SOC 4 Study Guide 2
*DISCLAIMER
This is a study guide based upon my notes and my reading. As we have learnt through this class all of us may have different paradigms, therefore, this 'study guide' reflects me and may not be 100% accurate. This in short, is not a substitute for class, nor reading the assigned material.
Chapt 4
3 purposes of research
1) exploration attempt to develop a rough understanding of some phenomenon
2) description the precise measurement and reporting of characteristics of some population or phenomenon.
3) explanation the discovery and reporting of relationships among different aspects of the phenomenon under study
Nomothetic causality
3 criteria
1) Variables must be empirically associated and correlated
2) the causal variable must occur earlier in time than the variable it is said to affect
3) the observed effect cannot be explained as the effect of a different variable
the variables must be nonspurious
the variables must be nonspurious
4) the causal mechanism must be specified
ex: summertime: ice cream sales go up, crime rate also goes up at the same time
not necessarily correlated, and the mechanism in which relates to the variables are not specified.
not necessarily correlated, and the mechanism in which relates to the variables are not specified.
Necessary / Sufficient causes
-correlation does not 100% mean causation
- spurious causal relationship is in association that in reality is caused by 1 or more other variables
- cross section study is based on observations made at 1 point in time
- longitudinal study - observations are made at many times
- trend - observations made of samples from general populations
- cohort study - samples are drawn from more specific subpopulations
- panel studies - the same sample of a population is used each time
Chapt 5
- Variations between extremes
the degree of precision taken in considering the operationalizing of your given variables.
the degree of precision taken in considering the operationalizing of your given variables.
How precise do you want to define the attributes which compose your variables?
-look at the purpose and procedure of your study (specific or general)
-look at the purpose and procedure of your study (specific or general)
4 levels of measurement
1) Nominal - a variable whose attributes have only the characteristics of exhaustiveness and mutual exclusiveness
-describes a variable that has attributes that are merely different
-ex: gender [usually clear variables with 2 choices]
-ex: gender [usually clear variables with 2 choices]
2) Ordinal -describes a variable with attributes we can rank order along some dimension
ex: socioeconomic status (SES) [high, med, low]
ex: socioeconomic status (SES) [high, med, low]
3) internal - describes a variable whose attributes are rank ordered and have equal distances between adjacent attributes
-ex:10-18 (degrees F) same as 89-90
4) ratio measure - describes a variable with attributes that have all the qualities of nominal/ordinal/internal measures and in addition are based on a true zero point
-ex: age
-ex: length of residence
-ex: # of Arabian friends
-ex: length of residence
-ex: # of Arabian friends
-comparing 2 subjects in terms of a ratio variable allows us to determine:
1) they are different or the same
2) one is more than the other
3) how much they differ
4) the ratio of one to another
2) one is more than the other
3) how much they differ
4) the ratio of one to another
Ratio is the highest level of measurement, descends to internal-ordinal-nominal
Nominal = the lowest level of measurement
highest level to lower = good/ok
lower to high = usually not possible/ and if it is its usually = bad data
Quality of a measurement relies on precision and accuracy
accuracy has to do with the truth, and precision is the fineness of distinction
Reliability - the quality of measurement that suggests that the same data would have been collected each time in repeated observations of the same phenomenon
validity - describes a measure that accurately reflects the concept it is intended to measure
testing for reliability:
-test retest method
to make the same measurement more than once.
ex: questionnaire then a follow up questionnaire with the same ?'s
-test retest method
to make the same measurement more than once.
ex: questionnaire then a follow up questionnaire with the same ?'s
split half method
ex: randomly assign those 10 items to 2 sets of five. (if the 2 sets of items classify ppl differently you most likely have a problem of reliability in your measure of the variable)
ex: randomly assign those 10 items to 2 sets of five. (if the 2 sets of items classify ppl differently you most likely have a problem of reliability in your measure of the variable)
established measures
ex: U.S. Census Bureau
ex: U.S. Census Bureau
reliability of research workers
ex: do your coders know how to friggin' use microsoft excel?
ex: do your coders know how to friggin' use microsoft excel?
