Tiêu đề Data & Variables .pdf ✅

abc CHAPTER DATA & VARIABLES 1 on different values in different subjects. that takes on characteristic in same subject of nn in indÅ6dual different time. or object that varies. pressure. parity, education. occupation, knowledge, skill, habit 1 I aspects of population under study. Raw materials of statistics A yet 01 values or Information about n variable that is measured or recorded on study subjects. Factual tnformataon obtained through measurement or observation on study subjects. ata Variables i Blood Pressure 120 mmHg. 80 mmHg hypertensive, hypotensive, normotensive. In these data arithmetic mean is meaningful & sometimes ratio is also meaningful e.g. 2.5 children per family is meaningful. A 04 children family has twice as many children as a 02 children family is also meaningful. Average angina) episode per week is meningful. 2. Qualitative data [categorical datal : Can not be measured but can be counted • • • Data that varies in kinds and used simply as a label to distinguish one from another. It focuses on meaning & experience, having no magnitude and expressed as rate, ratio, percentage, proportion. They have scale of measurement and measured on nominal scale (mostly) or ordinal scale. Data that provides answer to the question what type? Types of qualitative data Age Body weight 30 yrs. 70 yrs. 20 yrs old, young. TO kg. 90kg, 40 kg over weight, obese. under weight. 'NPFSOF DATA/ VARIABLE: Based on characteristics 1 i. Quantitative dato Inurnerical dgca): Can be measured or counted Datal that vanes in amount and can be measured numerically in terms of quantity. • It focuses on numbers & frequencies and expressed numerically as mean, range, etc. • They have scale of measurement and measured on ratio scale (mostly) or interval scale. Data that provides answer to the question how much? Types of quantative data: a. Continuous data Data that can take any (countless) value, even fraction or decimal within certain range. They are generated by measurement on continuous scale. e.g. Blood pressure, height, weight, blood glucose concentration. In these data anthmatic mean and ratio is meaningful. b. Discrete data Data that take only the whole (integral) numerical values or limited number of discrete values in a gxven Interval. They are generated by counting or enumeration. e.g. Number of children in family, hospital bed, number of student in class, number of patient admitted, number of pregnancy in reproductive life, number of asthma attack per week. number of previous pregnancy, number of angina) episode per week etc. a. b. Nominal data: These are unranked categorical/qualitative data Generated by counting only. Usually binary (dichotomous) showing only two categories e.g. sex. May be polychotomous showing more than two categories e.g. religion, blood group. Unordered (unranked) and mutually exclusive. m e.g. Sex, religion, blood group, marital status, eye color. In these data arithmetic mean & ratio is not meaningful. e.g. Ifunmarried, married & divorced are coded as 1, 2, 3 respectively and if you talk with an unmarried & a divorced subject; it will be meaningless to say that on average you talked with a married individual. Again you can't say divorced is 3 times good than unmarried. Ordinal data: These are ranked categorical/qualitative data Generated by counting and ranking. Have meaningful numerical order, so can speak about the rank. Mutually exclusive and ordered. e.g. Tumor grading, opininon (agree, disagree, neutral), prognosis (better, same, worse), official designation (Asst. professor, Assoc. professor), academic performance (GPA 5, GPA 4), disease condition (mild, moderate, severe) etc. In these data arithmetic mean & ratio is not meaningful e.g. if mild, moderate & severe are coded as 1,2, 3, respectively and if you have two patients one mild (1) and another severe (3); it will make no sense to say that on average the patients are moderate. Again you can not say severe patient is 3 times more risky than mild patient. Continuous data that looks discrete In practice, sometimes continuous data may look discrete because of the way they are measured and reported. For example; gestational age, ESR, blood pressure etc. Gestational age of babies is often reported in whole weeks, such as 38 week, 39 week, 40 week and so on. Therefore, gestational age appears to be discrete. It is however continuous data because it could be reported to a greater degree of accuracy through decimal, such as 38.5 weeks. Same is true for blood pressure and ESR. TYPES OF VARIABLE: Based on causal relationship with each other 1 I. Independent variable • Variable(s) that can be manipulated (changed) by the researcher to influence (cause) some change on the dependent variable. It is synonymous with cause, exposure, input, predictor, risk factor, explanatory variable, and stimulus. 08 09
abc biost•tist methodolog abc E. of research TYPES OF DATA: Based on the scale of measurement. Page 13 2. Dependent variable 1 Variable(s) that are influenced by alteration of independent variables. that describe the measurable outcome of the manipulation of independent variable. It is synonymous with outcome, output, effect, response. 3. Intervening (intermediate) variable Third set of variable through which independent variable affects the dependent variable. 1. Nominal data measured on nominal scale. 2. Ordinal data measured on ordinal scale. 3. Interval data measured on interval scale. 4. Ratio data measured on ratio scale. OTHER TERMS ASSOCIATED WITH DATA Alcohol intake Salt intake Hypertension Salt intake Poverty Less vitamin-A intake Blindness Decreased MI Increased HDL Hypertension MI (myocardial infarction) Hypertension Blindness Less vitamin-A intake Poverty Blindness Less vitamin-A intake Poverty 2. 3. 4. 5. 6. 7. 