statistically + itis called a dataset when entered into the computer ‘Types of data * Quantitative data Data that expressed in numerical terms Data can be explored statsteaily ~ el reslts that can be generalised to some larger population. ata often gathered trough surveys and questionnaires that ae carey developed and structured Qualitative data ata that represented ether in a verbal or narrative format ‘7 Rsimpe way to lok a qualitative data to think oft i the frm of words + ga are collected through ndvidal interviews and focus group dstasions ung semi-structured or unstructured topic guides, opened fended questionnaire temsDATA ANALYSIS What is data analysis? + itis a process of collecting and organizing data so that 2 researcher can come to a conclusion which can be used to support decision-making. + itis process of obtaining raw data and converting it into Information useful for decision-making by users. + itis the process of bringing order, structure and meaning to the mass of collected data, + Itisa time-consuming, creative, and fascinating process Why do we need data analysis? stthelps to: Answer questions ‘Describe and summarise data ‘Test hypotheses “Derive important information Data analysis process * Develop research questions + Develop research objectives + Data collection + Data processing ~ organizing, entry + Data cleaning + Data analysis Quantitative data analysis “Ios the ee ana and iteration of gues nd ‘numbers, and + Attempts to find ratlonale behind the emergence of main fingings Quantitative data analysis ra have + Asa first stop its essential to check whether the ‘been entered correctly. + Allinferences drawn and conclusions reached as a result of ‘statistical manipulation ofthe data rest on the assumption all measures were vad and reliable, and that the data had been entered correct + No amount of statistical manipulation ean ever make up for ‘erorsin data collection and entryQuantitative data analysis + Analysis begins with exploring and describing the data and confirming its characteristics. +The first step in doing s0 Isto obtain the frequencies. + An unexpected value in the frequency table (e.g. age 99 in a study (of children) indicates incorrect data entry or incorrect recording of the information, One needs to go back and check. + Similarly, missing values may become obvious from frequency tables and one may have to go back and do a careful check, + This step is referred to as data cleaning, Quantitative data analysis “The next step after cleaning the data is to generate descriptive statistics using summary statistics 2s appropriate ‘Measures of central tendency for quantitative (numerical) variables ‘7 Measures of dispersion for quantitative (numerical) variables ‘Proportions and percentages for qualtatve (categorical) variables + Subsequently inferential statistics can be carried out to explore the relationships between variables Qualitative data analy: + This is process that seeks to reduce and make sense of vast amounts of information, often from diferent sources, so ‘that mpremions that shed light on a rescarch question car emerge, “tt isa process where you take descriptive information and offer an explanation or interpretation. +The information can consist of interview transcripts, documents, surveys, pictures, videos etc Qualitative data analysis + Involves the identification of common patterns within responses and analysing them critically in order to achieve research alms and objectives *Tt isa search for general statements about relationships among categories of data Qualitative data analysis * The first thing you need todo sto familarise yourself with Your data. This involves reading and re-reading your material (ata) ints entirety. “Make notes of thoughts that spring to mind and write summaries of each transcript or pece of data that you wil analyse. + As your aim is to condense al ofthis information to key ‘themes and topics that can shed light on your research ‘question, you need to start coding the material Qualitative data analysis + Acode Isa word or a short phrase that descriptively captures the tessence of elements of your material (e.g. quotation) and Is the frat step n your data reduction and interpretation, * Coding isa long, slow and repetitive process, and you are encouraged to merge, split up or rename codes as you progress. ‘There is no fxed rule on how many codes you should aim for, but if you have more than 100-120 codes, It is advisable that you begin to ‘merge some of your codes. "Once you have coded all of your material you need to start abstracting themes from the codes.Organising themes Working street children cannot afford to go to school Working street children are stressedData analysis software + Qualitative data analysis * Quantitative data analysis “vivo SPSS itias1 SAS “HyperRESEARCH YEpinfo Max QDA stata vite. mn ete. DATA PRESENTATION Data presentation "The best way to present your findings depends on: ‘The audience “The purpose “The data gathered \7The analysis undertaken ‘Ways of presenting quantitative data Tables pawns + Diagrams Frequency distribution tables + The primary abjecive sto enable the researcher have an Initial exploration ofthe data investigator observes the distribution of the entire sample. data at afferent values ofeach variable ‘+ Expressed in simple tables that usually has two main columns and a numberof rowsFrequency distribut in tables + Tables for qualitative variables ‘Easy and straight forward to construct + Tables for quantitative variables Values of variables could be many Easier to construct frequency tables for discrete variables than for continuous variables Qualities of a good frequency table + Simple and not too information loaded * Clear tite indicating what the table contains? Where the data was collected? and when the data was collected? Good labeling of rows and columns with specific units of measurements ekg for weight + Should be self explanatory and self sufficient ‘The distribution of health workers at the Care Centre, Ondo, 2010 Frequency distribution tables + To manage resources and also represent similar information egsov- demographics, reproductive history, contraceptive use history ete, you can Include mare than cone variable ina sable Presentation of ungrouped data in a frequency distribution table + For many continuous variables and even some discrete variables, its almost always necessary to regroup the data ‘over a range of values of the variables + This may be done using: 4, Grouping intervals or 2. Application ofthe STURGES rule to determine the ‘minimum number of cass intervals (ie classes) for & frequency table for any given set of data sing STURGES rule to deterr Presentation of ungrouped data in a frequency table ie the minimum number of class interval + STEP 1: Determine the number of intervals or class ke 1+ 3.3logon ‘Where n= sample sie and k = number of class intervalPresentation of ungrouped data in a frequency table using STURGES rule to determine the minimum number of class interval + STEP 2: Determine the class width (\.