Biostatistics: Data Collection & Sampling Methods PDF

Biostatistics rosmo DATA COLLECTION AND SAMPLING METHODS census – actual measurement of all elements from population...

Biostatistics rosmo DATA COLLECTION AND SAMPLING METHODS census – actual measurement of all elements from population – can be viewed as sample that includes entire population Key Words parameter – numerical characteristics of population statistics – collection, organization, presentation, analysis, and interpretation of numerical data to obtain information statistic – measured characteristic of sample – steps 1. data collection – obtaining information Sampling Methods 2. organization – presenting data in tables/graphs/charts probability sampling – every element in population has equal chance for logical and statistical conclusions – simple random: chance methods/random numbers 3. analysis – extracting relevant information – systematic: selecting every nth subject for formulating numerical description – stratified random: choosing representatives from each group 4. interpretation – drawing conclusions from data – cluster: choosing an entire group as subjects – 2 major areas nonprobability sampling – some degree of personal subjectivity 1. descriptive statistics – describing properties – convenience: cheap, quick, and chooses based on availability 2. inferential statistics – prediction, estimation, conclusion : for quick user opinion polls/pilot testing biostatistics – specialized for fields of biology, medicine, and health – quota: similar to stratified random sampling, – role of biostatisticians but requires people who match characteristics of segment a. specialists of data evaluation – purposive/judgmental: based on previous ideas of population b. make public health decisions from findings : intentional selection based on characteristics c. influence change at policy-making level Levels of Measurement d. improve science, bridging gap between theory and practice Note: Measurement assigns numerical value to variable. – importance in clinical research and public health nominal – only number for identifying membership in category a. design and development of research framework – mutually exclusive, exhausting categories without rankings b. data monitoring and management – e.g. zip code, gender c. data analysis and reporting ordinal – greater/lesser than; no precise differences between ranks d. clinical trial – e.g. preferences, rankings, grade e. enriching patient care interval – greater/lesser than with unit of measurement f. data trends – ranking and precise differences between units of measure g. epidemiological studies – no meaningful zero variable – characteristics/attribute with different values – e.g. IQ – qualitative/attributes ratio – interval with absolute zero and multiples are meaningful – quantitative – true ratios exist a. discrete –values in interval with gaps between them – e.g. height – whole number DEMOGRAPHY b. continuous – value at any point along interval – decimals Definition of Demography data – values that variables can assume – size, composition, and geographic distribution of populations – primary: generated by a researcher – big three for stability/change: birth, death, migration – methods of collecting primary data – central component of societal contexts and change a. surveys – solicit information from people Composition of Demography ▪ self-administered/mail survey – cover letter + postage basic demographic features – age, sex, family, household status ▪ personal interview – flexible but expensive social and economic – language, education, ethnicity, religion, income ▪ telephone interview Distribution of Demography b. direct observation – watch/listen  count/measure multiple levels – local, regional, national, global c. experiments – identify cause-and-effect types of boundaries – political, economic, geographic – secondary: gather by someone else for other purpose Tools of Demography – methods of collecting secondary data count – absolute # that is the basic of statistics a. internal – generated by own firm/organization rates – frequency – e.g. annual reports – crude rates: computed for an entire population b. external – gathered outside firm for other purpose – specific rates: computed for a specific subgroup – e.g. government agencies, published sources, ratio – subgroup to subgroup relationship commercial suppliers proportion – subgroup to population relationship questionnaire/data collection instrument – filled out by respondent or administered by interviewer constant (K) – unchanging, arbitrary # – 3 types of questions a. multiple choice c. open-ended cohort measures – events of a group with shared demographic exp. b. dichotomous – e.g. birth (most common; annually listed), marriage, school class – basic points a. must be short, simple, and clear period measures – events of all or part of a population; “snapshot” b. no leading questions Population Composition and Its Parameters c. no questions that may cause hesitation to answer Note: Population is a group of individuals of same species living and interbreeding within a given area. Inferential Statistics population size – # of individuals in a geographic range population/universe – entire set of objects of interest sample – smaller number/subset of objects within population 1 out of 2 Biostatistics rosmo 𝒏𝒐.𝒐𝒇 𝒊𝒏𝒅𝒊𝒗𝒊𝒅𝒖𝒂𝒍𝒔 graphical – visual/picture with numerical info density – population size and space ( ) 𝒖𝒏𝒊𝒕 𝒂𝒓𝒆𝒂 – histogram: contiguous vertical bars age structure – cohorts: age-specific categories (e.g. juvenile/subadult) of various heights fecundity – # of offspring produced : no gap between bars morality – # of deaths in a population, counterbalance to fecundity sex ratio – males and females Estimating and Projecting Population – frequency polygon: lines that connect points plotted for frequencies at midpoints of classes population estimate – annual size of a population between census periods or current year – data involved 1. population change (migration, fertility, mortality) 2. census results 3. change in population size projection – size of population in the future Basic Demographic Equation – ogive/cumulative frequency graph P2 = P1 + B – D + I – O population population natural increase net migration at any time at previous time (birth - death) (in - out) Vital Statistics vital statistics – vital events (e.g. birth/natality, death/ mortality, marriages/nuptiality, divorces) DATA PRESENTATION Presentation of Data Note: Raw data is not manipulated beyond original collection. textual – paragraphs that enumerate, emphasize, and identify data tabular – tables with columns and rows – e.g. 1. frequency distribution table – has classes and no. of cases 2. categorical frequency distribution table – specific categories like nominal/ordinal level data – stem-and-leaf plot: table with graph 𝒇 – frequency percentage: % = 𝒙 𝟏𝟎𝟎 : only for small data with positive values and >1 digit 𝒏 Other Concepts and Formulas – relative frequency: decimal/fraction equivalent of total # 𝑧2 ▪ 𝑝 ▪ (1−𝑝) 3. grouped frequency distribution – used when data is large 𝑒2 sample size (n) = 𝑧2 ▪ 𝑝 ▪ (1−𝑝) – rules a. 5-20 classes 1+ 𝑒2 ▪ 𝑁 b. class width is odd N = population size c. classes are mutually exclusive, continuous, and exhaustive z = z-score d. classes have equal width, E = margin of error (in decimal) except when it is open-ended p = standard deviation (0.5 if none) – steps 1. range – 𝑹 = 𝑴𝑨𝑿 − 𝑴𝑰𝑵 𝑁 Slovin’s formula (n) = 1+𝑁𝑒 2 2. no. of classes – 𝒌 = 𝟏 + 𝟑. 𝟑𝟐𝟐 𝒙 𝐥𝐨𝐠(𝒏) sampling error – expected error from observations in sample – round off k non-sampling error – mistakes in acquisition of data or observations 𝑹 3. class size – 𝒄= – non-responsive error and selection bias 𝒌 – round off c magic number 30 – rule of thumb for minimum sample size 4. class interval – lower limit and upper limit – application of Central Limit Theorem 5. frequency of each class 6. true class boundaries – continuity property 𝑳𝑻𝑪𝑩+𝑼𝑻𝑪𝑩 𝑳𝑳+𝑼𝑳 7. class mark – 𝑪𝑴 = = 𝟐 𝟐 – midpoint of a class 8. cumulative frequency – total # of observations in >CF or