LE362 - Research Methods - Data Collection-Quantitative.pdf

Kwame Nkrumah University of Science & Technology, Kumasi, Ghana LE 362: Research Methods II Data Collection-Quantitative Anthony Adjei-Twum, PhD Department of Land Economy Faculty of Built Environment, CABE  COURSE OUTLINE In...

Kwame Nkrumah University of Science & Technology, Kumasi, Ghana LE 362: Research Methods II Data Collection-Quantitative Anthony Adjei-Twum, PhD Department of Land Economy Faculty of Built Environment, CABE  COURSE OUTLINE Introduction to Research Data Collection Data Analysis & Interpretation Ethics Dissertation report writing www.knust.edu.gh www.knust.edu.gh  INTRODUCTION Two categories of data collection will be discussed here ⮚ Collection of Quantitative Data ⮚ Collection of Qualitative Data www.knust.edu.gh www.knust.edu.gh Steps Involved In Collecting Quantitative Data Administer data collection Identify or develop instrument Specify needed information Obtain Permission Identify people & www.knust.edu.gh www.knust.edu.gh place to study  Identify People & Place to Study ⮚ Identify unit of analysis: ⮚ It is the levels at which data is to be collected – this involves identifying the individuals (e.g. students, teachers, parents, adults, farmers, employees, managers, etc.) or organizations (e.g. schools, banks, etc.) from which data is to be collected. www.knust.edu.gh www.knust.edu.gh  Specify the Population and Sample ⮚ Specify the Population and Sample Selected sample should be representative of target population to ensure appropriate conclusion about the population. ❖ Target population (sampling frame) is a group of individuals (or a group of organizations with some common defining characteristic that the researcher can identify and study. ❖ Sample is a subgroup of the target population that the researcher plans to study for generalizing about the target population. www.knust.edu.gh www.knust.edu.gh  Reasons for Specifying Sample ⮚ When it would be impracticable to survey the entire population. ⮚ When budget constraints prevent surveying the entire population. ⮚ When time constraints prevent surveying the entire population. ⮚ When the needed data has been collected but www.knust.edu.gh www.knust.edu.gh the results is quickly needed.  Sampling ⮚ Sampling is the process by which you select a number of individual cases from a larger population. ⮚ It addresses the two questions: 1. Who is in your study? Studies that involve the use of living data from the subjects, respondents, participants, or collaborators. 2. What is in your study? Studies that involve the use of non living data (e.g., content analysis of text or images) www.knust.edu.gh www.knust.edu.gh  SAMPLING STRATEGIES Probability Non-probability sampling Sampling Multistage Simple cluster Random Convenience Snowball Systematic Stratified Sampling Sampling Selecting individuals because they are available, Selecting individuals from the convenient, and represent www.knust.edu.gh www.knust.edu.gh population who are representative some characteristic the of that population. investigator seeks to study  Probability Sampling Simple Random sampling Selecting participants (or units, such as schools) for the sample so that any individual has an equal chance of being selected from the population. Procedure 1. Assign a number to each individual (or site) in the population. 2. Use a random numbers table, available in many statistics books or Research Randomizer (an online software to generate random numbers), to select the individuals (or www.knust.edu.gh www.knust.edu.gh sites) for the sample. Illustration Random Number Generator www.knust.edu.gh www.knust.edu.gh https://stattrek.com/statistics/random- number-  50 Random Numbers 0956 0508 0450 0795 0280 0700 0033 0439 0694 0791 0736 0936 0606 0294 0676 0389 0868 0803 0856 0962 0374 0253 0157 0530 0776 0050 0771 0709 0670 0654 0073 0968 0761 0148 0855 0326 0807 0366 0598 0408 0179 0411 0431 0295 0191 0192 0768 0327 0064 0341 Specs: This table of 50 random numbers was produced according to the following specifications: ❑ Numbers were randomly selected from within the range of 1 to 1000. ❑ Duplicate numbers were not allowed. www.knust.edu.gh www.knust.edu.gh ❑ This table was generated on 7/4/2022. Systematic Sampling Involves choosing every nth individual or site in the population (sample frame) until you reach your desired sample size. Use sample fraction to Procedure determine frequency of selection Calculate sample fraction Use random number to select the first case Number each of www.knust.edu.gh