8. Primary data: (original data) 4. Confounding (extraneous) variable : Page 328 Variable(s) other than independent variable (IV) that may influence the dependent variable (DV) and exert confounding effect (mixing effect) between IV (exposure) and DV (outcome). Third variable(s) that might affect the relationship between independent variable (exposure) and dependent variable (outcome) and thereby distort the study result leading to spurious association or lack of association between independent variable and dependent variable. Confounders are independently related to both dependent (outcome) and independent (exposure) variable but not in a causal chain. Their effects can not be separated from the effects of independent variable. So, they need to be eliminated or controlled. Independent Dependent Confounding Obesity Hypertension Diabetes Gender / Smoking Age Smoking * A variable can be independent variable in one context but dependent variable or intervening • Data generated first hand by the researcher himself from the original sources. These are generated by observation, measurement, counting, experiment, survey, census, interview, focus group discussion, participatory observation etc. Secondary data • Data already existing; generated by some one else and have already been passed through the statistical process. • These are collected from records, documents, journals, newspaper, books, other studies etc. m Derived data Data derived from primary data or secondary data through some mathematical process. e.g. BMI derived (calculated) from body weight & height. Disease rates etc. Binary (dichotomous) data : Nominal type of qualitative data having just two catetories Nominal type of qualitative data expressing only two mutually exclusive categories where all subjects fall into one or other of the categories. e.g. Sex (male / female), cured / not cured, dead / alive, pregnant / not pregnant, disease / no disease, symptoms (yes / no), satisfy (yes / no). Univariate data • Proposition made on a single variable that incorporate single item of information. e.g. Birth rate of male baby, birth weight of newborn, crime rate in Dhaka city. Sex of newborn, prevalence of MI in Dhaka city etc. Bivariate data • Proposition made by relating two variables that incorporate two linked pieces of information. e.g. Birth rate of Rh+ve male baby, gender & birth weight of newborn. Data showing hypertension and MI Multivariate data • Proposition made by relating more than two variables that incorporate more than two linked pieces of information e.g. Birth rate of Rh+ve male premature baby Mother's age, newborn gender & birth weight of newborn. Data showing obesity, hypertension and MI Outlier (extreme values) • Data which is distinct from the main body of data and incompatible with the rest of data. • Extreme & unusual value (s) in a data set. variable or confounding variable in other context. 10 11
abc error but be true also. In Of student. if the marks of students ranges between 60-70 and if on] student secure 03 another student secure : 9. with expressions (both qualitative and quantitative) Qualitative expression Quantitative expression Outafvariablc mmHg 160 Hypertensive Rich Biood pressure Income Body weight 10. Independent event Tk. 50.000/ month .'AO kg 00 vears Under weight Young Two or events where occurrence ofone event does not influence or exclude the occurrence Two or more events which can occur simultaneously. e.g. gender & blood group. A boy with blood group B is possible. Il. Mutually exclusive event (variable) Two or more where occurrence of one event excludes the possibility of the occurrence of other event(s) simultaneously. Two or more events that can not occur simultaneously. e.g. sex, blood group, religion, death and etc. one can not be both male & female samultaneously 12. Mutually exhaustive event (variable) Variable classified exhausting its all possible categories so that no body get missed from this classlLcauon. e.g. BMI: <18. 18-30, >30. It is mutually exclusive & exhaustive. BMI: IS-3(). >30. It is mutually exclusive but not exhaustive. Sex: male, female. It is mutually exclusive & exhaustive. Marital status: unmarried, married, divorced, widow, widower. It is mutually exclusive & exhaustive. Marital status: unmarried, married. It is mutually exclusive but not exhaustive. 13. Dependent event (non indedependent event) Page 191 Two events where occurrence of one event depends on whether the other event has occurred. e.g. Probability of enteric fever given that widal test is positive OPERATIONAL DEFINITION OF VARIABLE/ DATA Statement of a vanable in workable and measurable term in actual research situation. The way by which researcher clarifies and defines the variable(s) under investigation. Purposes of operational definition: It is important to Provide unambiguous and consistent meaning of the terms used that otherwise can be interpreted in different ways by different peoples Make data collection and data analysis more focused and efficient. Ensure the collection of specific data that we exactly want. Reduce errors, 12 abc methodologt Data / variable Operational definitions Iodine deficient < 100 pg/L. Hypertension DBP>90 mHg Severe Iodine deficient UIE < 20 Under weight BMI < 18 Over weight BMI 25-29.9 Obese BMI 30 Anemic Hb < UIE : Urinary Iodine excretion BMI : Body mass index DBP : Diastolic blood presser Hb : Hemoglobin SCALES OF MEASUREMENT: Level of measurement 1 In ascending order of power over the preceding one; the scales of measurement are I. Nominal scale (nominal level) : Lowest level of measurement. m 2. Ordinal scale (ordinal level). 3. Interval scale (interval level). 4. Ratio scale (ratio level) • Highest level of measurement. * Data obtained from categorical variable are measured on nominal or ordinal scale and data obtained from numerical variable are measured on interval or ratio scale. NOMINAL SCALE (NOMINAL LEVEL) 1 It is a system of assigning numbers to the variable to label them only for identification and to distinguish one from another e.g. male-I, female-2. Here a variable is categorized in different catagories. Categories are designated by names (red, white, blue, green etc.) or numerals (ID number, room number, code number) but ordering of categories is meaningless. To work with non-numerical nominal data in statistical process we need to impose a numerical scheme on the data (e.g. male-I, female-2 or muslim.l, hindu-2, christian-3 or urban-I, rural-2, slum-3). In each of these cases numerical data have been artificially created but none of the numbers have any numerical meaning; so they are nominal because they are numerical in name only. Properties of nominal level of measurement. Categories are distinct and homogeneous. Mutually exclusive. Can not be measured or ordered but can be counted. Data can say one is different from another. Data can not say one is greater or smaller than another. ORDINAL SCALE (ORDINAL LEVEL) 1 It is a system of assigning numbers to the variable to label them for identification and ranking based on a scale having unequal interval size. • Here a variable is categorized in different catagories in ascending or descending order where intervals between the successive categories are unequal and unknown. So it expresses the relative quality of data. 13