5___6_measurement_com_200 - Measurement Measurement What do...

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: Measurement Measurement What do we measure in social science research? social In social science, observations often involve behaviors, behaviors attitudes, cognitions or artifacts. attitudes cognitions artifacts Human behaviors refer to the actions taken behaviors by an individual Voting Working Hitting Attitudes refer to an individual’s Attitudes evaluations of objects, people, behaviors, etc. Cognitions refer to the mental process of Cognitions acquiring and using knowledge acquiring awareness awareness memory memory perception perception problem solving problem reasoning reasoning e.g. awareness of sponsor of ad; know factual info about e.g. US politics US Artifacts refer to the physical output of Artifacts human activities. human Newspapers Movies Email messages Official documents Let’s say we are interested in the following statement… following Playing violent video games increases aggression in young children What concepts do we have to measure here? First: Define the concepts First: The concept “aggression” is hostile or The destructive behavior or actions. destructive (American Heritage Dictionary) Second: Determine examples of the concepts the Describe the types of behaviors that are “hostile or destructive.” Third: Compile list of indicators Third: The Hostile and destructive behaviors The listed are indicators of aggression. indicators Indicator: something that designates something the existence of a concept. Complex concepts may have more than one indicator. indicator Indicators in everyday life Indicators We use indicators to gauge: people’s moods what pedestrians will do what is important in class http://www.youtube.com/watch?v =U1VmGjJJFrc How do we collect indicators? How Direct observation People Artifacts Subject report Self-administered questionnaires - mail, Self-administered email, group email, Interviews - phone, email, face-to-face Fourth: Assign numbers to indicators Fourth: Indicators that can be quantified become Indicators variables. variables How? Once measurable indicators are identified, rules for Once assigning numbers based on these indicators are determined determined If multiple indicators are used, they need to be If combined (e.g. added together) combined Assigning numbers to indicators Assigning How do we put numbers to aggressive behaviors? • Watch people’s actions & count the number of Watch incidences within a 30-minute period incidences • Ask people about how often they act in certain Ask ways during a period ways Developing a Variable: Example Developing Concept -- Liberal ideology Concept Indicators – Responses to survey questions Indicators 1. Would you support increasing taxes to increase funding education? (1=yes, 0=no) education? 2. Would you support increasing taxes to increase funding for health care (1=yes, 0=no) for 3. Would you support increasing taxes to increase funding for Social Security? (1=yes, 0=no) for for Variable – Add responses together to form “liberalism” variable Variable Note on Hypotheses Note Hypotheses state relationships between variables (NOT concepts). variables (NOT Measurement Example Measurement Hypothesis Hypothesis Participation in structured intergroup Participation meetings with Palestinians is associated with an increase in Israeli-Jews’ guilt toward Palestinians. toward -Maoz and Ellis (2008) Method & Measures Method Telephone survey of 289 secular Israeli Jews in June 2003: 1. “Did you participate in an intergroup Did meeting?” meeting?” Yes / No Yes 2. “I feel guilty towards Palestinians” 2. feel Rate selves in 6 categories: “disagree strongly” to “agree strongly” Variable Participate in meeting Code 1=yes 0=no Feel guilt towards Palestinians 1=disagree strongly 2 3 4 5 6=agree strongly Types of variables Types Nominal variables: variables we can place in a category but do not have a logical ordering ordering e.g. Gender Coded as 1 (male) or 0 (female) Note: This binary variable (two possible Note: binary outcomes - 1, 0) is sometimes called a dummy variable. Ordinal variables: categories that may have Ordinal a natural ordering natural e.g. A survey response of “strongly agree, e.g. agree, neutral, disagree, strongly disagree” disagree” Code as 1=strongly agree, 2=agree, Code 3=neutral, 4=disagree, 5=strongly disagree disagree Nominal and Ordinal variables are sometimes referred to as categorical variables variables Interval variable: variable in which it makes Interval sense to talk about intervals, but not ratios E.g. temperature in F or C E.g. Ratio variables: A variable for which is Ratio makes sense to talk about ratios makes Unlike interval variables, ratio variables Unlike have a meaningful zero point have e.g. Annual Income in dollars More on ratio variables More Ratio variables: Continuous- Can be any number on an interval e.g. Age, income in dollars Discrete- Can count the possible responses; must be a whole number e.g. Number of car accidents on a highway Note: Ratio variables may