www.knust.edu.gh the cases uniquely Illustration Assumed that the Regional Land Administrator wants to study land owners’ satisfaction with the land registration process in the region. Will use an interval of 5 (i.e., 200/1000) to select land owners Identify owners on the register, say 1000 He may study a percentage, say 20% registered land owners www.knust.edu.gh www.knust.edu.gh Stratified Sampling Researchers divide (stratify) the population on some specific characteristic (e.g. gender) and then, using simple random sampling, samples from each subgroup (stratum) of the population (e.g. females and males). When is it use? ❖ When the population reflects an imbalance on a characteristic of a sample. ❖ When a simple random sampling procedure would yield fewer participants in a specific www.knust.edu.gh www.knust.edu.gh category. Procedure for Stratified Use simple random or Sampling systematic sample to select your sample Number each of the cases uniquely within each stratum Divide the population by the stratum (e.g. men and women) Choose stratification variable www.knust.edu.gh www.knust.edu.gh Multi-stage Sampling It involves a situation where the researcher chooses a sample in two or more stages because either the researcher cannot easily identify the population or the population is extremely large When is it use? ❖ Where it is difficult to obtain the complete list of members of the population. Example: A researcher wants to understand the farming habits among people in Ghana. For this www.knust.edu.gh www.knust.edu.gh study, he requires a sample size of 300.  Districts in Ghana Districts in Ghana Greater Accra = 33 Bono East = 11 Ashanti = 43 Savannah = 7 Western = 14 Northern = 16 Western North = 9 North East = 6 Eastern = 33 Upper West = 11 Central = 22 Upper East = 15 Ahafo = 6 1. Bono He chooses = 11 the number of regions using simple random sampling (or any other probability sampling). Select, say, 10 regions. 2. He then choose 6 districts within each region using the systematic sampling method (or any other probability sampling). 3. He then choose 5 households from each district using the systematic sampling or simple random sampling method. You will end up with 300 houses that you can include inwww.knust.edu.gh the sample www.knust.edu.gh group for research.  Non-Probability Sampling Convenience Sampling The researcher selects participants based on their willingness and availability for the study. The individuals may not be representative of the population Snowball Sampling In snowball sampling , the objective of the researcher is to ask participants to identify others www.knust.edu.gh www.knust.edu.gh to become part of the sample.  Specify Need Information This step involves: 1. the identification of the variables in the questions and the hypothesis - independent, dependent, and control variables 2. Defining the variables operationally – this is the specification of how you will define and measure the variable in your study. This may be from definitions used by other authors or previous studies. 3. Considering types of information and measures - (a) Measures of individual performance (b) Measures of individual attitude (c) Observation of individual behavior (d) Factual information  Types of Quantitative Data and Measures Types of Data Types of Tests, Definition of the Type of Test, Instruments, or Instruments, or Document Documents to Collect Data Measures of Various forms of tests or A test where the individual’s individual personality assessment. grade is a measure of how performance (e.g. Achievement test) well he or she did in comparison with a large group of test takers Measures of Affective scale An instrument that measures individual positive or negative effect for attitude or against a topic Observation Behavioural checklist An instrument used to record of individual observations about individual behaviour behaviour Factual Public documents Information from public www.knust.edu.gh www.knust.edu.gh information sources that provides data about a sample or population  Designing Instrument Is there an Can you modify the instrument available existing instrument to measure your to suite your study? variables? www.knust.edu.gh www.knust.edu.gh  Features of a Good Research Instrument A good research instrument must: ✓ Be valid and reliable ✓ Be based on a conceptual framework, or the researcher's understanding of how the particular variables in the study connect with each other ✓ Be able to gather data suitable for and relevant to the research topic ✓ Be able to test hypothesis and/or answer proposed research questions under investigation ✓ Be Free of bias and appropriate for the context, culture, and diversity of the study site ✓ Contain clear and definite instructions to use the instrument www.knust.edu.gh www.knust.edu.gh  Constructing Data Collection Instrument 1. Clearly define and individually list all the specific objectives, research questions or hypotheses, if any, to be tested. 