¢. width of class interval) ‘lass width = largest variable = smallest variable ¥ + STEP3:Sort the data into appropriate class intervals and count the ‘number of checks in each class interval to obtain the frequency Class Exercise + Question: The data on the weight (in kilograms) of 80 pregnart ‘women attending the antenatal clinic of Idere maternity heath ‘centre in May 2009 is as follows: + 88,62,76,93,90,68,82,75,84,68,59,71,85,75,93,60,73. 78,95,63,72,62,74,69,68,60,77,94,75,82,78,66,79,60,83,. 75,61,89,78,79,65,76,65,76,71,75,85,78,97,67,63,79,76,62,63, 74,78,88,57,73,80,85,75,76,85,77,63,72,81,73,67,67,86,72,72 + Present the above data ina frequency table using STURGES rule to determine the minimum number of class interval Answer to class exercise + STEP 1: Determine the numberof intervals or class k=143.3ogen where n=-sample size and k= number of cass interval k=1633 (2.9031) k=1+6.2802 k=7.2802 Answer to class exercise (Cont’d) + STEP 2: Determine the cass width (i.e. width of class interval) class width = largest variable ~ smallest variable k largest variable = 97 smallest variable = 57 Therefore the class width = 97-57. 7 Class width =5.7143 =6 ‘Answer to class exercise (Cont'd) + STEP3:Sort the data into appropriate class intervals and count the number of checks in each class interval to obtain the frequency Frequency cistibution table for weight of pregnant women attending Idore maternity health centre in May 2008 oe “0 ajayFrequency distribution table for weight of pregnant women attending Idere maternity health centre in May 2009 Percent Weight class intervals Frequency ayer eee he es ins ea mip 63-68 13 16.25 69-74 4 17.50 75-80 23 28.75 81-86 n 13.75 87-92 4 5.00 93-98 6 7.50 100 Total 80Presentation of data using diagrams or graphs Types of diagrams/graphs “Makes the information simple and easy te comprehend ‘quantitative variable“ Quatative variable +-Brings out cleary the dstrbutional form ofthe data which “Histogram YBaechart als id futher sais nalts “Frequency “Pie chart polgon/requency curve Seater diagram 1p ore sepa? Histogram “+ Are graphs drawn as slender rectangles adining each other fon a rectangular co-ordinate axis and used for presenting Sistribution of continuous variables (Le. data on interval or rato scale) + There ae no spaces between the bars + The width of each rectangle is equa tothe class interval + Histogram showing the serum HOL Cholesterol evel in 2 sample of 20-23 years old patients at State Specialist Hospital, Ondo in 2010 ™ Frequency polygons + Are graphical devices for understanding the shapes of ‘distributions + They serve the same purpose as histograms, but are especaly helpful for comparing sets of data + Useful in comparing several relative frequencies + Gotten by joining the mid-points of the tp ofthe rectangles or blocks ina histogram by straight lines + ifthe mid-point ae into many more groups there wil be ‘many points and the polygon wil presenta curve. The curve becomes smoother asthe number of asses increases CeF[['['“ = —————————— Frequency polygon + Frequency polygon showing the test Scores of 200 Level UUNIMED students ln Bostaistles, 2017 amen ereceseees Ts els ws ws oe wsFrequency polygon + Frequency polygon showing the height 0f 200 Level UUNIMED students in Biostatistics, 2017 raveney Pobre patting raguansay of aru Chalstara for ‘es hgesth Stand ges 38-8 Scatter diagram + Suitable for presenting the relationship between two ‘quantitative variables Scatter diagram- Positive relationship | + we Scatter diagram- Negative relationship 1 eae Scatter diagram- No relationship »ar chart + Suitable for presenting nominal data + Constructed as slender rectangles on two co-ordinate axes (x.y) + The rectangles do not adjoin each other like the histogram. ‘They are separate and distinct looking bars + Each bar represent each category of the categorical variables, being presented + The length or height of the bars correspond to the absolute or relative frequency of observations in each category of the variable Bar chart showing job distribution of health workers at State Specialist Hospital, Ondo in 2010 ‘Consultant Medical Officer = Midwife Nurse w CHEW. Bar chart showing job distribution of health workers at ‘State Specialist Hospital, Ondo in 2010 PIE CHART. + Useful type of diagram particularly when itis desired at 2 glance how a total population is made of its components + Usually for qualitative variables Pie chart showing job distribution of health workers at State Specialist Hospital, Ondo in 2010 Presentation of The imouaisaton cand was mentioned i all ‘qualitative d the FGD sessions as away of eeping + For qualitative data, tumunisaion records which anlar be sed Youpresentthe oases immunisation coveag. ase or sentence Brae Semmes FSD) rupert se hare shots dl gen Youre aking tio, hn wl ld ong Be cand 9 that hey oa wnt there Dat th cad as Fou mani” (FGD 2)vs/ul) DATA INTERPRETATION Data interpretation ‘Data interpretation isthe process of attaching meaning to 3 data “Interpretation of data demands fair and careful judgement «The same data can be interpreted in different ways “So itis helpful to involve others or take time to hear how different people interpret the same information Data interpretation = When interpreting your data, you position your study findings In ‘elation to other related studies So basically, you report what others have found in comparison to ‘what you found and tell your audience why you think tis Iss. + Part of interpreting information is identifying the lessons learnt, ‘reporting what is new, reporting what was expected, reporting findings that surprised you, reporting things that you do not ‘understand very well and making recommendations for further study