also be referred to as measurement variables. These definitions matter, because certain tools in statistics apply to certain types of variables variables Examples of Variables Examples Number of deaths by heart attack in a county in Number 2006 2006 Presidential candidate voted for in 2004 election Income categories: $0-19,999; $20,000-$29,999, Income $30,000-$44,999; etc. Religion: Mormon, Mainline Protestant, Atheist, etc Religion: “Extreme conservative”; “conservative”; “somewhat Extreme conservative”, “Moderate”, “somewhat liberal”; “liberal”, “extreme liberal” “liberal”, Many things can be measured! Many Religion; income; gender; ideology; how Religion; much a person agrees with a statement… much More ambiguous cases… More Church attendance? Handedness (right, left)? Handedness Survey Handedness Which hand do you use-right or left? Make Which a sheet with two columns: left and right. For each activity listed in the next slide: For Put + in the column if you use that hand Put +in both columns if you are really Put indifferent indifferent Put ++ in the column if you would never use Put never the other hand the Left Writing Throwing Scissors Broom (upper hand) Striking match (hand that holds match) Opening box (hand that holds lid) Right TOTAL: Put + in column if you use hand, + in both columns if you’re indifferent, ++ in column if you would never use other hand Now sum the number of +’s in each column and calculate: Right total-left total=__ Right Right total+left total=___ Handedness score = (Right-left) / (left + right) My handedness My Which do you use-right or left? Make a sheet with two columns: right and left. For each activity: Put + in the column if you use that hand, and + in both columns if you are really indifferent, put ++ in the column if you would in never use the other hand never Left Right Left Writing ++ Throwing ++ Scissors ++ Broom (upper hand) + Striking match (hand that holds the match) ++ Opening box (hand that holds the lid) + TOTAL: 0 10 10 Right-left=__ 10-0 Right+left=__10 + 0_ Right-left=__ Handedness score=(Right-left)/(left + right) = 10/10=1 Questions? Designing good measures Designing Today: Focus on surveys Yes, Minister We will discuss We 1. Using surveys for… Measuring behavior Measuring cognitions and attitudes 2. Some problems/issues with questions on 2. surveys surveys Measuring Behavior Measuring Ask about behavior -- questionnaire & interview The behavior needs to be something The individuals can give a response to. individuals It is best to give a time period and one they It can remember. can Examples of behavior questions Examples How many hours of sleep did you get last How night? night? Did you vote in the 2008 presidential Did election? election? Unobtrusive Measures Unobtrusive Surveys may be inappropriate for cases that Surveys may involve the Hawthorne effect or may involve socially desirable or undesirable behaviors behaviors Hawthorne effect --The process of being a research subject causes a change in behavior. behavior. Unobtrusive measures Unobtrusive Types of unobtrusive measures Archives -- existing material; e.g. official Archives vote records vote Observation -- Watching without the Observation subjects knowing subjects Measuring attitudes & cognitions attitudes Likert-type scales Thermometer scaling Likert-type scales scales Likert-type scale: Questionnaire item Likert-type requiring respondents to specify their level requiring of agreement to a statement of Developed by Rensis Likert Developed Also called summated rating Also summated Likert-type scales Likert-type The Republican Party better understands the country’s economic The problems than the Democratic Party. problems 1.____ 2.____ 3. ____ 4. ____ 5.____ 1.____ 2.____ 3. 4. 5. Strongly Disagree Strongly Disagree Neither Agree Agree/Disagree Agree/Disagree Strongly Agree I trust the Republican Party to handle foreign affairs more than trust the Democratic Party. the 1.____ 2.____ 3. ____ 4. ____ 5.____ 1.____ 2.____ 3. 4. 5. Strongly Disagree Strongly Disagree Neither Agree Agree/Disagree Agree/Disagree Strongly Agree Likert-type scales Likert scales are meant to be summed. They represent multiple indicators. represent Whether they are ordinal or interval is debated, but Whether most researchers treat them as interval. most If summing Likert scales, they must be positively If related Likert scale notes Likert Can reverse-code items before summing them if Can necessary necessary The Democratic Party is better suited to handle environmental problems than The the Republican Party. the 1.____ 2.____ 3. ____ 4. ____ 5.