Testing for validity
criterion related validity - the degree to which a measure relates to some external criterion (predicative validity)
construct validity - the degree to which a measure relates to other variables as expected within a system of theoretical relationships
content validity - the degree in which measurement covers the range of meanings included within a concept
criterion related validity - the degree to which a measure relates to some external criterion (predicative validity)
construct validity - the degree to which a measure relates to other variables as expected within a system of theoretical relationships
content validity - the degree in which measurement covers the range of meanings included within a concept
chapt 6
Index: a type of composite measure that summarizes and rank orders several specific observation and represents some general dimension (needs a sufficient level of variance) [make sure variables are not under represented or over]
Indexes are constructed when we accumulate scores assigned to individual indicators
Scale: a type of composite measure composed of several items that have a logical or empirical structure among them
scales/indexes both use ordinal measures of variables
-both based on more than one data item
-both based on more than one data item
scale assigns scores to patterns of responses (measures intensity of a variable)
EX: POKER - I put 20$ worth of chips into the pot, I get my hand cards aren't very high, I fold (index = chips, how many chips in) other guy puts $20 in the pot but has the same hand but feels that he'll bluff through anything, never say die attitude, intense dedication to the hand (scale)
Constructing an Index
Five steps
1) select items for a composite index
2) examine empirical relationships
3) assign scores for responses
4) handle missing data
5) does your data reflect reality?
Five steps
1) select items for a composite index
2) examine empirical relationships
3) assign scores for responses
4) handle missing data
5) does your data reflect reality?
Validating an index:
Item Analysis ‐ internal validation
External validation ‐ ranking of groups on the index should predict the ranking of groups in answering similar or related questions.
Item Analysis ‐ internal validation
External validation ‐ ranking of groups on the index should predict the ranking of groups in answering similar or related questions.
4 CONSIDERATIONS
1)face validity - Just like face validity for single measure variables. Items chosen need to make logical sense in order to be included.
2)Unidimensionally - Index of measures should focus on only one dimension at a Dimension.
• One index for Physical Dimension of Gender; Another for
Social Role Dimension of Gender.
3) General or Specific:
3) General or Specific:
Only Domestic Life Social Role Specifically, or Social Role in General (equal measures for Domestic, Work, Relationship)
4) Variance - Make sure indicators produce a sufficient distribution of responses.
Constructing a scale:
Constructed by assigning scores to patterns of responses, recognizing that some items reflect a weak degree of the variable while others reflect something stronger.
(there are multiple techniques)
Bogardus social distance scale - measures the willingness of people to participate in social relations of varying degrees of closeness.
Consider for a moment a typical Latino/a-American
Do you mind if they live in this country?
Do you mind if they live in the same neighborhood?
Do you mind if they live next to you?
Would you have them as a close friend?
Would you mind them marrying into your family?
Would you personally date them?
Would you marry them?
Thurstone scales - judges determine the intensity of different indicators.
Researcher selects several indicators of a variable and gives them to a panel of judges.
Each judge assigns a value, say 1-13, ranking the strength of the indicator.
Say 1 for the weakest indicator of the variable and a 13 for the strongest.
The researcher then selects those measures with the widest agreement for the scale, and disregards those with the lowest.
Likert scaling - uses standardized response categories.
Strongly agree, agree, disagree, strongly disagree.
Strongly agree, agree, neutral, disagree, strongly disagree.
Strongly agree, agree, neutral, disagree, strongly disagree, I am not familiar with this.
Guttman scaling - uses an empirical intensity structure (most common).
A Guttman scale only holds for the sample for which it is created.
Typology - The classification (typically nominal) of observations in terms of their attributes on two or more variables.
The classification of newspapers as liberal-urban, liberal-rural, conservative-urban, or conservative-rural would be an example.
The classification of newspapers as liberal-urban, liberal-rural, conservative-urban, or conservative-rural would be an example.