2. For each objective, research question or hypothesis, list all the associated questions that you want to answer through your study. 3. Take each question that you identified in Step 2 and list the information required to answer it. 4. Formulate question(s) that you want to ask of your respondents to obtain the required information. www.knust.edu.gh www.knust.edu.gh Steps Use clear Administer the language, instrument make sure the to a small number answer options do of individuals and not overlap, and make changes Include personal, pose questions based on their attitudinal, and that feedback behavioural are applicable to Perform a pilot questions; sensitive all participants test of the questions; and Use strategies for questions closed-and open- good question ended questions construction Write different types www.knust.edu.gh www.knust.edu.gh of questions.  Scales of Measurement ❑ These are response options to questions that measure or observe variables in categorical or continuous units. ❑ Understanding scales of measurement enables assessment of instrument quality and determination of appropriate statistics for data analysis. Types of Scales of Measurement i. Nominal Scales \ Categorical Scales ii. Ordinal Scales iii. Ratio Scales www.knust.edu.gh www.knust.edu.gh iv. Interval Scales Continuous Scales  Nominal Scales or Categorical Scales ❖ It involves provision of response options for participants to check one or more categories that describe their traits, attributes or characteristics ❖ They have no order Examples 1. Sex, divided into the two categories of male and female 2. A checklist of “yes” or “no” responses Examples of Questions 1. How much education have you completed? ___No college___ Some college ___Bachelor’s degree ___ Graduate or professional work 2. What is your class rank? _______ First ______ Second Upper www.knust.edu.gh www.knust.edu.gh _______ Second Lower ______ Pass  Ordinal Scales or Ranking scales ✔ It provides response options where participants rank from best or most important to worst or least important some trait, attribute, or characteristic ✔ The scales have an implied intrinsic order Examples 1. A researcher recording individual performance in a race for each runner from first to last place. 2. Attitudinal measures asking participants to rank in order of importance (“highly important” to “of no importance”) or the extent (“to a great www.knust.edu.gh www.knust.edu.gh extent” or “to a little extent”) of topics.  Examples of Questions Using Ordinal Scale 1. Has the Estate Officer addressed your concerns? _______ Not at all ______ To a small extent _______ To some extent ______ To a great extent _______ To a very great extent 2. Rank your preference for type of undergraduate- level instruction from 1 to 4 _______ Activity-based learning _______ Lecture _______ Small-group learning _______ Discussion www.knust.edu.gh www.knust.edu.gh Interval Scales or rating scales or continuous scales ✔ They provide “continuous” response options to questions with assumed equal distances between options. ✔ These scales may have three, four, or more response options. Example ❖ Likert scale (strongly agree to strongly disagree) is an example of scales which theoretically, has equal intervals among responses. www.knust.edu.gh www.knust.edu.gh  Ratio Scales or true zero scale ❑ It is a response scale in which participants check a response option with a true zero and equal distances between units. Examples ❑ the height of individuals (e.g. 40 metres, 50 metres). ❑ income levels (from zero cedis to GHS 10,000 in increments of GHS 2,000). www.knust.edu.gh www.knust.edu.gh Note: ✓ Researchers who undertake quantitative research usually employ both categorical and continuous scales on a single instrument ✓ In terms of carrying out strong statistical analysis, interval scales are the best as they tend to provide the most variation of responses. ✓ You can always convert continuous scales to ordinal or nominal scales but not from ordinal or nominal scales to continuous scales www.knust.edu.gh www.knust.edu.gh Examples of Questions Using Quasi interval or interval/ratio scale 1. School is a place where I am thought of as a person who matters. _______ Strongly agree _______ Agree _______ Undecided _______ Disagree _______ Strongly disagree 2. Colleges and universities should conduct research to solve economic problems of cities. ______ Strongly agree _______ Agree ______ Undecided _______ Disagree www.knust.edu.gh www.knust.edu.gh ______ Strongly disagree  Types of Questions Background or Demographic Questions ❖What is your age?