____ Strongly Disagree Disagree Neither Agree Strongly Agree Strongly Agree/Disagree Agree/Disagree Likert scales notes (con) Likert Determine relationship b/w Likert scales Determine with factor analysis factor Factor analysis shows that the items are related and therefore measuring the same concept. same Can also be used for other types of Can indicators-not just Likert scales indicators-not We will not cover factor analysis in this We class class Likert-type scales Likert-type How many options should be offered in How these scales? these The more the better, BUT: Language is limited in describing iintervals. ntervals. Seven intervals seem to be good. Seven Likert scale example: Likert Poll of white registered voters Using Likert Scales Using Attitudes about Energy Attitudes Conservation Conservation The last person leaving a room should always turn off the lights. the Strongly agree -- 5 Agree -- 4 Neither agree nor disagree -- 3 Disagree -- 2 Strongly disagree -- 1 Everyone should drive hybrid automobiles. hybrid Strongly agree -- 5 Agree -- 4 Neither agree nor disagree -- 3 Disagree -- 2 Strongly disagree -- 1 The United States should invest more money in developing the generation of wind energy. of Strongly agree -- 5 Agree -- 4 Neither agree nor disagree -- 3 Disagree -- 2 Strongly disagree -- 1 Thermometer scaling Thermometer This is one of a variety of graphic This presentations that can be used. presentations Typically, this is used with a zero to 100 Typically, scale. scale. Often used in marketing and advertising Thermometer Thermometer 0 25 50 75 100 Feeling Thermometer Attitudes cannot always be measured with surveys measured Example: theory that people who are Example: extremely homophobic may be repressing homosexual urges homosexual Think: American Beauty This attitude is not something homophobes This would or even could admit would How to test this? Study of homophobia Study Have men watch sexually explicit videos Have while measuring how “turned on” the subjects were subjects How to do this? With a plethysmograph: a rubber ring placed With around the penis that is sensitive to changes in circumference-ie erections changes Homophobic men were more likely to get erections during male homosexual film than non-homophobic men than This is a controversial measure as there is This evidence that anxiety can cause sexual arousal arousal Short Survey Results Short Does question wording matter? Does wording In the survey I handed out, there were two versions of In the wallet question: the 1. If you found a wallet with $20 in it, would you keep If it? YES NO it? 2. If you found a wallet with $20 in it, would you do the If honest thing and return it? YES NO honest Will the responses be different? “Wallet” question 79% wouldn’t “keep it” 88% would “do the honest thing and return 88% it” it” Do you think radical Muslims critical of the United States should be allowed to speak on campus? on Do you think radical Muslims critical of the Do United States should be prohibited from speaking on campus? speaking “Radical Muslims” question Radical 62% would “allow” 74% would NOT “prohibit” These question wording effects are surprising because the two questions are logically equivalent logically Does question order matter? order Is the diameter of the moon MORE or LESS than Is 5,000 miles? Is the diameter of the moon MORE or LESS than 1,000 miles? Later, I ask you to guess the diameter of the moon. Anchoring Anchoring 5000 anchor: 5600 1000 anchor: 4900 (Actually, I omitted outliers...more on this in a (Actually, couple of classes) couple Actual diameter of the moon: 2160 miles Question Wording Effects Question Deliberate bias “Do you agree that abortion, the murder of innocent Do beings, should be outlawed?” beings, “Do you agree that there are circumstances under Do which abortion should be legal, to protect the rights of the mother?” the Unintentional Bias Questions that are easy to misinterpret “Do you take drugs?” prescription drugs? Illegal drugs? prescription Question wording effects Question Asking the uninformed Are you conservative, moderate, liberal? Unnecessary complexity Unnecessary “Do you sometimes find that you have Do arguments with your family and coarguments workers?” Question ordering Question Torture example Torture Other effects Other Sensitive topics e.g. racial attitudes/sexual behavior Desire to please Did you vote in 2008? Open-ended v. closed-ended questions Open-hard to code; many different responses Closed-options offered tend to be selected more than Closed-options they would have with closed option they When offered the option, 29% mentioned “the environment”; 0 When spontaneously mentioned this in open-ended spontaneously Evaluating our Measures Evaluating What is a good measure of liberalism, What political knowledge, Facebook use, performance in school, etc.? performance Criteria for Evaluating Variables Criteria Validity: How well a measure actually measures what you How want to measure want Com 200 exam Reliability: whether the measure gives approximately the whether same answer time after time IQ tests IQ Bias: whether a measure is systematically off the mark in whether one direction one Are you very conservative, somewhat conservative, Are moderate, somewhat liberal, or very liberal moderate, Any questions? Any ...
View Full Document

This note was uploaded on 01/27/2011 for the course COM 200 taught by Professor Tamborini during the Fall '09 term at Michigan State University.

Ask a homework question - tutors are online