____________________ ❖How many years of service have your completed?_________ Attitudinal or opinion Questions ❖How much do you agree or disagree with this statement: Most days I am enthusiastic in serving clients. ______ Strongly agree ______ Agree ______ Neither agree nor disagree ______ Disagree ______ Strongly disagree Behavioural Questions Do you smile while serving clients? ______ Yes www.knust.edu.gh www.knust.edu.gh ______ No Sensitive Questions Must be developed and use with care ❖Instead of : Have you taken bribe before? ❖In the past surveys, men have reported that at some point in their lives they have taken bribe. This could have happened before you joined the Commission, during adolescence, or as an adult. Have you ever? Open-and Closed-Ended Questions ❖Surveys consist mainly of closed-ended questions: There are many reasons why people wish to register their lands. What is your most important reason for coming to register your land? ______ To secure my interest in the land ______ To enable me secure a loan www.knust.edu.gh www.knust.edu.gh ______ To make it easy to transfer the land  Problems Associated with Constructing Questions i. Multiple questions in one ii. Includes overly technical language iii. Overlapping responses iv. Not all questions are applicable to all participants v. Mismatch between the question and the answers vi. Negatively worded question vii.Wordy question viii.Unbalanced response options ix. Question with jargons www.knust.edu.gh www.knust.edu.gh x. Unclear question  Examples Poor Questions Poor Questions Do you support gun Do you believe that guns and control? knives do not belong in _____ Yes schools? _____ No _____ Yes _____ Don’t know _____ No _____ Don’t know Improved Questions Improved Questions Do you believe that guns do Do you believe knives do not not belong in schools? belong in schools? _____ Yes _____ Yes _____ No _____ No www.knust.edu.gh _____ Don’t know _____ Don’t knowwww.knust.edu.gh  Examples Poor Questions How many times have you seen a student carry a handgun? _____ 0 times _____ 1-2times _____ 2-3 times _____ More than 3 times Improved Questions How many times have you seen a student carry a handgun? _____ 0 times _____ 1-2 times _____ 3-4 times _____ More than 4 times www.knust.edu.gh www.knust.edu.gh  Reliability and Validity of Instruments ❑ Reliability of an instrument means that the instrument is able to produce stable and consistent scores. ❑ Scores from the instruments should be the same irrespective of the number of times that the instrument is administered. ❑ The instrument must also be consistent, i.e. the respondents should respond to items and related items in the same way. www.knust.edu.gh www.knust.edu.gh  Reliability and Validity of Instruments ❑ Validity is the degree to which all the evidence points to the intended interpretation of test scores for the proposed purpose. ❑ Thus, the validity of an instrument is focused on the consequences of using the scores from the instruments. ❑ It is about developing evidence that is sound in demonstrating that the test interpretation (of scores about the concept or construct that the test is assumed to measure) matches its proposed use. NB: ❑ Score must be reliable before they can be valid or meaningful. ❑ Hence, the more reliable the scores from an www.knust.edu.gh instrument www.knust.edu.gh are, the more valid the scores will be.  Factors that influence Reliability ❑ Items on instruments are ambiguous and unclear. ❑ Procedures of test administration vary and are not standardized. ❑ Participants are fatigued, are nervous, misinterpret questions, or guess on tests (Rudner, 1993) www.knust.edu.gh www.knust.edu.gh  Administer Data Collection ❑ Send questionnaires through mails (emails or postal) to participants. ❑ Hand deliver questionnaires to individual people. ❑ Administration of surveys in person or over the telephone. ❑ Complete checklists for observational study. ❑ Use instruments for face to face interviews with individuals or for a group of individuals. www.knust.edu.gh www.knust.edu.gh Thank You www.knust.edu.gh www.knust.edu.gh